CN117734702A

CN117734702A - Parking space searching method, module and vehicle based on side ultrasonic radar

Info

Publication number: CN117734702A
Application number: CN202311782596.0A
Authority: CN
Inventors: 赵林峰; 李志刚; 李旭辰; 汪彧涛; 徐玉娟; 张绍康; 陈梓俊; 梅震; 丰肖; 裴石渊
Original assignee: Hefei University of Technology
Current assignee: Hefei University of Technology
Priority date: 2023-12-22
Filing date: 2023-12-22
Publication date: 2024-03-22

Abstract

The invention belongs to the field of automatic driving of automobiles, and particularly relates to a parking space searching method, a module and a vehicle based on a side ultrasonic radar. According to the scheme provided by the invention, the ultrasonic radar echo signals are firstly obtained and then mapped into the map space, the preliminary segmentation of the obstacle is realized according to the rising edge and the falling edge, the ultra-wide obstacle is distinguished, and the linear fitting and the pole recognition are carried out on the discrete signals of the ultra-wide obstacle, so that the secondary segmentation of the ultra-wide obstacle is realized. And finally, carrying out corner correction on the secondary segmentation result by combining with the FOV angle of the ultrasonic radar, generating virtual corners among the obstacles, connecting the boundaries of the obstacles, and finally searching for available parking spaces outside the boundaries of the obstacles. The method follows the strategy of generating the vertical parking spaces first and then generating the parallel parking spaces, can maximally find all available parking spaces, and solves the problems that the prior scheme cannot accurately distinguish the actual length of the obstacle and the judgment of the type of the parking spaces is inaccurate.

Description

Parking space searching method, module and vehicle based on side ultrasonic radar

Technical Field

The invention belongs to the field of automatic driving, and particularly relates to a parking space searching method and module based on a side ultrasonic radar, a corresponding automatic parking method and a corresponding vehicle.

Background

With the continuous development of technology, automobile automatic driving technology has become an important development direction of the automobile industry, and the automatic driving technology is mature at present, so that the large-scale commercial application is about to draw a preamble. Some applications in typical autopilot systems, such as full automatic parking APA, automatic emergency braking system AEB, etc., have been commercially deployed and become an important technical selling point for intelligent automobiles to attract customers.

The precondition for achieving automatic parking is that obstacle recognition is required for the environment around the vehicle, which requires the use of radar or vision sensors or the like arranged around the vehicle. The ultrasonic radar is waterproof and dustproof, even if a small amount of sediment is shielded, the influence is avoided, the influence from the environment is small, and the stability is good; the detection range is between 0.2 and 5.0 meters, and the precision is high and the real-time performance is good. The obstacle recognition method based on the ultrasonic radar has the advantages of relatively low cost, high detection precision and good real-time performance, but has some disadvantages. For example, current automatic parking systems based on ultrasonic radars tend to only care about the corner points of obstacles at two sides of a space parking space when searching for the parking space, and consider that the side reference obstacles of the parking space are in simple rectangular shapes, so that the corner points of the space parking space and the planned route when parking are calculated. In actual life, the obstacles at two sides of the parking space are often composite obstacles consisting of a plurality of rectangles, and the space provided for parking is not simple horizontal and vertical. This may lead to misidentification of the vehicle and release of some errant spaces that may interfere with the proper passage of other vehicles.

In addition, the detection range of the ultrasonic radar is a sector area, so that the identified obstacle is too wide when the obstacle detection is performed, a narrow parking space can not be identified, the actual range of the safety area can be reduced, and the parking path planning is failed. Therefore, how to further excavate the detection signal of the ultrasonic radar, to divide the types of the combined obstacle, accurately identify the real contour information of the obstacle, and accurately divide the obstacle space and the parking space is one of the problems that the sensing module of the automatic parking system needs to be researched.

Disclosure of Invention

The invention provides a parking space searching method and module based on a side ultrasonic radar, and a corresponding automatic parking method and a vehicle, aiming at solving the problems that the existing obstacle recognition technology based on the ultrasonic radar cannot accurately distinguish the type of an obstacle and the search result of the vehicle is inaccurate.

The invention is realized by adopting the following technical scheme:

a parking space searching method based on side ultrasonic radar utilizes detection data of the ultrasonic radar on the side of a vehicle to accurately position the position of an obstacle, and then the idle parking space is identified. The parking space searching method provided by the invention comprises the following steps:

S1: when the vehicle starts the parking space searching function, the echo signal of the lateral ultrasonic radar and the detection data of the inertial sensor are synchronously acquired. The original data point format of the echo signal is (t, y) _t )。

Wherein t represents the moment; y is _t Representing the lateral distance of the obstacle from the vehicle.

S2: and taking the vehicle driving direction as the X-axis direction and the vehicle lateral direction as the Y-axis, and establishing a map space xOy of the search area.

S3: combining the detection data of the inertial sensor, and integrating each data point (t, y) corresponding to each echo signal of the ultrasonic radar _t ) Mapping into map space to obtain each continuous signal point u _t ：u _t ＝(x _t ，y _t )。

Wherein x is _t And the signal point set U is formed by representing the running distance of the vehicle.

S4: identifying corner points corresponding to obstacle boundaries in the map space according to the signal point set U, wherein the process is as follows in detail:

s41: y based on adjacent signals _t The values identify the respective rising and falling edge signals in the set of signal points U.

S42: marking a signal point corresponding to the rising edge signal as a marked front corner b _j The signal point corresponding to the falling edge signal is marked as a rear corner point c _j The method comprises the steps of carrying out a first treatment on the surface of the The area between each group of front corner points and rear corner points is marked as an obstacle B _j 。

Wherein the subscript j denotes the number of the corner or obstacle.

S43: the respective obstacles detected in the signal point set U are classified into a normal obstacle and an ultra-wide obstacle according to the width of the obstacle.

S44: fitting all signal points corresponding to the ultra-wide obstacle to obtain a fitting curve s (t), and calculating a first-order derivative function s (t)', of the fitting curve.

S45: setting a threshold K, and when the neighborhood satisfies the following conditions: s (t) _i )'-s(t _i-1 )'|>K, consider that there is an obstacle boundary; where i denotes the sequence number of the signal points in the signal point set U, i=2, 3,4 … n.

Dividing the original ultra-wide barrier into two parts at each boundary, generating a corresponding rear corner point and a corresponding front corner point, and renumbering the subsequent corner points and the barriers.

S5: for each obstacle B according to the FOV angle of the ultrasonic radar _j Corresponding to the front corner b _j And rear corner point c _j Correcting to obtain accurate front corner point b _j 'and accurate rear corner point c' _j 。

S6: supplementing a virtual corner d between every two adjacent obstacles _j Sequentially connecting each accurate front corner point b _j ' accurate rear corner point c _j ' and virtual corner d _j A barrier boundary in the form of a square wave is obtained.

S7: the space from the boundary of the obstacle to one side of the vehicle running track is a safety space, and an available parking space is generated in the safety space according to the following strategy:

(1) When the gap width of any two barriers is larger than the length of the standard parking space, an available longitudinal parking space or a transverse parking space is generated at the corresponding position according to the types of the front parking space and the rear parking space.

(2) When the gap width of any two barriers is larger than the standard parking space width and smaller than the standard parking space length and the longitudinal parking spaces of both front and rear vehicles are met, an available longitudinal parking space is generated.

(3) When the gap between any two vehicles is insufficient to generate a parking space, continuously judging whether an ultra-wide barrier exists or not, and simultaneously meeting the following conditions:

a. the width of the ultra-wide barrier is larger than the length of the standard parking space,

b. the minimum distance between the ultra-wide obstacle and the current vehicle is larger than the width of a standard parking space;

if yes, an available transverse parking space is generated at the corresponding position;

otherwise, judging that no available parking space exists on the driving path.

As a further development of the invention, in step S1, the inertial sensor is used to detect the speed, acceleration and angle of rotation signals of the vehicle during its travel, the data format being as follows (v _t ，a _t ，β _t ). In step S3, the travel distance x of the vehicle _t The calculation formula of (2) is as follows:

as a further improvement of the present invention, in step S41, the method for judging the rising edge signal and the falling edge signal is as follows:

for any two adjacent signal points u _t And u _t+1 Defining y when no obstacle exists in the detection range of the ultrasonic radar _t =0, then:

(1) When meeting: y is _t > 0, and y _t+1 When=0, define u _t Is a falling edge signal;

(2) When meeting: y is _t =0, and y _t+1 At > 0, define u _t Is a rising edge signal.

As a further improvement of the present invention, in step S3, the method of dividing the normal obstacle and the ultra-wide obstacle is as follows:

(1) The width of a preset standard parking space is W0;

(2) Calculating the width W of each obstacle in the signal point set U by _j ：

W _j ＝x _j,2 -x _j,1

In the above, x _j,1 Represents the jth obstacle B _j Front corner b of (b) _j Is the abscissa of (2); x is x _j,2 Represents the jth obstacle B _j Back corner c of (2) _j Is defined by the abscissa of the (c).

(3) When any obstacle B _j Width W of (2) _j The method meets the following conditions: w (W) _j And if the energy is more than or equal to 2W0, defining the energy as an ultra-wide obstacle, otherwise, defining the energy as a conventional obstacle.

As a further improvement of the invention, in step S44, a least square method or a B spline curve fitting method is adopted to fit discrete signal points, so as to obtain a corresponding fitting curve.

As a further development of the invention, in step S5, the front corner b is refined _j 'and accurate rear corner point c' _j The update process of (2) is as follows:

s51: for any obstacle B _j Its corresponding front corner b _j And rear corner point c _j Is marked as (x) _j,1 ，y _j,1 ) And (x) _j,2 ，y _j,2 )。

S52: distance between front and rear corner points and ultrasonic radar is takenMinimum value of separation is the highest risk distance y _1min 。

S53: when the FOV angle of the ultrasonic radar is theta, the accurate front corner b is calculated by _j ' and accurate back corner point c _j ' abscissa x _j,1 ' and x _j,2 '：

S54: the corrected accurate front corner b _j 'and accurate rear corner point c' _j Is (x) _j,1 '，y _1m ) And (x) _j,2 '，y _1m )。

As a further improvement of the present invention, in step S6, the virtual corner d _j The generation method of (2) is as follows:

(1) For any two adjacent obstacles B _j And B _j+1 Acquisition of B _j The coordinates of the precise front corner point and the precise rear corner point are recorded as follows: (x) _j,1 ，y _2min ) And (x) _j,2 ，y _2min ) Acquisition of B _j+1 The coordinates of the precise front corner point and the precise rear corner point are recorded as follows: (x) _j+1,1 ，y _3min ) And (x) _j+1,2 ，y _3min )。

Wherein y is _2min Is an obstacle B _j Is the minimum of the echo distance, y _3min Is an obstacle B _j+1 Is the echo distance minimum of (2).

(2) Taking the abscissa of the larger one of the minimum echo distance values of the two corner points at the boundary of the obstacle as the abscissa of the virtual corner point, namely:

when y is _2min >y _3min The abscissa of the virtual corner point is x _j,2 ；

When y is _3min >y _2min Virtual corner d _j Is x in the coordinate of (2) _j+1,1 。

(3) The smaller one of the minimum echo distance values between two adjacent obstacles is taken as the ordinate of the virtual corner, namely:

when y is _2min >y _3min Virtual corner d _j Is y _2min ；

When y is _3min >y _2min Virtual corner d _j Is y _3min 。

The invention also comprises a parking space searching module based on the side ultrasonic radar, which comprises a memory, a processor and a computer program stored on the memory and running on the processor. When the processor executes the computer program, the steps of the parking space searching method based on the side ultrasonic radar are realized, and the idle parking spaces in the target area are identified.

The invention also includes an automatic parking method comprising the steps of:

(1) Firstly, a candidate parking space is obtained by adopting the parking space searching method based on the side ultrasonic radar.

(2) And then generating a parking path which reaches the candidate parking space from the current position and meets the collision-free constraint in the safety space by any path planning algorithm.

(3) And finally driving the vehicle to travel into the candidate parking space according to the planned parking path.

The invention also comprises a vehicle, wherein a computer program capable of driving the vehicle to automatically park is integrated in a control system of the vehicle, and the computer program realizes the automatic parking method when being executed.

The technical scheme provided by the invention has the following beneficial effects:

The method provided by the invention can overcome the problem of effectively losing parking spaces caused by the over-wide estimation of the obstacle by the traditional ultrasonic radar by correcting the accurate angular points. Meanwhile, the method combines the segmentation of the ultra-wide obstacle to realize the accurate classification of the type of the obstacle, thereby overcoming the problem that the normal communication of other vehicles is influenced by the error release of the transverse parking space.

When the parking space is generated in the safety area, the strategy of searching the parking space with a wide gap, searching the parking space with a narrow gap and searching the side parking space without gaps is followed, so that all available parking spaces can be searched to the maximum.

The parking space searching scheme provided by the invention can only rely on the detection signal of a single ultrasonic radar to achieve the effect of machine vision or human eye recognition. Compared with other existing schemes based on ultrasonic radars, the scheme is finer and more accurate, and is more in line with the evaluation of drivers on actual conditions, so that the application of the side ultrasonic radars in automatic parking and other auxiliary driving technologies can be greatly improved.

Drawings

Fig. 1 is a case image of a narrower parking space loss caused by conventional ultrasonic radar detection in embodiment 1 of the present invention.

Fig. 2 is a case image of a wrong release of a transverse parking space caused by conventional ultrasonic radar detection in embodiment 1 of the present invention.

Fig. 3 is a case image of a lateral parking space allowed to be generated in the detection scene corresponding to fig. 2.

Fig. 4 is a flow chart of steps of a parking space searching method based on a side ultrasonic radar according to embodiment 1 of the present invention.

Fig. 5 is a position distribution diagram of a map space of an established search area.

Fig. 6 is a schematic view of the positions of corner points characterizing the boundaries of a vehicle.

Fig. 7 is a signal diagram for distinguishing an obstacle from a rising edge and a falling edge of an ultrasonic radar echo signal.

Fig. 8 is a signal comparison diagram of a conventional obstacle and an ultra-wide obstacle in an ultrasonic radar echo signal.

Fig. 9 is a typical case image for realizing corner position revision in embodiment 1 of the present invention.

Fig. 10 is a comparison diagram of two types of barrier boundaries obtained before and after generating virtual corner points.

Fig. 11 is a logic block diagram of generating different types of available parking spaces according to embodiment 1 of the present invention.

Fig. 12 is a case image in which a longitudinal parking space can be generated in a wider gap.

Fig. 13 is a case image in which a lateral parking space can be generated in a wider gap.

Fig. 14 only allows the generation of case images of longitudinal parking spaces in a narrow gap.

Detailed Description

The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.

Example 1

In the prior art, two main defects generally exist when the ultrasonic radar is used for searching a parking space based on the inherent characteristics of an echo signal of the ultrasonic radar; the problems of the loss of the narrower parking space and the error release of the transverse parking space are respectively solved.

Fig. 1 is a view of a first defect. In fig. 1, the vehicle identifies a region on the left side along the travel route, and a sector region at the front left wheel of the vehicle is a detection range of an ultrasonic radar for identifying an obstacle mounted on the side of the vehicle. From the figures it can be seen that: the ultrasonic sensor has a certain FOV (fan angle), and the FOV of the ultrasonic sensor is defined as θ. Therefore, when the ultrasonic radar detects an obstacle on the side, the actual echo signal detects the obstacle before the ultrasonic radar is flush with the rear edge of the obstacle, and after the vehicle runs until the ultrasonic radar is flush with the rear edge of the obstacle, the obstacle can still be detected in the angle of FOV. Due to the presence of this situation, the width of the obstacle detected in the echo signal of the ultrasonic type is always larger than the actual width of the obstacle. For example, in fig. 1, the wire frame around the vehicle is the boundary of an obstacle detected by the ultrasonic radar, which is wider than the actual boundary of the vehicle.

The obstacle boundary detected by the ultrasonic radar is wider than the actual boundary, so that the safety redundancy of the vehicle is improved to a certain extent, and the collision of the vehicle is further avoided. However, in the field of parking space recognition and search, such inaccurate estimation of obstacle boundaries creates another disadvantage. For example, in fig. 1, the gap between B car and D car is large enough, and the gap between B and D is recognized enough as one complete parking space even if there is a "overestimate" situation of the vehicle by the ultrasonic radar, so C car can be parked between B car and D car. The gap between the a car and the B car is narrower than the gap between the B car and the D car, and although the gap between the a car and the B car is sufficient to park one car, the car can judge that the car cannot park between the a car and the B car in an actual detection signal because of the problem of 'overestimation' of the ultrasonic radar on the car, which results in that a search result can appear that a valid parking space is lost for a narrower parking space.

Fig. 2 is a view of a second defect. In fig. 2 it can be seen that the a-car and the B-car are parked further back than the C-car, which results in that the space at the front end of the a-car and the B-car can park exactly one car sideways. Because of the problem of over-wide estimation, the existing ultrasonic radar cannot distinguish between two normal vehicles that have been parked at A, B, but rather recognizes them as an ultra-wide obstacle. If, as shown in fig. 3, A, B is actually an ultra-wide obstacle, such as a wall or a transversely parked vehicle, then the identified empty space is correct, but in the scenario of fig. 2, it is obviously inappropriate to generate an empty space at the front ends of the vehicles a and B, which would affect the normal traffic of the vehicle A, B, and the identified space is incorrect.

The embodiment provides a parking space searching method based on a side ultrasonic radar, which utilizes detection data of the ultrasonic radar on the side of a vehicle to accurately position the position of an obstacle, so as to realize the identification of an idle parking space. According to the embodiment, on the basis of the existing scheme, the echo signals of the ultrasonic radar are further analyzed, and the defects that a narrower parking space cannot be accurately identified and a transverse parking space is wrongly released are overcome.

Specifically, as shown in fig. 4, the parking space searching method provided in this embodiment includes the following steps:

s1: when the vehicle starts the parking space searching function, the echo signal of the lateral ultrasonic radar and the detection data of the inertial sensor are synchronously acquired.

The detection principle of the ultrasonic radar is to judge the target area by generating ultrasonic waves into the detection environment and receiving echo signalsThe distance of the obstacle from itself, so that the original data point format of the obtained echo signal is (t, y _t ). Wherein t represents the transmission/reception timing of the signal; y is _t Representing the lateral distance of the obstacle from the vehicle.

In vehicles, inertial sensors are typically integrated with the bucket, which sense the movement of an object, detect linear displacement or angular rotation of the object, and convert this reaction into an electrical signal that is amplified and processed by electronic circuitry. Accelerometers and gyroscopes are the most common two broad classes of MEMS inertial sensors. An accelerometer is a sensor that detects axial acceleration and converts it into a usable output signal; a gyroscope is a sensor that is capable of detecting the angular velocity of motion of a motion body relative to an inertial space. The combination of three MEMS accelerometers and three MEMS gyroscopes results in a system that can achieve detection of linear acceleration and angle in 3 directions, i.e., an inertial sensor. In the present embodiment, inertial sensor is used to detect the speed v of the vehicle during driving _t Acceleration a _t And angle of rotation beta _t The signal, the data format obtained is as follows (v _t ，a _t ，β _t )。

In order to characterize the map space around the vehicle, as shown in fig. 5, the embodiment establishes a map space of a search area, in which the running direction of the vehicle is the X-axis direction and the directions in which the left and right sides of the vehicle extend outward are the Y-axis directions. In this coordinate system, all parking spaces are present in the respective quadrants of the coordinate system.

It should be specifically noted that the parking space searching algorithm is described based on a single ultrasonic radar installed on the left side of the vehicle. In the practical application process, the left side and the right side of the vehicle can be provided with the existing ultrasonic radar, and the search can be synchronously started at the left side and the right side of the vehicle. For ease of illustration, the following description will proceed with the example of a single-sided search scheme based on a single ultrasonic radar.

S3: the map space xOy established in step S2 is a two-dimensional map space, and in step S1, the present embodiment obtains time-domain synchronized echo data of the ultrasonic radar and detection data of the inertial sensor, and the moment of signal and distance information along the Y-axis are recorded in the data point of the echo signal of each ultrasonic radar, and the travel distance information of the signal along the X-axis can be obtained according to the inertial sensor. Therefore, in combination with the detection data of the inertial sensor, the data point (t, y) corresponding to each echo signal of the ultrasonic radar can be obtained _t ) Mapping into map space to obtain each continuous signal point u _t : and forming a signal point set U. The data format of the converted signal points is as follows:

u _t ＝(x _t ，y _t )。

wherein x is _t Representing the distance travelled by the vehicle in the X-axis direction, y _t Indicating the distance between the obstacle and the vehicle in the Y-axis direction. Considering that the vehicle speed is slow in the process of searching for a parking space and basically keeps constant speed, the driving distance x of the vehicle in the embodiment _t The calculation formula of (2) is as follows:

in this way, the present embodiment maps the detection signal of the ultrasonic radar into the map space representing the parking space search area, so that the obstacle and the available space in the map space xOy are identified and analyzed in the subsequent process.

S4: in the conventional parking space recognition scheme, the ultrasonic radar recognizes the area around the vehicle instead of the actual boundary of the vehicle, and the embodiment can recognize the actual boundary of the vehicle, so that the problem of over-wide estimation in the conventional scheme is solved. In addition, in order to save computational resources, the method adopted in identifying the obstacle in this embodiment does not depict the complete outline of the obstacle according to the echo signal. The strategy shown in fig. 6 is adopted to identify the boundaries at two sides of the front end of the vehicle, and the corner points (points at two sides of the vehicle head in fig. 6) corresponding to the obstacles are recorded; and finally, taking the connecting line of the two corner points as the real boundary of the obstacle. The way in which the corner points are used to estimate the boundaries of obstacles is not universal, but in the field of space searching this is very efficient because the front end of the vehicle is essentially flush, and building structures in the space of the space, such as pillars and walls, generally conform to this situation. Therefore, the obstacle boundary estimation strategy for only recording the corner points is reasonable and effective in the parking space searching field.

According to the embodiment, the corner points of all corresponding obstacle boundaries in the map space can be identified according to the signal point set U, and the process is as follows in detail:

In this embodiment, the method for determining the rising edge signal and the falling edge signal is as follows:

In colloquial terms, the ultrasonic radar can receive the echo signal (y _t > 0), but at the next moment the echo signal disappears (y _t+1 =0), the signal at this time is a falling edge signal, which indicates that the ultrasonic radar has left from the obstacle.

Conversely, when the ultrasonic radar does not receive the echo signal (y _t =0), but at the next moment the echo signal appears (y _t+1 > 0), the signal at this time is a rising edge signal, which means that the ultrasonic radar has just encountered a new obstacle at this time.

S42: as can be seen from the foregoing, the rising edge refers to the obstacle just beginning to appear, and the falling edge refers to the obstacle disappearing, so the area between the rising edge and the falling edge is the area of the obstacle. Therefore, as shown in fig. 7, the signal points corresponding to the rising edge signal are marked as the marked front corner b in the present embodiment _j The signal point corresponding to the falling edge signal is marked as a rear corner point c _j The method comprises the steps of carrying out a first treatment on the surface of the The area between each group of front corner points and rear corner points is marked as an obstacle B _j . Wherein the subscript j denotes the number of the corner or obstacle.

S43: in the process of identifying obstacles according to the rising edge and the falling edge of the signal, the situation that the width of some obstacles is very wide occurs, for example, in fig. 1, the echo signal of the a car is similar to the signal pattern in fig. 7, while the echo signals of the B, C, D cars are not three independent standard signals comprising the rising edge and the falling edge, because the front end of the C car is already present in the echo signals before the right end of the B car leaves the detection area of the ultrasonic radar, and the echo signal of the ultrasonic radar does not present the rising edge or the falling edge at the junction of the two, but shows that a very wide obstacle is encountered. The scenario of fig. 1 is thus in the pattern shown in fig. 8 in the ultrasonic radar detection signal, and the detection result shows that there is one obstacle with a wider width and one obstacle with a narrower width on the vehicle side.

In order to divide the types of the obstacles, the present embodiment first divides each obstacle detected in the signal point set U into a normal obstacle and an ultra-wide obstacle according to the width of the obstacle. The method for dividing the conventional obstacle and the ultra-wide obstacle is shown in fig. 9, and comprises the following steps:

(1) The width of a preset standard parking space is W0;

W _j ＝x _j,2 -x _j,1

In this embodiment, the obstacle with a width larger than the width of two standard parking spaces is regarded as an ultra-wide obstacle, and the obstacle with a width smaller than the width of two standard parking spaces is regarded as a conventional obstacle. The criteria of this division is mainly to facilitate the later recognition of whether an ultra-wide obstacle displayed in an echo signal is a combination of a plurality of common vehicles approaching each other or a real large-sized obstacle. In general, an obstacle with a width less than two standard parking spaces needs not to be continuously identified, and can be identified as the same obstacle, and a plurality of situations may exist when the width exceeds the widths of the two standard parking spaces, and the special situations will be further identified later in this implementation.

The region between a rising edge signal and a falling edge signal may be an obstacle or a combination of obstacles, and thus, further segmentation of the ultra-wide obstacle is required. In an actual parking system, the echo point set is the distance feedback from the vehicle to the obstacle, and as shown in fig. 8, the echo data may reflect the local contour of the obstacle, so the present embodiment further performs the segmentation of the contour of the obstacle according to the data change of the echo point set.

The implementation of the barrier segmentation method in this embodiment is: firstly, fitting discrete signal points corresponding to the ultra-wide obstacle into a continuous curve, then finding out a signal abrupt change part (pole) in a smooth curve, namely a part where two obstacles are connected, and if the balance curve fitted by the ultra-wide obstacle is the smooth curve, the part (pole) without the signal abrupt change is the object, and then explaining that the obstacle is a complete obstacle. The specific data processing procedure of the obstacle segmentation method is realized through steps S44 and S45.

S44: fitting all signal points corresponding to the ultra-wide obstacle to obtain a fitting curve s (t), wherein the embodiment can use a least square method or a B spline curve fitting method to fit discrete signal points to obtain a corresponding fitting curve.

The least squares method determines the coefficients of the model by minimizing the sum of squares of the errors. For a model with n data points, its least squares method can be expressed as:wherein y is _i Is the actual point, f (x _i ) Is a fitting value. The B-spline curve fitting method is a polynomial-based method that segments intervals, which can be expressed as:

wherein P is _i (x) Is the i th segment polynomial +.>To indicate a function.

Of course, in addition to the two ways, any tool or method that can fit discrete data to a curve can be used for processing. After the fitted curve is obtained, the first order derivative function s (t)', of the fitted curve is further calculated.

S45: next, a threshold K is set, when the neighborhood satisfies: s (t) _i )'-s(t _i-1 )'|>K, consider that there is an obstacle boundary; where i denotes the sequence number of the signal points in the signal point set U, i=2, 3,4 … n.

In this embodiment, the original ultra-wide obstacle is divided into two parts at each boundary, a corresponding rear corner point and front corner point are generated, and the subsequent corner points and obstacles are renumbered. For example, in fig. 8, two boundaries can be identified in the ultra-wide obstacle, dividing the ultra-wide obstacle into 3 regular obstacles, i.e., 4 obstacles and 4 diagonal points in fig. 8.

S5: for each obstacle B according to the FOV angle of the ultrasonic radar _j Corresponding to the front corner b _j And rear corner point c _j Correcting to obtain accurate front corner pointb _j ' and accurate back corner point c _j ′。

As can be seen from fig. 1 and 2, the corner points extracted from the echo signal are not real boundaries of the obstacle, and the real corner points corresponding to each obstacle should be inside two corner points in the image of the echo signal due to the existence of the phenomenon of "overestimation" of the ultrasonic radar. The effect of this step is to correct the true positions of all pairs of corner points that appear, separated in step S4.

Specifically, the exact front corner b _j ' and accurate back corner point c _j The update procedure of' is as follows:

S52: taking the minimum value of the distance between the front corner point and the rear corner point and the ultrasonic radar as the highest risk distance y _1min 。

The intrinsic logic of the corner correction of this embodiment is: firstly, unifying the ordinate of the front corner point and the rear corner point as the minimum value of the distance detected by the ultrasonic radar, namely the highest risk distance y _1min . Then, based on the highest risk distance y _1min And calculating the deviation amount of the front foot point and the rear corner point in the echo signal in the X-axis direction by the FOV angle of the ultrasonic radar. And finally, correcting the abscissa of the front corner point and the rear corner point with errors by utilizing the deviation amount. Thereby obtaining an obstacleA true exact front corner and an exact rear corner of the object.

Fig. 9 is a case image of corner position correction, the upper half of the figure is the front corner and rear corner of three vehicles determined according to radar echo signals, and the lower figure is the combined risk maximum distance y _1min And front and rear corner points of three vehicles after FOV angle correction of the ultrasonic radar. As is evident from the figure, the radar signal has the phenomenon of "overestimation" of the obstacle, and the corrected corner point is closer to the real boundary of the vehicle.

S6: by utilizing the steps, accurate front corner points and accurate rear corner points of each obstacle encountered on a driving path can be identified according to the signal point set U corresponding to the radar echo signal, and the boundary of one obstacle can be approximately obtained by connecting the corner points. But the barrier boundaries obtained by such corner points being connected are not in fact accurate. This is because if the Y-axis directions of the two vehicles are not flush, the head-to-head connection will divide the triangle at the junction of the two vehicles into the boundaries of the obstacle, which reduces the range of the vehicle safety area. Thereby affecting the planning of the parking path.

As shown in fig. 10, in order to overcome the problem that the false boundary where corner points are connected first easily divides the inter-vehicle triangle into boundaries, the present embodiment introduces the concept of virtual corner points between adjacent obstacles. Supplementing a virtual corner d between every two adjacent obstacles _j Sequentially connecting each accurate front corner point b _j ' accurate rear corner point c _j ' and virtual corner d _j A barrier boundary in the form of a square wave is obtained. The barrier boundary of the right-angle wave shape is attached to the real situation, and no redundant triangular area exists.

Specifically, in this embodiment, the virtual corner d _j The generation method of (2) is as follows:

(1) For any two adjacent obstacles B _j And B _j+1 Acquisition of B _j The coordinates of the precise front corner point and the precise rear corner point are recorded as follows: (x) _j,1 ，y _2min ) And (x) _j,2 ，y _2min ) Acquisition of B _j+1 Is a precise front corner and a precise rear corner of (a)Is noted as: (x) _j+1,1 ，y _3min ) And (x) _j+1,2 ，y _3min )。

when y is _2min >y _3min The abscissa of the virtual corner point is x _j,2 ；

when y is _2min >y _3min Virtual corner d _j Is y _2min ；

When y is _3min >y _2min Virtual corner d _j Is y _3min 。

S7: the space from the boundary of the obstacle to one side of the vehicle running track is a safe space, and the available parking space is generated in the safe space according to the strategy shown in fig. 11, and the specific contents are as follows:

When the space between two obstacles is detected to be longer than the standard parking space, the vehicle can be stopped vertically at the position, and can also be stopped transversely at the position. The specific type of parking space to be generated depends on the parking mode of the front and rear obstacles, and if the front and rear vehicles are both conventional obstacles, it is indicated that they are both parked longitudinally, and a longitudinal parking space should be generated at this time, as shown in fig. 12. As shown in fig. 13, if both front and rear vehicles are ultra-wide obstacles, it is indicated that they are both parked laterally, and a lateral parking space should be generated at this time.

As shown in fig. 14, this criterion refers to that if the gap between two vehicles allows only one vehicle to be parked longitudinally, and the vehicles on both sides are also parked longitudinally, a longitudinal parking space can be created.

otherwise, judging that no available parking space exists on the driving path.

This criterion corresponds to the situation of fig. 3, where if there is no gap between adjacent vehicles but there is an ultra-wide obstacle far from itself, it is stated that it is possible here to allow a transverse parking space to be set, the triggering condition being that the space is longer in width than the standard parking space and deeper than the standard parking space.

Therefore, the method provided by the invention can overcome the problem of effectively losing parking spaces caused by the over-wide estimation of the obstacle by the traditional ultrasonic radar by correcting the accurate angular points. Meanwhile, the method combines the segmentation of the ultra-wide obstacle to realize the accurate classification of the type of the obstacle, thereby overcoming the problem that the normal communication of other vehicles is influenced by the error release of the transverse parking space. Compared with other schemes, the parking space searching scheme provided by the embodiment is finer and more accurate, and is more in line with the evaluation of the actual conditions by drivers, so that the application of the side ultrasonic radar in the auxiliary driving technology such as automatic parking can be greatly improved.

Example 2

The embodiment provides a parking space searching module based on a side ultrasonic radar, which comprises a memory, a processor and a computer program stored on the memory and running on the processor. When the processor executes the computer program, the steps of the parking space searching method based on the side ultrasonic radar according to embodiment 1 are implemented, and the idle parking space in the target area is identified.

The parking space searching module based on the side ultrasonic radar of the mirror is essentially a computer device for realizing data processing and instruction generation, and comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor. The computer device provided in this embodiment may be an embedded model capable of executing a computer program, or may be an intelligent terminal capable of executing a program, such as a tablet computer, a notebook computer, a desktop computer, a rack-mounted server, a blade server, a tower server, or a rack-mounted server (including an independent server or a server cluster formed by multiple servers), or the like. The computer device of the present embodiment includes at least, but is not limited to: a memory, a processor, and the like, which may be communicatively coupled to each other via a system bus.

In this embodiment, the memory (i.e., readable storage medium) includes flash memory, hard disk, multimedia card, card memory (e.g., SD or DX memory, etc.), random Access Memory (RAM), static Random Access Memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory may be an internal storage unit of a computer device, such as a hard disk or memory of the computer device.

In other embodiments, the memory may also be an external storage device of a computer device, such as a plug-in hard disk, smart Media Card (SMC), secure Digital (SD) Card, flash Card (Flash Card) or the like, which are provided on the computer device. Of course, the memory may also include both internal storage units of the computer device and external storage devices. In this embodiment, the memory is typically used to store an operating system and various application software installed on the computer device. In addition, the memory can be used to temporarily store various types of data that have been output or are to be output.

The processor may be a central processing unit (Central Processing Unit, CPU), an image processor GPU (Graphics Processing Unit), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor is typically used to control the overall operation of the computer device. In this embodiment, the processor is configured to execute the program code stored in the memory or process the data.

Example 3

The embodiment provides an automatic parking method, which comprises the following steps:

(1) A candidate parking space is first obtained by using the parking space searching method based on the side ultrasonic radar according to embodiment 1.

Example 4

The present embodiment provides a vehicle, in which a computer program capable of driving the vehicle to automatically park is integrated in a control system of the vehicle, and when the computer program is executed, the automatic parking method as in embodiment 3 is implemented.

The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.

Claims

1. The utility model provides a parking stall searching method based on side ultrasonic radar, its detection data that utilizes the ultrasonic radar of vehicle side carries out accurate location to the position of barrier, and then realizes discerning idle parking stall, its characterized in that, parking stall searching method includes following steps:

s1: when the vehicle starts the parking space searching function, synchronously acquiring the echo signal of the lateral ultrasonic radar and the detection number of the inertial sensorAccording to the above; the original data point format of the echo signal is (t, y) _t ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein t represents the moment; y is _t Representing the lateral distance of the obstacle from the vehicle;

s2: taking the driving direction of the vehicle as the X-axis direction and the lateral direction of the vehicle as the Y-axis, and establishing a map space XOY of the search area;

s3: combining the detection data of the inertial sensor, and integrating each data point (t, y) corresponding to each echo signal of the ultrasonic radar _t ) Mapping into map space to obtain each continuous signal point u _t ：u _t ＝(x _t ，y _t )，x _t A signal point set U which represents the running distance of the vehicle and is formed;

s4: identifying corner points corresponding to obstacle boundaries in the map space according to the signal point set U, wherein the process is as follows:

s41: y based on adjacent signals _t The value identifies each rising edge signal and falling edge signal in the signal point set U;

S42: marking a signal point corresponding to the rising edge signal as a marked front corner b _j The signal point corresponding to the falling edge signal is marked as a rear corner point c _j The method comprises the steps of carrying out a first treatment on the surface of the The area between each group of front corner points and rear corner points is marked as an obstacle B _j The method comprises the steps of carrying out a first treatment on the surface of the Subscript j denotes the number of the corner or obstacle;

s43: dividing each obstacle detected in the signal point set U into a conventional obstacle and an ultra-wide obstacle according to the width of the obstacle;

s44: fitting all signal points corresponding to the ultra-wide obstacle to obtain a fitted curve s (t), and calculating a first-order derivative function s (t)', of the fitted curve;

s45: setting a threshold K, and when the neighborhood satisfies the following conditions: s (t) _i )'-s(t _i-1 )'|>When K, considering that an obstacle boundary exists at the position, dividing the original ultra-wide obstacle into two parts at each boundary, generating a corresponding rear corner point and a corresponding front corner point, and renumbering the subsequent corner points and the obstacles;

wherein i represents the sequence number of the signal points in the signal point set U, i=2, 3,4 … n;

s5: for each of the FOV angles of the ultrasonic radarObstacle B _j Corresponding to the front corner b _j And rear corner point c _j Correcting to obtain accurate front corner point b _j 'and accurate rear corner point c' _j ；

S6: supplementing a virtual corner d between every two adjacent obstacles _j Sequentially connecting each accurate front corner point b _j 'accurate rear corner point c' _j And virtual corner d _j Obtaining an obstacle boundary in a right-angle wave shape;

s7: the space from the obstacle boundary to one side of the vehicle running track is a safety space, and an available parking space is generated in the safety space according to the following strategy:

(1) When the gap width of any two barriers is larger than the length of a standard parking space, an available longitudinal parking space or a transverse parking space is generated at the corresponding position according to the types of the front parking space and the rear parking space;

(2) When the gap width of any two obstacles is larger than the standard parking space width and smaller than the standard parking space length and the longitudinal parking spaces of both front and rear vehicles are met, an available longitudinal parking space is generated;

otherwise, judging that no available parking space exists on the driving path.

2. The side ultrasonic radar-based parking space searching method according to claim 1, wherein: in step S1, the inertial sensor is used to detect the speed, acceleration and rotation angle signals during the running of the vehicle, and the data format is as follows (v _t ，a _t ，β _t ) The method comprises the steps of carrying out a first treatment on the surface of the In step S3, the travel distance x of the vehicle _t The calculation formula of (2) is as follows:

3. the side ultrasonic radar-based parking space searching method according to claim 1, wherein: in step S41, the method for determining the rising edge signal and the falling edge signal is as follows:

4. The side ultrasonic radar-based parking space searching method according to claim 1, wherein: in step S3, the method for dividing the normal obstacle and the ultra-wide obstacle is as follows:

(1) The width of a preset standard parking space is W0;

W _j ＝x _j,2 -x _j,1

In the above, x _j,1 Represents the jth obstacle B _j Front corner b of (b) _j Is the abscissa of (2); x is x _j,2 Represents the jth obstacle B _j Back corner c of (2) _j Is the abscissa of (2);

5. The side ultrasonic radar-based parking space searching method according to claim 1, wherein: in step S44, a least square method or a B-spline curve fitting method is adopted to fit the discrete signal points, so as to obtain a corresponding fitting curve.

6. The side ultrasonic radar-based parking space searching method according to claim 1, wherein: in step S5, the precise front corner b _j 'and accurate rear corner point c' _j The update process of (2) is as follows:

s51: for any obstacle B _j Its corresponding front corner b _j And rear corner point c _j Is marked as (x) _j,1 ，y _j,1 ) And (x) _j,2 ，y _j,2 )；

S52: taking the minimum value of the distance between the front corner point and the rear corner point and the ultrasonic radar as the highest risk distance y _1min ；

S53: when the FOV angle of the ultrasonic radar is theta, the accurate front corner b is calculated by _j 'and accurate rear corner point c' _j X of the abscissa of (2) _j,1 ' and x _j,2 '：

S54: the corrected accurate front corner b _j ' and accurate back corner point c _j ' has the coordinates (x _j,1 '，y _1m ) And (x) _j,2 '，y _1m )。

7. The side ultrasonic radar-based parking space searching method according to claim 6, wherein: in step S6, virtual corner d _j The generation method of (2) is as follows:

(1) For any two adjacent obstacles B _j And B _j+1 Acquisition of B _j The coordinates of the precise front corner point and the precise rear corner point are recorded as follows: (x) _j,1 ，y _2min ) And (x) _j,2 ，y _2min ) Acquisition of B _j+1 The coordinates of the precise front corner point and the precise rear corner point are recorded as follows: (x) _j+1,1 ，y _3min ) And (x) _j+1,2 ，y _3min )；

Wherein y is _2min Is an obstacle B _j Is set to be the minimum value of the echo distance,y _3min is an obstacle B _j+1 Is the echo distance minimum of (2);

When y is _2min >y _3min The abscissa of the virtual corner point is x _j,2 ；

When y is _3min >y _2min Virtual corner d _j Is x in the coordinate of (2) _j+1,1 ；

when y is _2min >y _3min Virtual corner d _j Is y _2min ；

When y is _3min >y _2min Virtual corner d _j Is y _3min 。

8. A parking space searching module based on a side ultrasonic radar, which comprises a memory, a processor and a computer program stored on the memory and running on the processor, and is characterized in that: when the processor executes the computer program, the steps of the parking space searching method based on the side ultrasonic radar according to any one of claims 1-7 are realized, and the idle parking space in the target area is identified.

9. An automatic parking method, characterized in that it comprises the following steps:

(1) Firstly, acquiring a candidate parking space by adopting the parking space searching method based on the side ultrasonic radar according to any one of claims 1-7;

(2) Generating a parking path which reaches the candidate parking space from the current position and meets the collision-free constraint in the safety space through any path planning algorithm;

(3) And driving the vehicle to travel into the candidate parking space according to the parking path.

10. A vehicle having a computer program integrated into its control system for driving the vehicle to automatically park, the computer program, when executed, implementing the automatic parking method of claim 9.