roboticsProject

RRT*-guided robot learning for planning

•   Setup maze environment with Turtlebot in Pybullet
  

•   Run RRT* to gather data, including local lidar observations
  

•   Implement an imitation learning policy to generate the next waypoint toward the goal using only the local lidar observations
  

You can find the results and documentation in the REPORT

Instructions

• Run python3 mazeV.py for checkpoint 3 Checkpoint 3 includes code for Collection of RRT data, Behaviour Cloning without PPO expert, DAgger with PPO expert

• Run python3 maze.py for checkpoint 2

• Conda Env : condaEnv.txt Project files :

• maze.py - Checkpount 1 & 2

• mazeV.py - Checkpoint 3

• turtleEnv.py - Containes the turtlebot environment along with the 2 maze environemnts with robot

• dagger.py - sample code for runing one episode of dagger

• bc.py - sample code for running one episode of behaviour cloning

• path.docx - file for storing RRT paths

• transitions.pkl - contains the observations (LIDAR, Position(included only for maze 1)) and actions space

• RRTStar.py - code for implementing RRT*

WIP : Imitation Learning