Multi-robot environment mapping

The goal of this project was to construct two robots, able to map unknown environments as quickly as possible. The considered environments are exclusively composed of straight black lines on a white background, intersecting each other with right angles. The robots being only allowed to move along those black lines during the exploration phase.

The developed solution follows a master-slave architecture. Each robot (slave) follows a black line while keeping track of the travelled distance. Once an intersection detected, the robot stops and sends an information message to the master. The message contains the type of intersection encountered and the travelled distance since the previous intersection. This information is then used by a program (master), running locally on a computer, in order to update its internal representation of the environment (i.e. map). The next instruction (stop, straight, turn left or right, uturn) for the robot is then computed and send back to the rboto. This procedure is repeated until the environment is entirely explored.

   

The left-part of the animation above shows a live screen recording of the master constructing the map, corresponding to the robot exploration performed (displayed in the right-part). A video of the complete demonstration can be found here. The implementation and results are described in the report. The different steps of the construction of the robots are documented here.

Simulator

Testing the mapping algorithm with the exploration being performed by physical robots would have been time-consuming. Instead, the exploration was simulated on random environments.

The algorithm of the master is explained in the report) (pages 8-11). A simulation can be lauched with python main.py in the folder code/simulator/. The flag is_simulation needs to be set to true in the config file.

Robot

The control of the robot was tested with a simple challenge: following a curved line at a constant speed. To this end, a speed and steering controller were developed (described in the report, pages 20-24). The result is shown in the animation below.

The robot was then further augmented to loop through a serie of steps during the exploration phase. The high-level view of the main algorithm (serie of steps) is implemented in the file code/robot/mapping/mapping.ino, which is based on low-level features implemented in the files located in the folder code/robot/NewCOEX/. The acronym COEX stands for cooperative explorer.

Usage

The simulator can be launched as follows:

1. Set the flag is_simulation to True in the config file.
2. Run the command python main.py (in the folder code/simulator/).

The physical exploration can be launched as follows:

1. Copy the folder code/robot/NewCoex in your Arduino folder, and load it as a library (how to load a library).
2. Install the library corresponding to the QTR-X sensors.
3. Upload the program code/robot/mapping/mapping.ino to the micro-controller.
4. Set the flag is_simulation to False in the config file.
5. Check that the ports in the config files are correctly configured with the names of the bluetooth modules.
6. Switch the power of the robots on and calibrate the line sensors.
7. Run the command python main.py (in the folder code/master/).