Welcome to MRST’s documentation!

Multi-robot Reinforcement Learning Scalable Training School (MRST) is a training and evaluation platform for reinforcement learning reasearch.

Check out the paper “From Multi-agent to Multi-robot: Scalable Training Platform for Multi-robot Reinforcement Learning” for background on some of the project goals.

Simple Example

1. Launch the simulation environment

roslaunch mrst_simulation turtlebot3_autorace_roundabout.launch
roslaunch mrst_simulation turtlebot3_autorace_control_roundabout.launch

2. A simple code example for training

from Env import Env
def main():
   env=Env(scenario="roundabout")
   n_episodes = 100
   n_agents=12
   episode_length=15
   for e in range(n_episodes):
      env.reset()
      for et_i in range(episode_length):
            print(et_i)
            actions=[[1] for i in range(n_agents)]
            next_obs, rewards, dones, speeds = env.step(actions, isTeamReward=True)
if __name__ == "__main__":
   main()

Contents:

• Getting Start with MRST
• MRST Scenarios
  • Single-vehicle Lane Tracking
  • Multi-vehicle Cooperation
• How to customize the robot
  • RL-friendly Interface
  • Initialize the RL-based Model
  • Configuration of the Training
• Support for ROS Developer
• MRRL Research
  • Single-robot Reinforcement Learning
  • Multi-robot Reinforcement Learning