In this chapter, we saw how to apply RL algorithms to train an agent to learn to drive a car autonomously. We installed the TORCS racing-car simulator and also learned how to interface it with Python, so that we can train RL agents. We also did a deep dive into the state space for TORCS and the meaning of each of these terms. The DDPG algorithm was then used to train an agent to learn to drive successfully in TORCS. The video rendering in TORCS is really cool! The trained agent was able to drive more than seven to eight laps around the racetrack successfully. Finally, the use of PPO for the same problem of driving a car autonomously was also explored and left as an exercise for the interested reader; code for this is supplied in the book's repository.
This concludes this chapter as well as the book. Feel free to read upon more material online on the application of...
