A Survey on Policy Search for Robotics

to Multiple Tasks 2.3.2Learning Multiple Solutions for a Single Motor Task 2.4 Policy Update Strategies 2.4.1 Policy Gradient Methods 2.4.2Expectation-Maximization Policy Search Approaches 2.4.3Information-theoretic Approaches 2.4.4Miscellaneous Important Methods 2.5 Real Robot Applications with Model-free Policy Search 2.5.1 Learning Baseball with eNAC 2.5.2Learning Ball-in-the-Cup with PoWER 2.5.3Learning Pan-Cake Flipping with PoWER/RWR 2.5.4Learning Dart Throwing with CRKR 2.5.5 Learning Table Tennis with CRKR 2.5.6 Learning Tetherball with HiREPS 3 Model-based Policy Search