Learning model-free robot control by a Monte Carlo EM algorithm
Learning model-free robot control by a Monte Carlo EM algorithm
Peer-Reviewed Journal Publication
Δημοσίευση σε Περιοδικό με Κριτές
2015-03-232009enWe address the problem of learning robot control by model-free reinforcement learning (RL). We adopt the probabilistic model of Vlassis and Toussaint (2009) for model-free RL, and we propose a Monte Carlo EM algorithm (MCEM) for control learning that searches directly in the space of controller parameters using information obtained from randomly generated robot trajectories. MCEM is related to, and generalizes, the PoWER algorithm of Kober and Peters (2009). In the finite-horizon case MCEM reduces precisely to PoWER, but MCEM can also handle the discounted infinite-horizon case. An interesting result is that the infinite-horizon case can be viewed as a ‘randomized’ version of the finite-horizon case, in the sense that the length of each sampled trajectory is a random draw from an appropriately constructed geometric distribution. We provide some preliminary experiments demonstrating the effects of fixed (PoWER) vs randomized (MCEM) horizon length in two simulated and one real robot control tasks.
http://creativecommons.org/licenses/by/4.0/Autonomous Robots227123-130Vlassis_et_al_Autonomous_Robots_27_2009.pdfChania [Greece]Library of TUC2015-03-23application/pdf751 KB10.1007/s10514-009-9132-0free
Toussaint Marc
Kontes Georgios
Piperidis Savvas
Πιπεριδης Σαββας
Vlassis Nikos
Springer Verlag
Reinforcement learning