End-to-End Learning Approaches
{"description": "Slides and recording used to introduce end-to-end learning approaches in Duckietown such as reinforcement learning and imitation learning.", "keywords": "markov decision process, policy iteration, q-learning, reinforcement learning, imitation learning, machine learning, ML, AI, embedded AI"}
End-to-End Learning Approaches#
In this section, we depart from the traditional abstractions of perception, estimation, planning, and control and instead try to directly learn how to control an autonomous vehicle directly through data. We will primarily focus on two paradigms (although there are others): reinforcement learning (RL) and imitation learning (IL).