This project involves working in a team of 2 to build an autonomous Robot that can navigate a maze towards an unknown but marked goal, and “learns” the maze to find the optimal way to go back to the starting point.
The problem to be solved is a maze navigation problem. The only information we are given here:
- a maze
- a starting point (on the maze)
- a final destination point
Our job was to come up with a strategy, and implement the corresponding "algorithm", so that the robot will go from the starting point to the goal/destination and then make its way directly back to the start without making a wrong turn.
The implementation of our project is heavily influenced by the robotic architecture called Subsumption architecture, which the control is divided into layers corresponding to levels of behavior. The idea of subsumption is that not only do more complex layers depend on lower, more reactive levels, but that they could also influencetheir behavior. Within subsumption architecture, the controlling structure is an arbitrator. The arbitrator looks through a list of behaviors, and depending on the current conditions, will fire off a certain behavior.
- Sensors
- TouchSensor
- LightSensor
- Ultrasonic Sensor
- Explore Class
- defines and initiates the default action to perform in different cases when walking through the maze
- Avoid Class
- takes Touch Sensor and Ultrasonic Sensor
- will be called on once an obstacle is detected
- ReachGoal Class
- takes Light Sensor
- will be called on once it gets light reflected from the white(goal) cell
- will walk back to the starting cell with the optimal solution, and play a song
- Cell Class
- holds information for coordinates regarding its location in the maze
- World Class
- holds the grid and knowledge built from beginning or gained during exploration
- and DFS algorithm for generating an optimal path