Winner SuperAI SS3 Level2 Hackathon 3 ROBOTICS - OBSTACLE AVOIDANCE: Mobile Robot With Reinforcement Learning
Making TurtleBot3 run toward its goal without colliding with obstacles, using only reinforcement learning
- We created a custom environment, using GYM(OpenAI) to connect with ROS2 Gazebo, which's robot operating system of TurtleBot3.
- trained robot by Proximal Policy Optimization(PPO) model from Stable Baseline3.
- visualized result via TensorBoard and tuned model by changing/calculating observation (state in reinforcement learning) and reward functions.
