WO2020196962A1 - 인공 지능 청소기 및 그의 동작 방법 - Google Patents
인공 지능 청소기 및 그의 동작 방법 Download PDFInfo
- Publication number
- WO2020196962A1 WO2020196962A1 PCT/KR2019/003636 KR2019003636W WO2020196962A1 WO 2020196962 A1 WO2020196962 A1 WO 2020196962A1 KR 2019003636 W KR2019003636 W KR 2019003636W WO 2020196962 A1 WO2020196962 A1 WO 2020196962A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- cleaning
- artificial intelligence
- area
- cleaner
- areas
- Prior art date
Links
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 104
- 238000000034 method Methods 0.000 title claims description 48
- 238000004140 cleaning Methods 0.000 claims abstract description 568
- 230000004807 localization Effects 0.000 claims abstract description 5
- 238000013507 mapping Methods 0.000 claims abstract description 5
- 238000013528 artificial neural network Methods 0.000 claims description 70
- 238000013135 deep learning Methods 0.000 claims description 34
- 238000013145 classification model Methods 0.000 claims description 32
- 238000004891 communication Methods 0.000 claims description 23
- 238000010801 machine learning Methods 0.000 claims description 23
- 238000012549 training Methods 0.000 description 47
- 239000010410 layer Substances 0.000 description 43
- 238000010586 diagram Methods 0.000 description 24
- 230000006870 function Effects 0.000 description 20
- 210000002569 neuron Anatomy 0.000 description 17
- 238000002372 labelling Methods 0.000 description 16
- 238000001514 detection method Methods 0.000 description 12
- 239000000428 dust Substances 0.000 description 9
- 230000008569 process Effects 0.000 description 9
- 238000005516 engineering process Methods 0.000 description 8
- 238000005457 optimization Methods 0.000 description 8
- 239000000126 substance Substances 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 239000003795 chemical substances by application Substances 0.000 description 4
- 230000007774 longterm Effects 0.000 description 4
- 230000004913 activation Effects 0.000 description 3
- 238000010295 mobile communication Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000002787 reinforcement Effects 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000012706 support-vector machine Methods 0.000 description 3
- 210000000225 synapse Anatomy 0.000 description 3
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 2
- 230000009471 action Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000007418 data mining Methods 0.000 description 2
- 238000003066 decision tree Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000011478 gradient descent method Methods 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 239000002356 single layer Substances 0.000 description 2
- 238000012800 visualization Methods 0.000 description 2
- 241001465754 Metazoa Species 0.000 description 1
- XUIMIQQOPSSXEZ-UHFFFAOYSA-N Silicon Chemical compound [Si] XUIMIQQOPSSXEZ-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 238000013529 biological neural network Methods 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 210000003169 central nervous system Anatomy 0.000 description 1
- 238000010224 classification analysis Methods 0.000 description 1
- 230000001149 cognitive effect Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000006837 decompression Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000011086 high cleaning Methods 0.000 description 1
- 238000012880 independent component analysis Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011017 operating method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 229910052710 silicon Inorganic materials 0.000 description 1
- 239000010703 silicon Substances 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000000946 synaptic effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Classifications
-
- A—HUMAN NECESSITIES
- A47—FURNITURE; DOMESTIC ARTICLES OR APPLIANCES; COFFEE MILLS; SPICE MILLS; SUCTION CLEANERS IN GENERAL
- A47L—DOMESTIC WASHING OR CLEANING; SUCTION CLEANERS IN GENERAL
- A47L11/00—Machines for cleaning floors, carpets, furniture, walls, or wall coverings
- A47L11/40—Parts or details of machines not provided for in groups A47L11/02 - A47L11/38, or not restricted to one of these groups, e.g. handles, arrangements of switches, skirts, buffers, levers
- A47L11/4011—Regulation of the cleaning machine by electric means; Control systems and remote control systems therefor
-
- A—HUMAN NECESSITIES
- A47—FURNITURE; DOMESTIC ARTICLES OR APPLIANCES; COFFEE MILLS; SPICE MILLS; SUCTION CLEANERS IN GENERAL
- A47L—DOMESTIC WASHING OR CLEANING; SUCTION CLEANERS IN GENERAL
- A47L9/00—Details or accessories of suction cleaners, e.g. mechanical means for controlling the suction or for effecting pulsating action; Storing devices specially adapted to suction cleaners or parts thereof; Carrying-vehicles specially adapted for suction cleaners
- A47L9/28—Installation of the electric equipment, e.g. adaptation or attachment to the suction cleaner; Controlling suction cleaners by electric means
- A47L9/2836—Installation of the electric equipment, e.g. adaptation or attachment to the suction cleaner; Controlling suction cleaners by electric means characterised by the parts which are controlled
- A47L9/2852—Elements for displacement of the vacuum cleaner or the accessories therefor, e.g. wheels, casters or nozzles
-
- A—HUMAN NECESSITIES
- A47—FURNITURE; DOMESTIC ARTICLES OR APPLIANCES; COFFEE MILLS; SPICE MILLS; SUCTION CLEANERS IN GENERAL
- A47L—DOMESTIC WASHING OR CLEANING; SUCTION CLEANERS IN GENERAL
- A47L11/00—Machines for cleaning floors, carpets, furniture, walls, or wall coverings
- A47L11/40—Parts or details of machines not provided for in groups A47L11/02 - A47L11/38, or not restricted to one of these groups, e.g. handles, arrangements of switches, skirts, buffers, levers
- A47L11/4061—Steering means; Means for avoiding obstacles; Details related to the place where the driver is accommodated
-
- A—HUMAN NECESSITIES
- A47—FURNITURE; DOMESTIC ARTICLES OR APPLIANCES; COFFEE MILLS; SPICE MILLS; SUCTION CLEANERS IN GENERAL
- A47L—DOMESTIC WASHING OR CLEANING; SUCTION CLEANERS IN GENERAL
- A47L9/00—Details or accessories of suction cleaners, e.g. mechanical means for controlling the suction or for effecting pulsating action; Storing devices specially adapted to suction cleaners or parts thereof; Carrying-vehicles specially adapted for suction cleaners
- A47L9/28—Installation of the electric equipment, e.g. adaptation or attachment to the suction cleaner; Controlling suction cleaners by electric means
- A47L9/2894—Details related to signal transmission in suction cleaners
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J11/00—Manipulators not otherwise provided for
- B25J11/008—Manipulators for service tasks
- B25J11/0085—Cleaning
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/0003—Home robots, i.e. small robots for domestic use
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1602—Programme controls characterised by the control system, structure, architecture
- B25J9/161—Hardware, e.g. neural networks, fuzzy logic, interfaces, processor
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1656—Programme controls characterised by programming, planning systems for manipulators
- B25J9/1664—Programme controls characterised by programming, planning systems for manipulators characterised by motion, path, trajectory planning
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/0088—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots characterized by the autonomous decision making process, e.g. artificial intelligence, predefined behaviours
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0212—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory
- G05D1/0219—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory ensuring the processing of the whole working surface
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0212—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory
- G05D1/0221—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory involving a learning process
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0268—Control of position or course in two dimensions specially adapted to land vehicles using internal positioning means
- G05D1/0274—Control of position or course in two dimensions specially adapted to land vehicles using internal positioning means using mapping information stored in a memory device
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
- G06T7/55—Depth or shape recovery from multiple images
- G06T7/579—Depth or shape recovery from multiple images from motion
-
- A—HUMAN NECESSITIES
- A47—FURNITURE; DOMESTIC ARTICLES OR APPLIANCES; COFFEE MILLS; SPICE MILLS; SUCTION CLEANERS IN GENERAL
- A47L—DOMESTIC WASHING OR CLEANING; SUCTION CLEANERS IN GENERAL
- A47L2201/00—Robotic cleaning machines, i.e. with automatic control of the travelling movement or the cleaning operation
- A47L2201/04—Automatic control of the travelling movement; Automatic obstacle detection
Definitions
- the present invention relates to an artificial intelligence cleaner and a method of operating the same. More specifically, the present invention relates to an artificial intelligence cleaner that classifies a zone that reduces cleaning efficiency using cleaning data, and resets a cleaning path or a cleaning mode within the cleaning path using zone classification information.
- a robot cleaner is a device that automatically cleans by inhaling foreign substances such as dust from a floor surface while traveling by itself in an area to be cleaned without a user's manipulation. Such a robot cleaner sets a cleaning path according to a built-in program, and performs a cleaning operation while driving along the set cleaning path.
- a robot cleaner does not consider the entire cleaning area, but only considers the environment of a certain radius (for example, a radius of 25) based on the cleaner, and performs cleaning while avoiding obstacles. Therefore, the robot cleaner often gets lost in a previously wandered area such as an area that takes a long time to clean or an area that repeatedly hits an obstacle. For example, in an area with many obstacles, cleaning may take a lot of time or may be restricted due to movement restrictions. In addition, certain areas are cleaned more than other areas.
- An object of the present invention is to provide an artificial intelligence vacuum cleaner that separates and cleans a space that is difficult to clean and a space that can be easily cleaned from the cleaning record of the artificial intelligence cleaner, and an operating method thereof.
- the present invention is to provide an artificial intelligent vacuum cleaner and an operation method thereof that set a cleaning mode and a cleaning sequence of each cleaning area in consideration of the characteristics of the area, and clean according to a cleaning movement line in consideration of the set cleaning mode and cleaning sequence.
- the artificial intelligence cleaner divides a cleaning space into a plurality of cleaning areas using a plurality of cleaning records, sets a cleaning movement line in consideration of each cleaning area, and performs cleaning according to the set cleaning movement line. can do.
- each cleaning area is divided into area types, and a cleaning movement line may be set in consideration of a cleaning mode for each area type and a cleaning priority between area types.
- cleaning may be performed more suitably to the shape or characteristic of the space.
- a cleaning mode set for each area type and a cleaning priority between area types are set, it is possible to provide a cleaning mode suitable for the characteristics of each space, and cleaning from a high priority area. You can increase the satisfaction of cleaning and lower the probability of cleaning failure.
- FIG. 1 is a block diagram showing the configuration of an artificial intelligence cleaner 100 according to an embodiment of the present invention.
- FIG. 2 is a perspective view of an artificial intelligence cleaner 100 according to an embodiment of the present invention.
- FIG 3 is a bottom view of the artificial intelligence cleaner 100 according to an embodiment of the present invention.
- FIG. 4 is a block diagram showing the configuration of an artificial neural network learning apparatus 200 according to an embodiment of the present invention.
- FIG. 5 is a flowchart illustrating a method of operating an artificial intelligence cleaner according to an embodiment of the present invention.
- FIG. 6 is a diagram showing an example of a cleaning record according to an embodiment of the present invention.
- FIG. 7 is a diagram showing an example of a cleaning record according to an embodiment of the present invention.
- FIG. 8 is a diagram showing an example of a cleaning record according to an embodiment of the present invention.
- FIG. 9 is a diagram illustrating an example of a cleaning record according to an embodiment of the present invention.
- FIG. 10 is a flowchart illustrating an example of a step S503 of dividing the cleaning space shown in FIG. 5 into a plurality of cleaning areas.
- FIG. 11 is a diagram illustrating an example of a rule-based cleaning area classification model according to an embodiment of the present invention.
- FIG. 12 is a table showing an example of a deep learning-based cleaning area classification model according to an embodiment of the present invention.
- FIG. 13 to 15 are views illustrating examples of cleaning spaces in which cleaning areas are separated from each other according to an embodiment of the present invention.
- 16 is an operation flowchart showing an example of a step S505 of determining a cleaning movement line shown in FIG. 5.
- 17 and 18 are diagrams illustrating an example of setting a cleaning mode through a user input in an embodiment of the present invention.
- 19 is a diagram illustrating an example of setting cleaning priority through user input in an embodiment of the present invention.
- 20 and 21 are diagrams showing examples of cleaning movement lines of the artificial intelligence cleaner 100 according to an embodiment of the present invention.
- AI Artificial intelligence
- artificial intelligence does not exist by itself, but is directly or indirectly related to other fields of computer science.
- attempts are being made very actively to introduce artificial intelligence elements in various fields of information technology and to use them in solving problems in that field.
- Machine learning is a branch of artificial intelligence, a field of research that gives computers the ability to learn without explicit programming.
- machine learning can be said to be a technology that studies and builds a system that learns based on empirical data, performs prediction, and improves its own performance, and algorithms for it.
- Machine learning algorithms do not execute strictly defined static program instructions, but rather build specific models to derive predictions or decisions based on input data.
- machine learning' can be used interchangeably with the term'machine learning'.
- the decision tree is an analysis method that charts decision rules into a tree structure and performs classification and prediction.
- Bayesian network is a model that expresses the probabilistic relationship (conditional independence) between multiple variables in a graph structure. Bayesian networks are suitable for data mining through unsupervised learning.
- the support vector machine is a model of supervised learning for pattern recognition and data analysis, and is mainly used for classification and regression analysis.
- An artificial neural network is an information processing system in which a number of neurons, called nodes or processing elements, are connected in a layer structure by modeling the operation principle of biological neurons and the connection relationship between neurons.
- Artificial neural networks are models used in machine learning, and are statistical learning algorithms inspired by biological neural networks (especially the brain among animals' central nervous systems) in machine learning and cognitive science.
- the artificial neural network may refer to an overall model having problem-solving ability by changing the strength of synaptic bonding through learning by artificial neurons (nodes) that form a network by combining synapses.
- artificial neural network may be used interchangeably with the term neural network.
- the artificial neural network may include a plurality of layers, and each of the layers may include a plurality of neurons.
- artificial neural networks may include synapses that connect neurons and neurons.
- Artificial neural networks generally have three factors: (1) the connection pattern between neurons in different layers (2) the learning process to update the weight of the connection (3) the output value from the weighted sum of the input received from the previous layer. It can be defined by the activation function it creates.
- the artificial neural network may include network models such as DNN (Deep Neural Network), RNN (Recurrent Neural Network), BRDNN (Bidirectional Recurrent Deep Neural Network), MLP (Multilayer Perceptron), CNN (Convolutional Neural Network). , Is not limited thereto.
- DNN Deep Neural Network
- RNN Recurrent Neural Network
- BRDNN Bidirectional Recurrent Deep Neural Network
- MLP Multilayer Perceptron
- CNN Convolutional Neural Network
- the term'layer' may be used interchangeably with the term'layer'.
- a general single-layer neural network is composed of an input layer and an output layer.
- a general multilayer neural network is composed of an input layer, one or more hidden layers, and an output layer.
- the input layer is a layer that receives external data
- the number of neurons in the input layer is the same as the number of input variables
- the hidden layer is located between the input layer and the output layer, receives signals from the input layer, extracts characteristics, and transfers them to the output layer. do.
- the output layer receives a signal from the hidden layer and outputs an output value based on the received signal.
- the input signal between neurons is multiplied by each connection strength (weight) and then summed. If the sum is greater than the neuron's threshold, the neuron is activated and the output value obtained through the activation function is output.
- a deep neural network including a plurality of hidden layers between an input layer and an output layer may be a representative artificial neural network implementing deep learning, a type of machine learning technology.
- the term'deep learning' can be used interchangeably with the term'deep learning'.
- the artificial neural network can be trained using training data.
- learning means a process of determining parameters of an artificial neural network using training data in order to achieve the purpose of classifying, regressing, or clustering input data.
- parameters of an artificial neural network include weights applied to synapses or biases applied to neurons.
- the artificial neural network learned by the training data may classify or cluster input data according to patterns of the input data.
- an artificial neural network trained using training data may be referred to as a trained model in this specification.
- the following describes the learning method of artificial neural networks.
- Learning methods of artificial neural networks can be classified into supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning.
- Supervised learning is a method of machine learning to infer a function from training data.
- outputting a continuous value is called regression, and predicting and outputting the class of an input vector can be called classification.
- an artificial neural network is trained with a label for training data.
- the label may mean a correct answer (or result value) that the artificial neural network must infer when training data is input to the artificial neural network.
- the correct answer (or result value) to be inferred by the artificial neural network is referred to as a label or labeling data.
- labeling setting a label on training data for learning an artificial neural network is referred to as labeling the training data with labeling data.
- the training data and the label corresponding to the training data constitute one training set, and may be input to the artificial neural network in the form of a training set.
- the training data represents a plurality of features
- labeling of the training data may mean that a label is attached to the feature represented by the training data.
- the training data may represent the characteristics of the input object in the form of a vector.
- the artificial neural network can infer a function for the correlation between the training data and the labeling data using the training data and the labeling data.
- parameters of the artificial neural network may be determined (optimized) through evaluation of a function inferred from the artificial neural network.
- Unsupervised learning is a type of machine learning, where no labels are given for training data.
- the unsupervised learning may be a learning method of training an artificial neural network to find and classify patterns in the training data itself, rather than an association relationship between training data and a label corresponding to the training data.
- unsupervised learning examples include clustering or independent component analysis.
- Examples of artificial neural networks using unsupervised learning include Generative Adversarial Network (GAN) and Autoencoder (AE).
- GAN Generative Adversarial Network
- AE Autoencoder
- a generative adversarial neural network is a machine learning method in which two different artificial intelligences compete and improve performance, a generator and a discriminator.
- the generator is a model that creates new data and can create new data based on the original data.
- the discriminator is a model that recognizes a pattern of data, and may play a role of discriminating whether input data is original data or fake data generated by a generator.
- the generator learns by receiving data that cannot be deceived by the discriminator, and the discriminator can learn by receiving deceived data from the generator. Accordingly, the generator can evolve to deceive the discriminator as well as possible, and the discriminator can evolve to distinguish between the original data and the data generated by the generator.
- Auto encoders are neural networks that aim to reproduce the input itself as an output.
- the auto encoder includes an input layer, at least one hidden layer and an output layer.
- data output from the hidden layer goes to the output layer.
- the dimension of the data increases, and accordingly, decompression or decoding is performed.
- the auto-encoder controls the connection strength of neurons through learning, so that the input data is expressed as hidden layer data.
- the hidden layer information is expressed with fewer neurons than in the input layer, but being able to reproduce the input data as an output may mean that the hidden layer found and expressed a hidden pattern from the input data.
- Semi-supervised learning is a kind of machine learning, and may mean a learning method using both labeled training data and unlabeled training data.
- Reinforcement learning is the theory that, given an environment in which an agent can judge what action to do at every moment, it can find the best way to experience without data.
- Reinforcement learning can be mainly performed by the Markov Decision Process (MDP).
- MDP Markov Decision Process
- the structure of the artificial neural network is specified by the configuration of the model, activation function, loss function or cost function, learning algorithm, optimization algorithm, etc., and hyperparameters are pre-trained. It is set, and then, a model parameter is set through learning, so that the content can be specified.
- factors determining the structure of an artificial neural network may include the number of hidden layers, the number of hidden nodes included in each hidden layer, an input feature vector, a target feature vector, and the like.
- Hyperparameters include several parameters that must be initially set for learning, such as initial values of model parameters. And, the model parameter includes several parameters to be determined through learning.
- the hyperparameter may include an initial weight value between nodes, an initial bias value between nodes, a mini-batch size, a number of learning iterations, and a learning rate.
- the model parameters may include weights between nodes and biases between nodes.
- the loss function can be used as an index (reference) for determining an optimal model parameter in the learning process of the artificial neural network.
- learning refers to the process of manipulating model parameters to reduce the loss function, and the purpose of learning can be seen as determining model parameters that minimize the loss function.
- the loss function may mainly use a mean squared error (MSE) or a cross entropy error (CEE), but the present invention is not limited thereto.
- MSE mean squared error
- CEE cross entropy error
- the cross entropy error may be used when the correct answer label is one-hot encoded.
- One-hot encoding is an encoding method in which the correct answer label value is set to 1 only for neurons corresponding to the correct answer, and the correct answer label value is set to 0 for non-correct answer neurons.
- learning optimization algorithms can be used to minimize loss functions, and learning optimization algorithms include Gradient Descent (GD), Stochastic Gradient Descent (SGD), and Momentum. ), NAG (Nesterov Accelerate Gradient), Adagrad, AdaDelta, RMSProp, Adam, Nadam, etc.
- Gradient descent is a technique for adjusting model parameters in the direction of reducing the loss function value by considering the slope of the loss function in the current state.
- the direction to adjust the model parameter is called the step direction, and the size to be adjusted is called the step size.
- the step size may mean a learning rate.
- a gradient is obtained by partial differentiation of a loss function into each model parameter, and model parameters are updated by changing the acquired gradient direction by a learning rate.
- the stochastic gradient descent method is a technique that increases the frequency of gradient descent by dividing training data into mini-batch and performing gradient descent for each mini-batch.
- Adagrad, AdaDelta, and RMSProp are techniques that increase optimization accuracy by adjusting the step size in SGD.
- momentum and NAG are techniques to increase optimization accuracy by adjusting the step direction.
- Adam is a technique that improves optimization accuracy by adjusting the step size and step direction by combining momentum and RMSProp.
- Nadam is a technique that improves optimization accuracy by adjusting step size and step direction by combining NAG and RMSProp.
- the learning speed and accuracy of an artificial neural network are highly dependent on hyperparameters as well as the structure of the artificial neural network and the type of learning optimization algorithm. Therefore, in order to obtain a good learning model, it is important not only to determine an appropriate artificial neural network structure and learning algorithm, but also to set appropriate hyperparameters.
- hyperparameters are experimentally set to various values to train an artificial neural network, and as a result of learning, the hyperparameter is set to an optimal value that provides stable learning speed and accuracy.
- FIG. 1 is a block diagram showing the configuration of an artificial intelligence cleaner 100 according to an embodiment of the present invention.
- the artificial intelligence cleaner 100 includes an image sensor 110, a microphone 120, an obstacle detection unit 130, a wireless communication unit 140, a memory 150, and a driving device.
- a driver 170 and a processor 190 may be included.
- the artificial intelligence cleaner 100 may be referred to as the terminal 100.
- the image sensor 110 may acquire image data about the surroundings of the artificial intelligence cleaner 100.
- the image sensor 110 may include one or more of the depth sensor 111 and the RGB sensor 113.
- the depth sensor 111 may detect that light irradiated from the light emitting unit (not shown) is reflected on an object and returned.
- the depth sensor 111 may measure a distance to an object based on a time difference when the returned light is sensed, an amount of the returned light, and the like.
- the depth sensor 111 may acquire 2D image information or 3D image information around the cleaner 100 based on the measured distance between objects.
- the RGB sensor 113 may acquire color image information about objects around the cleaner 100.
- the color image information may be a photographed image of an object.
- the RGB sensor 113 may be referred to as an RGB camera.
- the microphone 120 may receive a user's voice.
- the received user's voice may be analyzed for intention information through a voice server (not shown).
- the user's voice may be a voice for controlling the artificial intelligence cleaner 100.
- the obstacle detection unit 130 may include an ultrasonic sensor, an infrared sensor, a laser sensor, and the like.
- the obstacle detection unit 130 may irradiate laser light to the cleaning area and extract a pattern of the reflected laser light.
- the obstacle detection unit 130 may detect the obstacle based on the position and pattern of the extracted laser light.
- the configuration of the obstacle detection unit 130 may be omitted.
- the wireless communication unit 140 may include at least one of a wireless Internet module and a short-range communication module.
- the mobile communication module includes technical standards or communication methods for mobile communication (for example, GSM (Global System for Mobile communication), CDMA (Code Division Multi Access), CDMA2000 (Code Division Multi Access 2000)), EV-DO (Enhanced Voice-Data Optimized or Enhanced Voice-Data Only), WCDMA (Wideband CDMA), HSDPA (High Speed Downlink Packet Access), HSUPA (High Speed Uplink Packet Access), LTE (Long Term Evolution), LTE-A (Long Term) Evolution-Advanced), etc.), transmits and receives a radio signal with at least one of a base station, an external terminal, and a server.
- GSM Global System for Mobile communication
- CDMA Code Division Multi Access
- CDMA2000 Code Division Multi Access 2000
- EV-DO Enhanced Voice-Data Optimized or Enhanced Voice-Data Only
- WCDMA Wideband CDMA
- HSDPA High Speed Downlink Packet Access
- HSUPA High Speed Uplink Packet Access
- LTE Long Term Evolution
- LTE-A
- the wireless Internet module refers to a module for wireless Internet access, and may be built-in or external to the terminal 100.
- the wireless Internet module is configured to transmit and receive wireless signals in a communication network according to wireless Internet technologies.
- wireless Internet technologies include WLAN (Wireless LAN), Wi-Fi (Wireless-Fidelity), Wi-Fi (Wireless Fidelity) Direct, DLNA (Digital Living Network Alliance), WiBro (Wireless Broadband), WiMAX (World Interoperability for Microwave Access), High Speed Downlink Packet Access (HSDPA), High Speed Uplink Packet Access (HSUPA), Long Term Evolution (LTE), and Long Term Evolution-Advanced (LTE-A).
- WLAN Wireless LAN
- Wi-Fi Wireless-Fidelity
- Wi-Fi Wireless Fidelity
- Direct wireless Internet technologies
- DLNA Digital Living Network Alliance
- WiBro Wireless Broadband
- WiMAX Worldwide Interoperability for Microwave Access
- HSDPA High Speed Downlink Packet Access
- HSUPA High Speed Uplink Packet Access
- LTE Long Term Evolution
- LTE-A Long Term Evolution-Advanced
- the short range communication module is for short range communication, and includes Bluetooth (Bluetooth), RFID (Radio Frequency Identification), Infrared Data Association (IrDA), UWB (Ultra Wideband), ZigBee, NFC (Near Field Communication), Wi-Fi (Wireless-Fidelity), Wi-Fi Direct, and Wireless Universal Serial Bus (USB) technologies may be used to support short-range communication.
- Bluetooth Bluetooth
- RFID Radio Frequency Identification
- IrDA Infrared Data Association
- UWB Ultra Wideband
- ZigBee Ultra Wideband
- NFC Near Field Communication
- Wi-Fi Wireless-Fidelity
- Wi-Fi Direct Wireless Universal Serial Bus
- the memory 150 may store a SLAM map created through a Simultaneous Localization And Mapping (SLAM) algorithm.
- SLAM Simultaneous Localization And Mapping
- the movement detection sensor 160 may detect the movement of the artificial intelligence cleaner 100. Specifically, the movement detection sensor 160 may detect that the artificial intelligence cleaner 100 is lifted and moved by a user.
- the movement detection sensor 160 may include one or more of a floor detection sensor 161 and a gyro sensor 163.
- the floor detection sensor 161 may detect whether the artificial intelligence cleaner 100 has been moved by a user using infrared rays. A detailed description of this will be described later.
- the gyro sensor 163 may measure the angular velocity of the artificial intelligence cleaner 100 for each of the x-axis, y-axis, and z-axis. The gyro sensor 163 may detect movement of the artificial intelligence cleaner 100 by the user using the amount of change in angular velocity for each axis.
- the movement detection sensor 160 may include a wheel sensor, a cliff sensor, and the like, and may detect movement of the artificial intelligence cleaner 100 by a user by using this.
- the driving driving unit 170 may move the artificial intelligence cleaner 100 in a specific direction or by a specific distance.
- the driving driving unit 170 may include a left wheel driving unit 171 for driving the left wheel of the artificial intelligence cleaner 100 and a right wheel driving unit 173 for driving the right wheel.
- the left wheel driving unit 171 may include a motor for driving the left wheel
- the right wheel driving unit 173 may include a motor for driving the right wheel.
- the driving driving unit 170 includes the left wheel driving unit 171 and the right wheel driving unit 173, but it is not necessary to be limited thereto, and when there is only one wheel, only one driving unit may be provided. .
- the processor 190 may control the overall operation of the artificial intelligence cleaner 100.
- FIG. 2 is a perspective view of an artificial intelligence cleaner 100 according to an embodiment of the present invention.
- the artificial intelligence cleaner 100 may include a cleaner body 50 and an image sensor 110 provided on an upper surface of the cleaner body 50.
- the image sensor 110 may irradiate light to the front and receive the reflected light.
- the image sensor 110 may acquire depth information by using a time difference in which the received light returns.
- the cleaner body 50 may include other components other than the image sensor 110 among the components described in FIG. 1.
- FIG 3 is a bottom view of the artificial intelligence cleaner 100 according to an embodiment of the present invention.
- the artificial intelligence cleaner 100 may further include a cleaner body 50, a left wheel 61a, a right wheel 61b, and a suction unit 70.
- the left wheel 61a and the right wheel 61b may drive the cleaner body 50.
- the left wheel drive unit 171 may drive the left wheel 61a, and the right wheel drive unit 173 may drive the right wheel 61b.
- the artificial intelligence cleaner 100 may suck foreign substances such as dust or garbage through the suction unit 70.
- the suction unit 70 may be provided in the cleaner body 50 to suck dust from the bottom surface.
- the suction unit 70 may further include a filter (not shown) for collecting foreign substances from the suctioned airflow, and a foreign substance receiver (not shown) in which foreign substances collected by the filter are accumulated.
- FIG. 4 is a block diagram showing the configuration of an artificial neural network learning apparatus 200 according to an embodiment of the present invention.
- the learning device 200 is a device or server configured separately outside the artificial intelligence cleaner 100 and is configured to receive, classify, store, and output information to be used for data mining, data analysis, intelligent decision making, and machine learning algorithms. Can be.
- the machine learning algorithm may include a deep learning algorithm.
- the learning device 200 may communicate with at least one artificial intelligence cleaner 100, and may derive a result by analyzing or learning data on or in place of the artificial intelligence cleaner 100.
- the meaning of helping other devices may mean distribution of computing power through distributed processing.
- the learning device 200 of an artificial neural network is various devices for learning an artificial neural network, and may generally mean a server, and may be referred to as a learning device or a learning server.
- the learning device 200 may be implemented as a single server as well as a plurality of server sets, cloud servers, or a combination thereof.
- the learning device 200 may be configured in plural to form a learning device set (or cloud server), and at least one or more learning devices 200 included in the learning device set may analyze or learn data through distributed processing. Results can be derived.
- the learning device 200 may transmit the model learned by machine learning or deep learning to the artificial intelligence cleaner 100 periodically or upon request.
- the learning device 200 may store the learned model and transmit a result value derived by using the learned model to the artificial intelligence cleaner 100 when requested by the artificial intelligence cleaner 100.
- the learning device 200 includes a communication unit 210, an input unit 220, a memory 230, a learning processor 240, and a power supply unit. , 250), and a processor 260.
- the communication unit 210 may correspond to the wireless communication unit 140 of FIG. 1.
- the input unit 220 may obtain training data for model training and input data for obtaining an output using a trained model.
- the input unit 220 may obtain unprocessed input data.
- the processor 260 may pre-process the obtained data to generate training data or pre-processed input data that can be input to model training.
- the pre-processing of input data performed by the input unit 220 may mean extracting an input feature from the input data.
- the memory 230 stores models, training data, input data, etc. learned through a machine learning algorithm or a deep learning algorithm.
- the memory 230 may include a model storage unit 231 and a database 232.
- the model storage unit 231 stores the model being trained or trained through the learning processor 240 (or artificial neural network 231a), and stores the updated model when the model is updated through training.
- the model storage unit 231 may divide and store the learned model as a plurality of versions according to a learning time point or a learning progress, if necessary.
- the artificial neural network 231a shown in FIG. 4 is only an example of an artificial neural network including a plurality of hidden layers, and the artificial neural network of the present invention is not limited thereto.
- the artificial neural network 231a may be implemented in hardware, software, or a combination of hardware and software. When some or all of the artificial neural network 231a is implemented in software, one or more instructions constituting the artificial neural network 231a may be stored in the memory 230.
- the database 232 stores input data obtained from the input unit 220, training data (or training data) used for model training, and a model learning history.
- the input data stored in the database 232 may be not only processed data suitable for model learning, but also raw input data itself.
- the learning processor 240 may train (train, or learn) the artificial neural network 231a using training data or a training set.
- the learning processor 240 learns the artificial neural network 231a by immediately acquiring preprocessed data of the input data acquired by the processor 260 through the input unit 220 or acquires preprocessed input data stored in the database 232 Thus, the artificial neural network 231a can be trained.
- the learning processor 240 may determine optimized model parameters of the artificial neural network 231a by repeatedly learning the artificial neural network 231a using various learning techniques described above.
- an artificial neural network whose parameters are determined by being trained using training data may be referred to as a learning model or a trained model.
- the learning model may infer the result value while being mounted on the learning device 200 of the artificial neural network, and may be transmitted to and mounted on another device such as the artificial intelligence cleaner 100 through the communication unit 210.
- the updated learning model may be transmitted and mounted to another device such as the terminal 100 through the communication unit 210.
- the power supply unit 250 supplies power.
- FIG. 5 is a flowchart illustrating a method of operating an artificial intelligence cleaner according to an embodiment of the present invention.
- the processor 190 of the artificial intelligence cleaner 100 collects a plurality of cleaning records for a cleaning space (S501).
- the processor 190 may generate cleaning records through information collected from various sensors included in the artificial intelligence cleaner 100.
- various sensors may include an RGB camera, an RGB-D camera, and a cliff sensor.
- Each cleaning record may include at least one of a cleaning time required for each cleaning unit in the cleaning space, a cleaning status, a cleaning result value, a cleaning number, whether an obstacle is detected, or whether a restraint has occurred.
- the cleaning unit may mean a unit for dividing the cleaning space, and the size may use a preset default value.
- the cleaning unit may be 5cm x 5cm.
- the time required for cleaning may mean a time that the artificial intelligence cleaner 100 stays in a specific cleaning unit.
- the artificial intelligence cleaner 100 can facilitate cleaning, the time required for cleaning may be reduced.
- the artificial intelligence cleaner 100 may perform more operations such as direction change, and thus the time required for cleaning may be large.
- Whether to clean may mean whether the artificial intelligent cleaner 100 has ever arrived or has stayed in a specific cleaning unit.
- whether or not to clean can only indicate whether or not the time required for cleaning is 0, and in this case, whether or not to clean can be understood as a sub-concept of the time required for cleaning.
- the cleaning result value may indicate an amount of foreign matter/dust acquired through sensors in a specific cleaning unit, or may indicate a reduction amount of foreign matter/dust according to cleaning.
- the cleaning result value may appear good, and in a space where the artificial intelligence cleaner 100 is difficult to reach due to many obstacles, the cleaning result value may appear bad. have.
- the number of times of cleaning may mean how many times the artificial intelligence cleaner 100 has arrived or how many times has stayed in a specific cleaning unit.
- the cleaning frequency may be low.
- the passage connecting the two rooms has a high possibility of overlapping cleaning as the artificial intelligence cleaner 100 reciprocates the two rooms, and thus the cleaning frequency may be high.
- Whether the obstacle is detected may mean whether the artificial intelligence cleaner 100 detects an obstacle in a specific cleaning unit.
- whether or not an obstacle is detected may indicate not only whether there is an obstacle, but also whether there is a threshold or a height difference.
- whether an obstacle is detected may be determined as a value of 1 when an obstacle is detected, or as 0 when an obstacle is not detected.
- Whether or not restraint occurs may mean whether or not the artificial intelligence cleaner 100 can no longer proceed due to a certain factor in a specific cleaning unit.
- whether or not restraint occurs may mean whether the artificial intelligence cleaner 100 does not deviate from a plurality of adjacent cleaning units.
- the specific adjacent plurality of cleaning units are relatively small spaces compared to the entire cleaning space.
- the artificial intelligence cleaner 100 when the artificial intelligence cleaner 100 is caught in an electric wire and cannot move, or when it enters under the bed and cannot come out again, it may be determined to be restrained, and the value may be determined as 1.
- Each cleaning record can be collected based on the SLAM map for the cleaning space.
- information collected from sensors may be stored as a cleaning record together with the coordinates inside the cleaning space.
- the cleaning record may include coordinate information in the SLAM map for the cleaning space.
- the size of the cleaning unit serving as a reference in the cleaning record is different from the default value, information on the size value of the cleaning unit may be further included in the cleaning record.
- the cleaning unit defaults to 5cm x 5cm, but when a specific cleaning record is collected for a cleaning unit of 10cm x 10cm, the specific cleaning record may include 10cm x 10cm as size information of the cleaning unit.
- the processor 190 divides the cleaning space into a plurality of cleaning areas by using the collected cleaning records (S503).
- the reason why the processor uses the plurality of cleaning records when dividing the cleaning space into a plurality of cleaning areas is to prevent the cleaning areas from being divided into a distorted state due to a temporary obstacle by a user or a temporarily placed object.
- Each of the divided cleaning areas may be classified as one of a plurality of preset area types.
- the area types include a basic cleaning area, and at least one or more of an obstacle area, a constrained area, a complex area, or a passage area may be further included.
- the cleaning space may be divided into an area without obstacles and an area with many obstacles, an area without obstacles may be classified as a basic cleaning area, and an area with many obstacles may be classified as an obstacle area.
- an area that takes a long time to clean due to an obstacle or a structure of a space may be divided into a complex area.
- each of the divided cleaning areas may be divided so as not to include sub cleaning areas that are not connected to each other. This means that each of the separated cleaning areas is a continuous area.
- the plurality of areas may be separated from each other and divided into respective cleaning areas.
- the processor 190 determines a cleaning movement of the artificial intelligence cleaner 100 in consideration of the divided cleaning areas (S505).
- the processor 190 may set different cleaning modes for each of the divided cleaning areas.
- the cleaning mode may include a general cleaning mode, a simple cleaning mode, and a non-cleaning mode.
- the cleaning mode may further include a strong cleaning mode (or a thorough cleaning mode).
- a cleaning area classified as a general cleaning area may be cleaned in a general cleaning mode
- a cleaning area classified as an obstacle area may be cleaned in a simple cleaning mode
- the processor 190 may determine a cleaning movement line corresponding to each cleaning mode.
- the processor 190 may set the interval between cleaning movement lines in the simple cleaning mode to be wider than the interval between cleaning movement lines in the general cleaning mode. Likewise, the processor 190 may set the interval between the cleaning movement lines in the strong cleaning mode to be narrower than the interval between the cleaning movement lines in the general cleaning mode.
- the processor 190 may determine the number of times of cleaning the same point according to the cleaning mode.
- the processor 190 determines a cleaning flow line to perform at least one cleaning at one point in the simple cleaning mode, and determines a cleaning flow line to perform at least two cleaning at one point in the general cleaning mode, and For the cleaning mode, it is possible to determine the cleaning path to perform at least three cleanings at one point.
- the processor 190 may determine a cleaning movement line according to a cleaning priority order among the divided cleaning spaces.
- the processor 190 may determine a cleaning movement for cleaning a cleaning space of a priority priority first and cleaning a space of a later priority.
- the processor 190 controls the driving driving unit 170 of the artificial intelligence cleaner 100 according to the determined cleaning movement line (S507).
- the determined cleaning movement line may be stored in the memory 150, and the processor 190 may control the driving driving unit 170 so that the artificial intelligence cleaner 100 may move along the cleaning movement line.
- the artificial intelligence cleaner 100 divides the cleaning space into a plurality of cleaning areas and cleans along the cleaning flow line in consideration of the type of each cleaning area, thereby reducing the time required for cleaning, and constrained in the middle of cleaning. It can effectively prevent interruptions.
- FIG. 6 is a diagram showing an example of a cleaning record according to an embodiment of the present invention.
- FIG. 6 is a visualization of the collected cleaning time for each cleaning unit together with a SLAM map.
- the cleaning record may include the cleaning time required for each cleaning unit, and the cleaning time required for each cleaning unit may be recorded together with the coordinates of the SLAM map of the cleaning space.
- the cleaning record may include information on the time required for cleaning in the form of (coordinates, time required for cleaning).
- the artificial intelligence cleaner 100 may provide the user terminal with a SLAM map as shown in FIG. 6 and image information indicating the required cleaning time in each cleaning unit on the SLAM map.
- FIG. 7 is a diagram showing an example of a cleaning record according to an embodiment of the present invention.
- FIG. 7 shows a heat map reflecting the number of cleaning times and the required cleaning time with respect to the SLAM map in the cleaning space.
- the artificial intelligence cleaner 100 may provide a heat map for the SLAM map as shown in FIG. 7 to the user terminal.
- FIG. 8 is a diagram showing an example of a cleaning record according to an embodiment of the present invention.
- FIG. 8 is a visualization of whether or not cleaning for each cleaning unit or cleaning result value is collected together with a SLAM map.
- Whether to clean a specific cleaning unit may mean whether the artificial intelligence cleaner 100 has reached or stayed at the specific cleaning unit for cleaning. Alternatively, whether or not cleaning may mean whether or not cleaning is necessary.
- the cleaning result value may mean the amount or change of the foreign matter/dust collected through the sensor.
- the cleaning result value may mean a value indicating whether cleaning has been sufficiently performed depending on whether the amount of foreign matter/dust collected through the sensor is less than or equal to a reference value.
- the cleaning result value can be set to 1.
- the cleaning record may include whether to clean each cleaning unit or a cleaning result value, and whether each cleaning unit is cleaned or a cleaning result value may be recorded together with the coordinates of the SLAM map of the cleaning space.
- the cleaning record may include information on the required cleaning time in the form of (coordinates, cleaning status), (coordinates, cleaning result values) or (coordinates, cleaning status, cleaning result values).
- the cleaning record may include the cleaning time required for each cleaning unit, and the cleaning time required for each cleaning unit may be recorded together with the coordinates of the SLAM map of the cleaning space.
- the cleaning record may include information on the time required for cleaning in the form of (coordinates, time required for cleaning).
- the artificial intelligence cleaner 100 may provide the user terminal with a SLAM map as shown in FIG. 8 and image information indicating whether cleaning is performed in each cleaning unit or a cleaning result value on the SLAM map.
- FIG. 9 is a diagram illustrating an example of a cleaning record according to an embodiment of the present invention.
- FIG. 9 shows a SLAM map reflecting whether the cleaning space is cleaned or a cleaning result value.
- the artificial intelligence cleaner 100 may provide a SLAM map as illustrated in FIG. 9 to a user terminal.
- FIG. 10 is a flowchart illustrating an example of a step S503 of dividing the cleaning space shown in FIG. 5 into a plurality of cleaning areas.
- the processor 190 determines whether the cleaning area classification model is a deep learning-based model (S1001).
- the model for classifying the cleaning area may be a deep learning (or machine learning)-based model or a rule-based model.
- a deep learning (or machine learning)-based model may mean an artificial neural network learned by a deep learning (or machine learning) algorithm.
- the rule-based model may mean a model consisting of conditions for several variables.
- rule-based model may refer only to conditions for several variables, but may refer to the entire process of classifying cleaning areas using these conditions.
- the cleaning area classification model may be stored in the memory 150.
- the processor 190 may determine whether the cleaning area classification model is a deep learning-based model or a rule-based model according to a preset value or a user's selection.
- the processor 190 may classify the cleaning area using the mounted cleaning area classification model without a separate determination process.
- step S1001 when the cleaning area classification model is not a deep learning-based model, the processor 190 classifies the cleaning areas according to the rule-based model.
- the processor 190 classifies the area type of each cleaning unit according to whether a preset condition for information included in the cleaning records is satisfied (S1003).
- the processor 190 combines cleaning units having the same area type to generate one cleaning area (S1005).
- the processor 190 may combine cleaning units adjacent (or connected) to each other while having the same area type.
- the first to tenth cleaning units may be classified as a complex area
- the 11th to twentieth cleaning units are a passage area
- the remaining cleaning units may be classified as a general cleaning area.
- the processor 190 is the first to fifth cleaning units.
- the sixth to tenth cleaning units may be combined to generate another cleaning area.
- each cleaning area can be classified into one area type, and each cleaning area does not include separate (not connected) areas.
- step S1001 when the cleaning area classification model is a deep learning-based model, the processor 190 classifies the cleaning areas according to the deep learning-based model.
- the processor 190 obtains a map in which a plurality of cleaning areas are classified using a deep learning-based cleaning area classification model (S1005).
- S1005 deep learning-based cleaning area classification model
- the processor 190 inputs map data for the cleaning space and a cleaning record for the cleaning space for a deep learning-based cleaning area classification model.
- a map in which the cleaning space is divided into a plurality of cleaning areas is obtained as an output corresponding to the data input through the deep learning-based cleaning area classification model.
- a deep learning-based cleaning area classification model will be described with reference to FIG. 12.
- FIG. 11 is a diagram illustrating an example of a rule-based cleaning area classification model according to an embodiment of the present invention.
- FIG. 11 shows conditions for information included in cleaning records in a rule-based cleaning area classification model and a result of classification of area types according to conditions.
- the rule-based cleaning area classification model may be implemented through a combination of various conditions.
- the processor 190 determines whether the confinement ratio is greater than or equal to the first reference value th 1 as a first condition in the cleaning area classification model, and when the confinement ratio is greater than or equal to the first reference value, the cleaning unit may be classified as a constrained area. .
- the confinement ratio may mean a ratio of a cleaning record in which the robot cleaner has been confined in a corresponding cleaning unit among a plurality of collected cleaning records.
- the processor 190 determines whether the obstacle ratio is equal to or greater than the second reference value (th 2 ) as a second condition when the restraint ratio is less than the first reference value, and when the obstacle ratio is greater than the second reference value, the cleaning unit is It can be divided into obstacle areas.
- the obstacle ratio may mean a ratio of cleaning records in which an obstacle has been detected in a corresponding cleaning unit among a plurality of collected cleaning records.
- the processor 190 determines whether the average cleaning time is equal to or greater than the third reference value (th 3 ) as a third condition when the obstacle ratio is less than the second reference value, and when the average cleaning time is more than the third reference value, the corresponding cleaning Units can be divided into complex areas.
- the average cleaning time may mean an average value of the cleaning time required for a corresponding cleaning unit obtained from a plurality of collected cleaning records.
- the processor 190 determines whether the average cleaning number is equal to or greater than the fourth reference value (th 4 ) as the fourth condition, and when the average cleaning time is greater than the fourth reference value, the corresponding The cleaning unit is divided into a passage area, and when the average number of cleaning is less than the fourth reference value, the cleaning unit may be classified as a general cleaning area.
- the average cleaning time may mean an average value of the number of cleaning times for a corresponding cleaning unit obtained from a plurality of collected cleaning records.
- a cleaning unit to which the artificial intelligence cleaner 100 is frequently constrained is first classified as a constrained area, and a cleaning unit in which an obstacle is frequently recognized is classified as an obstacle area.
- cleaning units that take a long time to clean are classified as complex areas.
- Cleaning units that do not take a long time to clean but have a large number of cleaning times are classified as passage areas.
- Cleaning units that do not take a long time to clean and have a low cleaning frequency are classified as general cleaning areas.
- a frequency of cleaning records having specific data among a plurality of cleaning records may be used.
- the processor 190 may determine whether the frequency of cleaning records recorded as constrained by a specific cleaning unit is greater than or equal to a reference value. That is, the processor 190 may use various statistical values using a plurality of cleaning records.
- FIG. 12 is a diagram illustrating an example of a deep learning-based cleaning area classification model according to an embodiment of the present invention.
- a deep learning-based cleaning area classification model 1203 may be configured as an artificial neural network.
- the deep learning-based cleaning area classification model 1203 may be a personalized model trained individually for each user.
- the deep learning-based cleaning area classification model 1203 may be a segmentation model.
- the deep learning-based cleaning area classification model 1203 may be implemented as a model such as a Fully Convolutional Network (FCN), U-net, SegNet, and DeepLab.
- FCN Fully Convolutional Network
- U-net U-net
- SegNet SegNet
- DeepLab DeepLab
- the deep learning-based cleaning area classification model 1203 is a model that outputs a map in which a plurality of cleaning areas are divided when map data corresponding to the cleaning space and cleaning records 1201 and 1202 in the cleaning space are input. .
- the map corresponding to the cleaning space may be a SLAM map for the cleaning space.
- each cleaning record may include at least one or more of cleaning time required for each cleaning unit in the cleaning space, cleaning status, cleaning result value, cleaning number, detection of obstacles, or occurrence of restraint. .
- a map is included in the inputted cleaning record data itself, but the map data and the cleaning record may be input as separate data.
- a SLAM map as map data, and coordinates in the SLAM map and information about each coordinate may be input as data for cleaning records.
- the deep learning-based cleaning area classification model 1203 is trained using training data.
- the learning data may be composed of input data for learning and labeling information corresponding to the input data for learning.
- the input data for training may be composed of a map corresponding to the cleaning space and cleaning records in the cleaning space.
- the labeling information may be a map in which a plurality of cleaning areas are divided corresponding to input data for learning.
- the labeling information may be manually set by a user, a developer, or a designer.
- the labeling information may include an area type.
- the deep learning-based cleaning area classification model 1203 may be trained in a direction of narrowing the difference between the output data and labeling information corresponding to the training input data when input data for training included in the training data is input.
- the difference between the output data and the labeling information may be expressed as a loss function or a cost function, and the deep learning-based cleaning area classification model 1203 is trained in a direction to minimize the loss function.
- the deep learning-based cleaning area classification model 1203 outputs a map for classifying cleaning areas based on the area type, and thus, each cleaning area may be classified into one area type.
- the deep learning-based cleaning area classification model 1203 may be a model that is learned and stored by the processor 190 of the artificial intelligence cleaner 100, but is learned through the external learning device 200, and the wireless communication unit 140 It may be a model received through.
- the artificial intelligence cleaner 100 may obtain a user's feedback, generate labeling information corresponding to a corresponding cleaning record based on the user's feedback, and generate the corresponding cleaning record and labeling information as learning data.
- the generated training data may be used to update the deep learning-based cleaning area classification model 1203.
- the deep learning-based cleaning area classification model 1203 may also be updated through the processor 190 of the artificial intelligence cleaner 100, but may also be performed through an external learning device 200.
- FIG. 13 to 15 are views illustrating examples of cleaning spaces in which cleaning areas are separated from each other according to an embodiment of the present invention.
- FIG. 13 is a diagram illustrating a cleaning space divided into a general cleaning area 1301 and a complex area 1302 on a SLAM map.
- FIG. 14 is a diagram illustrating a cleaning space divided into a general cleaning area 1401 and a complex area 1402 on a heat map.
- 15 is a diagram illustrating a cleaning space divided into general cleaning areas 1501 and 1503 and a passage area 1502.
- a cleaning area that requires a lot of time for cleaning due to a complex structure due to an obstacle, etc. may be divided into a complex area, and a passage connecting several spaces may be divided into a passage area.
- an area having a high cleaning frequency despite not being an obstacle area, a constrained area, or a complex area may be classified as a passage area. That is, the passage area is located in the middle of several spaces, so that the artificial intelligence cleaner 100 moves between the spaces and repeatedly passes, and accordingly, the number of cleanings may be high in the passage area.
- 16 is an operation flowchart showing an example of a step S505 of determining a cleaning movement line shown in FIG. 5.
- the processor 190 determines whether or not setting information for priority between region types has been received (S1601).
- the setting information on the priority between the types of areas may be received in the step of determining the cleaning movement line, it may be received in advance and stored in the memory 150.
- the step S1601 may mean determining whether or not there is setting information about the priority between the types received in the region.
- step S1601 If there is setting information on the priority between the received area types as a result of the determination in step S1601, the processor 190 prioritizes cleaning for each of the divided cleaning areas using the setting information on the received priority. The ranking is determined (S1603).
- step S1601 If there is no setting information on the priority between the received area types as a result of the determination in step S1601, the processor 190 prioritizes cleaning for each of the divided cleaning areas by using the setting information for the preset priority. The ranking is determined (S1605).
- the setting information on the cleaning priority may be information for setting the general cleaning area as 1st priority, the passage area 2nd, the complex area 3rd, the obstacle area 4th, and the restraint area 5th.
- each of the divided cleaning areas is classified into one area type, and thus, a cleaning priority order between the divided cleaning areas may be determined using the cleaning priority order between the area types.
- the processor 190 determines whether setting information on the cleaning mode for each area type has been received (S1607).
- the setting information on the cleaning mode for each area type may be received in the step of determining the cleaning movement line, but may be received in advance and stored in the memory 150.
- step S1607 may mean determining whether or not setting information for a cleaning mode for each type of area received by the area exists.
- the processor 190 determines the cleaning mode for each of the separated cleaning areas using the received setting information on the cleaning mode. Do (S1609).
- step S1601 If there is no setting information on the cleaning mode for each type of the received area as a result of the determination in step S1601, the processor 190 determines a cleaning mode for each of the divided cleaning areas using the setting information for the preset cleaning mode. Do (S1611).
- the setting information on the cleaning mode may be information for setting a general cleaning area to a general cleaning mode, a passage area, a complex area to a simple cleaning mode, and an obstacle area and a constrained area to a non-cleaning mode.
- the processor 190 determines a cleaning movement line in consideration of the priority for each cleaning area, a cleaning mode for each cleaning area, and a proximity between the cleaning areas (S1613).
- Each cleaning area is classified into one area type, and a plurality of cleaning areas may be classified into the same area type. That is, even though different cleaning areas are classified into the same area type, the same cleaning priority order and cleaning mode may be set.
- the processor 190 may additionally determine the cleaning movement line in consideration of the proximity between the cleaning areas.
- the processor 190 cleans
- the movement line may be determined in the order of the first cleaning area, the second cleaning area, and the third cleaning area, or vice versa.
- the processor 190 may additionally consider the current location of the artificial intelligence cleaner 100 or the location of the charging station of the artificial intelligence cleaner 100 in determining the cleaning movement.
- the processor 190 determines the cleaning movement in the order of the third cleaning area, the second cleaning area, and the first cleaning area. I can. However, in this case as well, the processor 190 may determine the cleaning movement line in the order of the first cleaning area, the second cleaning area, and the third cleaning area.
- 17 and 18 are diagrams illustrating an example of setting a cleaning mode through a user input in an embodiment of the present invention.
- the processor 190 may provide a cleaning mode setting window 1702 to the user terminal 1701 through the wireless communication unit 140. Further, the processor 190 may receive setting information on a cleaning mode for each area type or a specific area type from the user terminal 1701 through the wireless communication unit 140.
- the cleaning areas divided for the cleaning space are displayed on the SLAM map (1703), and items 1704 for setting the cleaning mode for a specific area type (eg, constrained area) are displayed. Can be included.
- a specific area type eg, constrained area
- the cleaning mode setting window 1702 provided to the user terminal 1701 may be provided as a graphic user interface (GUI).
- GUI graphic user interface
- the items 1704 for setting the cleaning mode may include an item for setting the constrained area to a non-cleaning mode, an item for setting the constrained area to a simple cleaning mode, and an item for setting the constrained area to a general cleaning mode.
- the processor 190 outputs a voice 1801 asking a user 1802 for a cleaning mode for each area type or for a specific area type through a speaker (not shown), and transmits the microphone 120. Through this, a response 1803 for setting a cleaning mode for each area type of the user or for a specific area type may be received.
- 19 is a diagram illustrating an example of setting cleaning priority through user input in an embodiment of the present invention.
- the processor 190 may provide a priority setting window 1902 to the user terminal 1901 through the wireless communication unit 140.
- the processor 190 may receive setting information about cleaning priority for each area type or a specific area type from the user terminal 1901 through the wireless communication unit 140.
- the cleaning areas divided for the cleaning space are displayed on the SLAM map (1903), and items 1904 for setting the cleaning priority for a specific area type (eg, constrained area) May be included.
- a specific area type eg, constrained area
- the cleaning mode setting window 1902 provided to the user terminal 1901 may be provided as a graphic user interface (GUI).
- GUI graphic user interface
- the items 1904 that can set the cleaning priority include items that are set to clean without distinction of the constrained area, the item that gives the priority of the constrained area as 1, and the priority of the constrained area as the last priority. Items may be included.
- setting a specific area type to be cleaned without distinction of areas may mean not giving priority to the corresponding area type. That is, an area to which priority is not assigned is not compared with other area types.
- the processor 190 outputs a voice to the user through a speaker (not shown) asking for cleaning priority for each area type or for a specific area type, and the microphone 120 for each user's area type. Alternatively, a response for setting a cleaning priority for a specific area type may be received.
- 20 and 21 are diagrams showing examples of cleaning movement lines of the artificial intelligence cleaner 100 according to an embodiment of the present invention.
- the cleaning space shown in FIGS. 20 and 21 includes two rooms 2002 and 2003 and one passage 2004 positioned therebetween.
- FIG. 20 illustrates a case where the cleaning space is not divided into two rooms (2002, 2003) and a passage (2004), or a case where the cleaning priority between the two rooms (2002, 2003) and the passage (2004) does not differ from each other.
- a cleaning movement line 2001 is shown.
- the processor 190 does not distinguish between the two rooms 2002 and 2003 and the passage 2004 in determining the cleaning movement line 2001. Therefore, even if cleaning is started in the left room (2002), the left room (2002) and the right room (2003) are cleaned out of sequence while repeatedly passing through the passage (2004).
- the artificial intelligence cleaner 100 passes the passage 2004 several times, and accordingly, the number of times of cleaning may be unnecessarily increased in the passage 2004.
- the left room 2002 is cleaned, the right room 2003 and the passage 2004 are cleaned, so that the cleaning movement is inefficient.
- the user may have doubts about the reliability of the cleaner's operation by seeing that the left room 2002 has not been cleaned but moves to another space.
- the processor 190 divides the cleaning space into several cleaning areas (2002, 2003, and 2004), the cleaning areas 2002, 2003 and 2004 have the same cleaning priority. Or, when it is set not to give priority, as shown in FIG. 20, the moving line 2001 may be determined.
- FIG. 21 shows a cleaning flow line 2101 for a case where the cleaning space is divided into two rooms 2002 and 2003 and a passage 2004, the passage 2004 is divided into a passage area, and the cleaning priority of the passage area is set low. ).
- the processor 190 separates the two rooms 2002 and 2003 and the passage 2004 from each other in determining the cleaning movement line 2101. Therefore, when cleaning is started in the left room (2002), even if it is located adjacent to the passage (2004), the cleaning of the left room (2002) is first finished, and then, through the passage (2004), it moves to the right room (2003). (2003) and finally the passage (2004).
- the artificial intelligence cleaner 100 does not unnecessarily pass through the passage 2004 and may clean the passage 2004 with a lower priority for cleaning in a later order.
- cleaning is performed sequentially for the right room 2003 and the passage 2004, so that the cleaning movement is efficient.
- the present invention described above can be implemented as a computer-readable code in a medium on which a program is recorded.
- the computer-readable medium includes all types of recording devices that store data that can be read by a computer system. Examples of computer-readable media include HDD (Hard Disk Drive), SSD (Solid State Disk), SDD (Silicon Disk Drive), ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc. There is this.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Theoretical Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Mechanical Engineering (AREA)
- Aviation & Aerospace Engineering (AREA)
- Remote Sensing (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Robotics (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Fuzzy Systems (AREA)
- Business, Economics & Management (AREA)
- Game Theory and Decision Science (AREA)
- Image Analysis (AREA)
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
Abstract
Description
Claims (14)
- 인공 지능 청소기에 있어서,청소 공간의 SLAM(simultaneous localization and mapping) 지도를 저장하는 메모리;상기 인공 지능 청소기를 주행시키는 주행 구동부; 및상기 청소 공간에 대한 복수의 청소 기록들을 수집하고, 상기 SLAM 지도 및 상기 수집된 복수의 청소 기록들을 이용하여 상기 청소 공간을 복수의 청소 영역들로 구분하고, 상기 구분된 청소 영역들을 고려하여 상기 인공 지능 청소기의 청소 동선을 결정하고, 상기 결정된 청소 동선에 따라 상기 주행 구동부를 제어하는 프로세서를 포함하는, 인공 지능 청소기.
- 청구항 1에 있어서,상기 구분된 복수의 청소 영역들 각각은미리 설정된 복수의 영역 유형들 중에서 하나로 분류되고, 서로 연결되지 않는 서브 청소 영역들을 포함하지 않는, 인공 지능 청소기.
- 청구항 2에 있어서,상기 복수의 영역 유형들은장애물 영역, 구속 영역, 복잡 영역 또는 통로 영역 중에서 적어도 하나 이상과 기본 청소 영역을 포함하는, 인공 지능 청소기.
- 청구항 3에 있어서,상기 프로세서는상기 기본 청소 영역으로 분류된 제1 청소 영역을 우선적으로 청소하고, 그 이후에 상기 기본 청소 영역으로 분류되지 않는 제2 청소 영역을 청소하도록 상기 청소 동선을 결정하는, 인공 지능 청소기.
- 청구항 2에 있어서,상기 프로세서는상기 각 영역 유형 별 청소 모드를 구분하여 설정하고, 상기 설정된 청소 모드에 따라 상기 주행 제어부를 제어하고,상기 청소 모드는적어도 일반 청소 모드, 간단 청소 모드 및 비청소 모드를 포함하는, 인공 지능 청소기.
- 청구항 5에 있어서,상기 프로세서는상기 영역 유형들 사이의 우선 순위를 결정하고, 상기 결정된 우선 순위, 상기 설정된 청소 모드 및 상기 구분된 청소 영역들 간 인접도를 고려하여 상기 청소 동선을 결정하는, 인공 지능 청소기.
- 청구항 6에 있어서,사용자 단말기와 통신하는 통신부를 더 포함하고,상기 프로세서는상기 통신부를 통하여 상기 사용자 단말기에 상기 구분된 청소 영역들에 대한 정보를 제공하는, 인공 지능 청소기.
- 청구항 7에 있어서,상기 프로세서는상기 통신부를 통하여 상기 사용자 단말기로부터 상기 각 유형 별 청소 모드에 대한 청소 모드 설정 정보를 수신하면, 상기 수신한 청소 모드 설정 정보에 따라 상기 각 영역 유형 별 청소 모드를 설정하고, 상기 설정된 청소 모드에 따라 상기 주행 제어부를 제어하는, 인공 지능 청소기.
- 청구항 7에 있어서,상기 프로세서는상기 통신부를 통하여 상기 사용자 단말기로부터 상기 우선 순위에 대한 우선 순위 설정 정보를 수신하면, 상기 수신한 우선 순위 설정 정보에 따라 상기 영역 유형들 사이의 상기 우선 순위를 설정하고, 상기 설정된 우선 순위를 고려하여 상기 청소 동선을 결정하는, 인공 지능 청소기.
- 청구항 2에 있어서,상기 각 복수의 청소 기록들은상기 청소 공간에서의 각 청소 단위에 대한, 청소 소요 시간, 청소 여부, 청소 결과 값, 청소 횟수, 장애물 감지 여부 또는 구속 발생 여부 중에서 적어도 하나 이상을 포함하는, 인공 지능 청소기.
- 청구항 10에 있어서,상기 프로세서는상기 청소 공간을 상기 복수의 청소 영역들로 구분할 때, 상기 청소 소요 시간, 상기 청소 여부, 상기 청소 횟수, 상기 장애물 감지 여부 또는 상기 구속 발생 여부 중에서 적어도 하나 이상에 대한 미리 설정된 조건을 이용하여 상기 각 청소 단위를 상기 미리 설정된 영역 유형들 중 하나로 분류하는, 인공 지능 청소기.
- 청구항 10에 있어서,상기 프로세서는머신 러닝 알고리즘 또는 딥 러닝 알고리즘을 이용하여 학습된 영역 구분 모델에 상기 청소 공간에 상응하는 지도 데이터와 상기 청소 기록들을 입력시키고, 그 결과로써 상기 청소 공간이 상기 복수의 청소 영역들로 구분된 지도 데이터를 획득하고, 상기 획득한 지도 데이터를 이용하여 상기 복수의 청소 영역들에 대한 구분 정보를 획득하고,상기 학습된 영역 구분 모델은인공 신경망으로 구성되는, 인공 지능 청소기.
- 인공 지능 청소기의 동작 방법에 있어서,청소 공간에 대한 복수의 청소 기록들을 수집하는 단계;상기 청소 공간의 SLAM(simultaneous localization and mapping) 지도 및 상기 수집된 복수의 청소 기록들을 이용하여 상기 청소 공간을 복수의 청소 영역들로 구분하는 단계;상기 구분된 청소 영역들을 고려하여 상기 인공 지능 청소기의 청소 동선을 결정하는 단계; 및상기 결정된 청소 동선에 따라 상기 인공 지능 청소기를 주행시키는 주행 구동부를 제어하는 단계를 포함하는, 인공 지능 청소기의 동작 방법.
- 인공 지능 청소기의 동작 방법을 수행하기 위한 프로그램이 기록된 기록 매체에 있어서,상기 인공 지능 청소기의 동작 방법은청소 공간에 대한 복수의 청소 기록들을 수집하는 단계;상기 청소 공간의 SLAM(simultaneous localization and mapping) 지도 및 상기 수집된 복수의 청소 기록들을 이용하여 상기 청소 공간을 복수의 청소 영역들로 구분하는 단계;상기 구분된 청소 영역들을 고려하여 상기 인공 지능 청소기의 청소 동선을 결정하는 단계; 및상기 결정된 청소 동선에 따라 상기 인공 지능 청소기를 주행시키는 주행 구동부를 제어하는 단계를 포함하는, 기록 매체.
Priority Applications (5)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/KR2019/003636 WO2020196962A1 (ko) | 2019-03-28 | 2019-03-28 | 인공 지능 청소기 및 그의 동작 방법 |
EP19921990.8A EP3949817B1 (en) | 2019-03-28 | 2019-03-28 | Artificial intelligence cleaner and operation method thereof |
US16/491,118 US11399685B2 (en) | 2019-03-28 | 2019-03-28 | Artificial intelligence cleaner and method of operating the same |
AU2019437767A AU2019437767B2 (en) | 2019-03-28 | 2019-03-28 | Artificial intelligence cleaner and operation method thereof |
KR1020190089911A KR20190094318A (ko) | 2019-03-28 | 2019-07-24 | 인공 지능 청소기 및 그의 동작 방법 |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/KR2019/003636 WO2020196962A1 (ko) | 2019-03-28 | 2019-03-28 | 인공 지능 청소기 및 그의 동작 방법 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2020196962A1 true WO2020196962A1 (ko) | 2020-10-01 |
Family
ID=67624568
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/KR2019/003636 WO2020196962A1 (ko) | 2019-03-28 | 2019-03-28 | 인공 지능 청소기 및 그의 동작 방법 |
Country Status (5)
Country | Link |
---|---|
US (1) | US11399685B2 (ko) |
EP (1) | EP3949817B1 (ko) |
KR (1) | KR20190094318A (ko) |
AU (1) | AU2019437767B2 (ko) |
WO (1) | WO2020196962A1 (ko) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112950617A (zh) * | 2021-03-24 | 2021-06-11 | 上海海事大学 | 基于连续旋转图像增强和条件生成对抗网络的潮流机叶片附着物识别方法 |
Families Citing this family (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR102331563B1 (ko) * | 2019-08-23 | 2021-11-26 | 엘지전자 주식회사 | 구역별 오염정보를 이용하여 청소를 수행하는 인공 지능 로봇 및 그 방법 |
KR102255712B1 (ko) | 2019-08-26 | 2021-05-26 | 엘지전자 주식회사 | 오염로그를 이용하여 청소를 수행하는 인공 지능 로봇 및 그 방법 |
KR20210040613A (ko) * | 2019-10-04 | 2021-04-14 | 삼성전자주식회사 | 전자 장치 및 그의 제어 방법 |
US20210282613A1 (en) * | 2020-03-12 | 2021-09-16 | Irobot Corporation | Control of autonomous mobile robots |
KR20220004453A (ko) * | 2020-07-03 | 2022-01-11 | 삼성전자주식회사 | 객체를 인식하는 전자 장치 및 그 동작 방법 |
CN115429154A (zh) * | 2021-06-04 | 2022-12-06 | 尚科宁家(中国)科技有限公司 | 一种清洁机器人的清洁控制方法和清洁机器人 |
KR20240110586A (ko) * | 2021-12-03 | 2024-07-15 | 엘지전자 주식회사 | 인공 지능 청소기 및 그의 동작 방법 |
CN116919235A (zh) * | 2022-04-01 | 2023-10-24 | 追觅创新科技(苏州)有限公司 | 区域清洁的控制方法和装置、存储介质及电子装置 |
US20240180383A1 (en) * | 2022-12-06 | 2024-06-06 | Robert Bosch Gmbh | Lifelong robot learning for mobile robots |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR100664059B1 (ko) | 2004-12-04 | 2007-01-03 | 엘지전자 주식회사 | 로봇청소기의 장애물 위치 인식시스템 및 방법 |
KR20130030943A (ko) * | 2011-09-20 | 2013-03-28 | 엘지전자 주식회사 | 로봇 청소기 |
KR20130091879A (ko) * | 2012-02-09 | 2013-08-20 | 삼성전자주식회사 | 로봇 청소기의 청소 작업을 제어하기 위한 장치 및 방법 |
KR20150004568A (ko) * | 2013-07-03 | 2015-01-13 | 삼성전자주식회사 | 자율 이동 로봇의 위치 인식 방법 |
US20160313741A1 (en) * | 2013-12-19 | 2016-10-27 | Aktiebolaget Electrolux | Prioritizing cleaning areas |
KR20180134230A (ko) * | 2017-06-08 | 2018-12-18 | 삼성전자주식회사 | 청소 로봇 및 그 제어 방법 |
Family Cites Families (26)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE4408328C2 (de) * | 1994-03-11 | 2002-09-26 | Siemens Ag | Verfahren zum Aufbau einer zellular strukturierten Umgebungskarte von einer selbstbeweglichen mobilen Einheit, welche sich mit Hilfe von auf Wellenreflexion basierenden Sensoren orientiert |
DE19641261C1 (de) * | 1996-10-07 | 1998-02-12 | Siemens Ag | Verfahren zur Bewertung der Meßqualität eines Entfernungsmeßsensors an einem autonomen mobilen System |
SE0100924D0 (sv) * | 2001-03-15 | 2001-03-15 | Electrolux Ab | Energy-efficient navigation of an autonomous surface treatment apparatus |
US7015831B2 (en) * | 2002-12-17 | 2006-03-21 | Evolution Robotics, Inc. | Systems and methods for incrementally updating a pose of a mobile device calculated by visual simultaneous localization and mapping techniques |
US7805220B2 (en) * | 2003-03-14 | 2010-09-28 | Sharper Image Acquisition Llc | Robot vacuum with internal mapping system |
US20050010331A1 (en) * | 2003-03-14 | 2005-01-13 | Taylor Charles E. | Robot vacuum with floor type modes |
US20050273967A1 (en) * | 2004-03-11 | 2005-12-15 | Taylor Charles E | Robot vacuum with boundary cones |
US8930023B2 (en) * | 2009-11-06 | 2015-01-06 | Irobot Corporation | Localization by learning of wave-signal distributions |
US9250081B2 (en) * | 2005-03-25 | 2016-02-02 | Irobot Corporation | Management of resources for SLAM in large environments |
KR100877072B1 (ko) * | 2007-06-28 | 2009-01-07 | 삼성전자주식회사 | 이동 로봇을 위한 맵 생성 및 청소를 동시에 수행하는 방법및 장치 |
JP4788722B2 (ja) * | 2008-02-26 | 2011-10-05 | トヨタ自動車株式会社 | 自律移動ロボット、自己位置推定方法、環境地図の生成方法、環境地図の生成装置、及び環境地図のデータ構造 |
US8961695B2 (en) * | 2008-04-24 | 2015-02-24 | Irobot Corporation | Mobile robot for cleaning |
KR101633889B1 (ko) * | 2009-02-18 | 2016-06-28 | 삼성전자주식회사 | 격자지도를 이용한 경로 생성 장치 및 방법 |
KR20110119118A (ko) * | 2010-04-26 | 2011-11-02 | 엘지전자 주식회사 | 로봇 청소기, 및 이를 이용한 원격 감시 시스템 |
EP3327697A1 (en) * | 2010-09-24 | 2018-05-30 | iRobot Corporation | Systems and methods for vslam optimization |
US8798840B2 (en) * | 2011-09-30 | 2014-08-05 | Irobot Corporation | Adaptive mapping with spatial summaries of sensor data |
TWM451103U (zh) * | 2012-10-30 | 2013-04-21 | Agait Technology Corp | 行走裝置 |
US9020637B2 (en) * | 2012-11-02 | 2015-04-28 | Irobot Corporation | Simultaneous localization and mapping for a mobile robot |
DE102013207904A1 (de) * | 2013-04-30 | 2014-10-30 | Bayerische Motoren Werke Aktiengesellschaft | Bereitstellen einer effizienten Umfeldkarte für ein Fahrzeug |
KR102376150B1 (ko) * | 2015-03-05 | 2022-03-21 | 삼성전자주식회사 | 청소 로봇 및 그 제어 방법 |
KR102431996B1 (ko) * | 2015-10-12 | 2022-08-16 | 삼성전자주식회사 | 로봇 청소기 및 그 제어 방법 |
JP6559535B2 (ja) * | 2015-10-22 | 2019-08-14 | 株式会社東芝 | 障害物マップ生成装置、その方法、及び、そのプログラム |
WO2017200353A1 (ko) * | 2016-05-20 | 2017-11-23 | 엘지전자 주식회사 | 로봇 청소기 |
KR102235271B1 (ko) * | 2017-02-27 | 2021-04-01 | 엘지전자 주식회사 | 이동 로봇 및 그 제어방법 |
JP6899668B2 (ja) * | 2017-03-01 | 2021-07-07 | パナソニック インテレクチュアル プロパティ コーポレーション オブ アメリカPanasonic Intellectual Property Corporation of America | 自走型掃除機の制御方法、制御装置、制御プログラム及び自走型掃除機 |
KR102021833B1 (ko) * | 2017-09-26 | 2019-09-17 | 엘지전자 주식회사 | 인공지능을 이용한 이동 로봇 및 이동 로봇의 제어방법 |
-
2019
- 2019-03-28 US US16/491,118 patent/US11399685B2/en active Active
- 2019-03-28 EP EP19921990.8A patent/EP3949817B1/en active Active
- 2019-03-28 AU AU2019437767A patent/AU2019437767B2/en active Active
- 2019-03-28 WO PCT/KR2019/003636 patent/WO2020196962A1/ko unknown
- 2019-07-24 KR KR1020190089911A patent/KR20190094318A/ko not_active Application Discontinuation
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR100664059B1 (ko) | 2004-12-04 | 2007-01-03 | 엘지전자 주식회사 | 로봇청소기의 장애물 위치 인식시스템 및 방법 |
KR20130030943A (ko) * | 2011-09-20 | 2013-03-28 | 엘지전자 주식회사 | 로봇 청소기 |
KR20130091879A (ko) * | 2012-02-09 | 2013-08-20 | 삼성전자주식회사 | 로봇 청소기의 청소 작업을 제어하기 위한 장치 및 방법 |
KR20150004568A (ko) * | 2013-07-03 | 2015-01-13 | 삼성전자주식회사 | 자율 이동 로봇의 위치 인식 방법 |
US20160313741A1 (en) * | 2013-12-19 | 2016-10-27 | Aktiebolaget Electrolux | Prioritizing cleaning areas |
KR20180134230A (ko) * | 2017-06-08 | 2018-12-18 | 삼성전자주식회사 | 청소 로봇 및 그 제어 방법 |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112950617A (zh) * | 2021-03-24 | 2021-06-11 | 上海海事大学 | 基于连续旋转图像增强和条件生成对抗网络的潮流机叶片附着物识别方法 |
Also Published As
Publication number | Publication date |
---|---|
EP3949817A1 (en) | 2022-02-09 |
EP3949817B1 (en) | 2024-05-01 |
US11399685B2 (en) | 2022-08-02 |
AU2019437767B2 (en) | 2023-09-28 |
EP3949817A4 (en) | 2022-12-07 |
KR20190094318A (ko) | 2019-08-13 |
US20210330163A1 (en) | 2021-10-28 |
AU2019437767A1 (en) | 2021-11-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2020196962A1 (ko) | 인공 지능 청소기 및 그의 동작 방법 | |
WO2020213750A1 (ko) | 객체를 인식하는 인공 지능 장치 및 그 방법 | |
WO2019151845A2 (ko) | 에어컨 | |
WO2020218632A1 (ko) | 인공지능 장치 | |
WO2021010509A1 (ko) | 인공지능 조리 기기 | |
WO2021025217A1 (ko) | 인공지능 서버 | |
WO2020138624A1 (en) | Apparatus for noise canceling and method for the same | |
WO2018097574A1 (ko) | 이동 로봇 및 그 제어방법 | |
WO2018038553A1 (ko) | 이동 로봇 및 그 제어방법 | |
WO2021006405A1 (ko) | 인공지능 서버 | |
WO2020204221A1 (ko) | 공기 조화기 | |
WO2019151846A2 (ko) | 공기 청정기 | |
AU2019262468B2 (en) | A plurality of robot cleaner and a controlling method for the same | |
WO2018139796A1 (ko) | 이동 로봇 및 그의 제어 방법 | |
WO2021010757A1 (en) | Mobile robot and control method thereof | |
WO2021020621A1 (ko) | 인공지능 무빙 에이전트 | |
WO2020213758A1 (ko) | 음성으로 상호작용하는 인공 지능 장치 및 그 방법 | |
WO2020230933A1 (ko) | 사용자의 음성을 인식하는 인공 지능 장치 및 그 방법 | |
WO2019182265A1 (ko) | 인공 지능 기기 및 그의 동작 방법 | |
WO2021006404A1 (ko) | 인공지능 서버 | |
WO2021029457A1 (ko) | 사용자에게 정보를 제공하는 인공 지능 서버 및 그 방법 | |
WO2020138523A1 (ko) | 세탁물 처리 장치 및 그의 세탁 코스 결정 방법 | |
WO2019212240A1 (en) | A plurality of robot cleaner and a controlling method for the same | |
WO2020184748A1 (ko) | 교통 정보에 기반한 오토 스탑 시스템을 제어하는 인공 지능 장치 및 그 방법 | |
WO2020184753A1 (ko) | 음성 추출 필터를 이용하여 음성 제어를 수행하는 인공 지능 장치 및 그 방법 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 19921990 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
ENP | Entry into the national phase |
Ref document number: 2019921990 Country of ref document: EP Effective date: 20211028 |
|
ENP | Entry into the national phase |
Ref document number: 2019437767 Country of ref document: AU Date of ref document: 20190328 Kind code of ref document: A |