Edinburgh Research Explorer

An Operational Method Toward Efficient Walk Control Policies for Humanoid Robots

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Original languageEnglish
Title of host publicationProceedings of the Twenty-Seventh International Conference on Automated Planning and Scheduling (ICAPS 2017)
Place of PublicationPittsburgh, USA
PublisherAAAI Press
Pages489-497
Number of pages9
Publication statusPublished - 23 Jun 2017
Event27th International Conference on Automated Planning and Scheduling - Pittsburgh, United States
Duration: 18 Jun 201723 Jun 2017
http://icaps17.icaps-conference.org/

Conference

Conference27th International Conference on Automated Planning and Scheduling
Abbreviated titleICAPS 2017
CountryUnited States
CityPittsburgh
Period18/06/1723/06/17
Internet address

Abstract

Optimizing policies for real-time control of humanoid robots is a difficult task due to the continuous and stochastic nature of the state and action spaces. In this paper, we propose a learning procedure to train a predictive motion model and RFPI, a solver for continuous state and action MDP. We use the predictive model as a transition model to train policies for a robot soccer. Our method requires no external hardware, a small amount of human work and manages to outperform the expert policy used by our team Rhoban winning the last 2016 edition of the Robocup in kid-size soccer league. Moreover, the proposed method is able to adapt to nonholonomic robots more efficiently than the expert approach. Our results are confirmed by both simulations and real robot experiments

Event

27th International Conference on Automated Planning and Scheduling

18/06/1723/06/17

Pittsburgh, United States

Event: Conference

ID: 70522683