Running Torcs With Ddpg Algorithm Keras

Summary

Trained on CG3 road with modified reward function. Trained an AI agent capable of high speed driving using Deep Deterministic Policy Gradient. We use Aalborg track as our training dataset. Trained with 1800 episode, no pre-training before. My computer is a bit slow so ... I trained on the wheel2 track with the modified reward function such that it does not require staying in the middle of the track. This video shows an RL-Agent playing The Open Racing Car Simulator. Naive Reinforcement Learning applied to Train car agent with Deep deterministic policy gradient (