My undergraduate thesis(2020). Implemented auto-driving in DonkeyCar simulator using reinforcement learning algorithms(DDPG, SAC, A3C).
In this work, I implemented some RL algorithms on DonkeyCar Simulator, finally I made the car drive in the track successful. And did some try on CARLA simulator.
After testing, the best performance in this scenario is the SAC algorithm.
CTE=1，alpha=0.1
The car drives smoothly, but cannot stay in the center of the track.
CTE=5, alpha=0.1
The car drives in the center of the track, but not very smooth.
Actually I think CARLA is a better simulator for this research, because it has more useful functions and it's more like the real world. But because of the communication problems and insufficient GPU, I couldn't do the training on CARLA, just made some interfaces.
