Autonomous Ball-Collecting Robot using Reinforcement Learning

1 · · Aug. 1, 2018, midnight
This is a project where we used reinforcement learning to train the robot to collect balls in a simple environment. The robot is a 4-Mecanum wheeled robot equipped with a camera and a 2D LiDAR sensor. The algorithms were implemented on Intel NUC and Nvidia Jetson TX2 board. This is my attempt at re-writing a write-up because a lot of recruiters ask me how the project was done and I sometimes forget (it was done in 2018). Optimal Policy and Optimal Value Functions Bellman optimality equation For...