Learning-based locomotion control fusing multimodal perception for a bipedal humanoid robot
The ability of bipedal humanoid robots to walk adaptively on varied terrain is a critical challenge for practical applications, drawing substantial attention from academic and industrial research communities in recent years. Traditional model-based locomotion control methods have high modeling compl...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-03-01
|
| Series: | Biomimetic Intelligence and Robotics |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S266737972500004X |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1823856661599617024 |
|---|---|
| author | Chao Ji Diyuan Liu Wei Gao Shiwu Zhang |
| author_facet | Chao Ji Diyuan Liu Wei Gao Shiwu Zhang |
| author_sort | Chao Ji |
| collection | DOAJ |
| description | The ability of bipedal humanoid robots to walk adaptively on varied terrain is a critical challenge for practical applications, drawing substantial attention from academic and industrial research communities in recent years. Traditional model-based locomotion control methods have high modeling complexity, especially in complex terrain environments, making locomotion stability difficult to ensure. Reinforcement learning offers an end-to-end solution for locomotion control in humanoid robots. This approach typically relies solely on proprioceptive sensing to generate control policies, often resulting in increased robot body collisions during practical applications. Excessive collisions can damage the biped robot hardware, and more critically, the absence of multimodal input, such as vision, limits the robot’s ability to perceive environmental context and adjust its gait trajectory promptly. This lack of multimodal perception also hampers stability and robustness during tasks. In this paper, visual information is added to the locomotion control problem of humanoid robot, and a three-stage multi-objective constraint policy distillation optimization algorithm is innovantly proposed. The expert policies of different terrains to meet the requirements of gait aesthetics are trained through reinforcement learning, and these expert policies are distilled into student through policy distillation. Experimental results demonstrate a significant reduction in collision rates when utilizing a control policy that integrates multimodal perception, especially in challenging terrains like stairs, thresholds, and mixed surfaces. This advancement supports the practical deployment of bipedal humanoid robots. |
| format | Article |
| id | doaj-art-567a1797276347e0bafd1c8e78b74010 |
| institution | Kabale University |
| issn | 2667-3797 |
| language | English |
| publishDate | 2025-03-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Biomimetic Intelligence and Robotics |
| spelling | doaj-art-567a1797276347e0bafd1c8e78b740102025-02-12T05:33:06ZengElsevierBiomimetic Intelligence and Robotics2667-37972025-03-0151100213Learning-based locomotion control fusing multimodal perception for a bipedal humanoid robotChao Ji0Diyuan Liu1Wei Gao2Shiwu Zhang3School of Engineering Science, University of Science and Technology of China, Hefei 230026, China; iFLYTEK Co., Ltd., Hefei 230088, ChinaiFLYTEK Co., Ltd., Hefei 230088, ChinaSchool of Engineering Science, University of Science and Technology of China, Hefei 230026, China; Corresponding authors.School of Engineering Science, University of Science and Technology of China, Hefei 230026, China; Corresponding authors.The ability of bipedal humanoid robots to walk adaptively on varied terrain is a critical challenge for practical applications, drawing substantial attention from academic and industrial research communities in recent years. Traditional model-based locomotion control methods have high modeling complexity, especially in complex terrain environments, making locomotion stability difficult to ensure. Reinforcement learning offers an end-to-end solution for locomotion control in humanoid robots. This approach typically relies solely on proprioceptive sensing to generate control policies, often resulting in increased robot body collisions during practical applications. Excessive collisions can damage the biped robot hardware, and more critically, the absence of multimodal input, such as vision, limits the robot’s ability to perceive environmental context and adjust its gait trajectory promptly. This lack of multimodal perception also hampers stability and robustness during tasks. In this paper, visual information is added to the locomotion control problem of humanoid robot, and a three-stage multi-objective constraint policy distillation optimization algorithm is innovantly proposed. The expert policies of different terrains to meet the requirements of gait aesthetics are trained through reinforcement learning, and these expert policies are distilled into student through policy distillation. Experimental results demonstrate a significant reduction in collision rates when utilizing a control policy that integrates multimodal perception, especially in challenging terrains like stairs, thresholds, and mixed surfaces. This advancement supports the practical deployment of bipedal humanoid robots.http://www.sciencedirect.com/science/article/pii/S266737972500004XBipedal humanoid robotDeep reinforcement learningMultimodal perception |
| spellingShingle | Chao Ji Diyuan Liu Wei Gao Shiwu Zhang Learning-based locomotion control fusing multimodal perception for a bipedal humanoid robot Biomimetic Intelligence and Robotics Bipedal humanoid robot Deep reinforcement learning Multimodal perception |
| title | Learning-based locomotion control fusing multimodal perception for a bipedal humanoid robot |
| title_full | Learning-based locomotion control fusing multimodal perception for a bipedal humanoid robot |
| title_fullStr | Learning-based locomotion control fusing multimodal perception for a bipedal humanoid robot |
| title_full_unstemmed | Learning-based locomotion control fusing multimodal perception for a bipedal humanoid robot |
| title_short | Learning-based locomotion control fusing multimodal perception for a bipedal humanoid robot |
| title_sort | learning based locomotion control fusing multimodal perception for a bipedal humanoid robot |
| topic | Bipedal humanoid robot Deep reinforcement learning Multimodal perception |
| url | http://www.sciencedirect.com/science/article/pii/S266737972500004X |
| work_keys_str_mv | AT chaoji learningbasedlocomotioncontrolfusingmultimodalperceptionforabipedalhumanoidrobot AT diyuanliu learningbasedlocomotioncontrolfusingmultimodalperceptionforabipedalhumanoidrobot AT weigao learningbasedlocomotioncontrolfusingmultimodalperceptionforabipedalhumanoidrobot AT shiwuzhang learningbasedlocomotioncontrolfusingmultimodalperceptionforabipedalhumanoidrobot |