The main objective of the project is to create a robot capable of touring a OU facility. For the achievement of this goal, we will use the ROS robotics framework.

For our implementation, we have decided to follow a hybrid approach to our robotical architecture. We will implement a 3T-based system where we have 3 main layers:

1. Deliberative:

• Our robot will use self-organizing snn's to create a representation of the world, and using an algorithm similar to A* the path planning will create dynamic plans.

2. Reactive:

• Alongside the robotical Executive module, there will be a reactive part to the robot where specific events will cause an override of actions and cause specific actions to happen.
  • Bumping against a wall.
  • Getting too close to an obstacle. (turn away or fly away.)
  • stop signal?

3. Coordinating layer:

• For the coordination of the two above layers, we will follow a "do-until-react" philosophy. What this means is that the robot will follow a path plan until there is a reaction. If there is a need to update the path plan, the the robot does so. We are expecting the low-energy requirements of the SNN path planning layer to help reduce the energy/computational expense. This will be a good research point for the future.

For the pourpose of path planning, we are researching into the posibility of the use of Spiking Neuronal Networks as a possible euclidean metric map representation. What we mean by this is using the timed-nature of neuronal networks to identify best possible paths to a goal based on who triggers first. Similar attempts have been performed in research/literature before as in Alamdari[1], bouganis et al., and Lobo et al.

The Executive layer is mainly governed by the coordinating layer. In this layer, the robot will decide what action to take next. The way the robot achieves this is with the following algorithm:

if there is next node:
  run to the node

• next_node is the response from a server from node trikh_planner called next_node
• reactions will specifically be the situation where the robot is about to bump against an obstacle. There exists 3 ways of reaction:
• Turn away.
• There is a way to get out
• Fly away.
• There is a flat wall in front of us.
• Stop.
• The robot has bumped against something and must stop. These reactions will be found in the trikh_grid_exec node.

To check for the reactions, the node uses a class called RobotReactive.

This class takes care of revising the environment for obstacles using a primitive approach: If robot is bumping or within offset distance from an obstacle, react.

The class uses a 3 bit flag called stimulus which is a private variable. This function acts as an accesor to the forementioned variable. The strcture of the flags is the following: 1st bit -> bumper, 2nd bit -> obstacle to the left, 3rd bit -> obstacle to the right.

For the creating of the world, the robot must be able to detect obstacles before encountering them by the bumper. For this is that we have decided to make use of obstacle-detection software such as OpenCV to detect obstacles and categorize them as needed.