Statistics for Machine Learning

Automotive vehicle control - self-driving cars

Self-driving cars are the new trend in the industry and many tech giants are working in this area now. Deep learning technologies, like convolutional neural networks, are used to learn Q-functions which control the actions, like moving forward, backward, taking left and right turns, and so on, by mixing and matching from the available action space. The entire algorithm is called a DQN (DeepQ Network). This approach can be used in playing games like Atari, racing, and so on. For complete details, please refer to the paper Deep Reinforcement Learning for Simulated Autonomous Vehicle Control by April Yu, Raphael Palesfky-Smith, and Rishi Bedi from Stanford University.