File Download
There are no files associated with this item.
Links for fulltext
(May Require Subscription)
- Publisher Website: 10.1109/LRA.2020.2972879
- Scopus: eid_2-s2.0-85081054811
- WOS: WOS:000526521500025
Supplementary
- Citations:
- Appears in Collections:
Article: Learning Natural Locomotion Behaviors for Humanoid Robots Using Human Bias
Title | Learning Natural Locomotion Behaviors for Humanoid Robots Using Human Bias |
---|---|
Authors | |
Keywords | learning from demonstration Deep learning in robotics and automation humanoid and bipedal locomotion |
Issue Date | 2020 |
Citation | IEEE Robotics and Automation Letters, 2020, v. 5, n. 2, p. 2610-2617 How to Cite? |
Abstract | © 2020 IEEE. This letter presents a new learning framework that leverages the knowledge from imitation learning, deep reinforcement learning, and control theories to achieve human-style locomotion that is natural, dynamic, and robust for humanoids. We proposed novel approaches to introduce human bias, i.e. motion capture data and a special Multi-Expert network structure. We used the Multi-Expert network structure to smoothly blend behavioral features, and used the augmented reward design for the task and imitation rewards. Our reward design is composable, tunable, and explainable by using fundamental concepts from conventional humanoid control. We rigorously validated and benchmarked the learning framework which consistently produced robust locomotion behaviors in various test scenarios. Further, we demonstrated the capability of learning robust and versatile policies in the presence of disturbances, such as terrain irregularities and external pushes. |
Persistent Identifier | http://hdl.handle.net/10722/288795 |
ISI Accession Number ID |
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Yang, Chuanyu | - |
dc.contributor.author | Yuan, Kai | - |
dc.contributor.author | Heng, Shuai | - |
dc.contributor.author | Komura, Taku | - |
dc.contributor.author | Li, Zhibin | - |
dc.date.accessioned | 2020-10-12T08:05:53Z | - |
dc.date.available | 2020-10-12T08:05:53Z | - |
dc.date.issued | 2020 | - |
dc.identifier.citation | IEEE Robotics and Automation Letters, 2020, v. 5, n. 2, p. 2610-2617 | - |
dc.identifier.uri | http://hdl.handle.net/10722/288795 | - |
dc.description.abstract | © 2020 IEEE. This letter presents a new learning framework that leverages the knowledge from imitation learning, deep reinforcement learning, and control theories to achieve human-style locomotion that is natural, dynamic, and robust for humanoids. We proposed novel approaches to introduce human bias, i.e. motion capture data and a special Multi-Expert network structure. We used the Multi-Expert network structure to smoothly blend behavioral features, and used the augmented reward design for the task and imitation rewards. Our reward design is composable, tunable, and explainable by using fundamental concepts from conventional humanoid control. We rigorously validated and benchmarked the learning framework which consistently produced robust locomotion behaviors in various test scenarios. Further, we demonstrated the capability of learning robust and versatile policies in the presence of disturbances, such as terrain irregularities and external pushes. | - |
dc.language | eng | - |
dc.relation.ispartof | IEEE Robotics and Automation Letters | - |
dc.subject | learning from demonstration | - |
dc.subject | Deep learning in robotics and automation | - |
dc.subject | humanoid and bipedal locomotion | - |
dc.title | Learning Natural Locomotion Behaviors for Humanoid Robots Using Human Bias | - |
dc.type | Article | - |
dc.description.nature | link_to_subscribed_fulltext | - |
dc.identifier.doi | 10.1109/LRA.2020.2972879 | - |
dc.identifier.scopus | eid_2-s2.0-85081054811 | - |
dc.identifier.volume | 5 | - |
dc.identifier.issue | 2 | - |
dc.identifier.spage | 2610 | - |
dc.identifier.epage | 2617 | - |
dc.identifier.eissn | 2377-3766 | - |
dc.identifier.isi | WOS:000526521500025 | - |
dc.identifier.issnl | 2377-3766 | - |