EUVIMEDEUVIMED entdecken
Crossref

Learning quadrupedal locomotion on deformable terrain

Suyoung Choi iD, Gwanghyeon Ji iD, Jeongsoo Park iD, Hyeongjun Kim iD, Juhyeok Mun iD, Jeong Hyun Lee iD, Jemin Hwangbo iD

Science Robotics · 25.01.2023 · 8 (74)

DOI: 10.1126/scirobotics.ade2256 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Science Robotics, erschienen am 25.01.2023. Verfasst von Suyoung Choi und 6 weiteren Autor:innen. Beteiligte Einrichtung: Robotics & Artificial Intelligence Lab, KAIST, Daejeon, Korea. Die Arbeit wurde bisher 194-mal zitiert (Crossref) und nennt 39 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Simulation-based reinforcement learning approaches are leading the next innovations in legged robot control. However, the resulting control policies are still not applicable on soft and deformable terrains, especially at high speed. The primary reason is that reinforcement learning approaches, in general, are not effective beyond the data distribution: The agent cannot perform well in environments that it has not experienced. To this end, we introduce a versatile and computationally efficient granular media model for reinforcement learning. Our model can be parameterized to represent diverse types of terrain from very soft beach sand to hard asphalt. In addition, we introduce an adaptive control architecture that can implicitly identify the terrain properties as the robot feels the terrain. The identified parameters are then used to boost the locomotion performance of the legged robot. We applied our techniques to the Raibo robot, a dynamic quadrupedal robot developed in-house. The trained networks demonstrated high-speed locomotion capabilities on deformable terrains: The robot was able to run on soft beach sand at 3.03 meters per second although the feet were completely buried in the sand during the stance phase. We also demonstrate its ability to generalize to different terrains by presenting running experiments on vinyl tile flooring, athletic track, grass, and a soft air mattress.

Autor:innen und Institutionen

  • Robotics & Artificial Intelligence Lab, KAIST, Daejeon, Korea.

ORCID-Profile: Suyoung Choi, Gwanghyeon Ji, Jeongsoo Park, Hyeongjun Kim, Juhyeok Mun, Jeong Hyun Lee, Jemin Hwangbo

Zitieren

APA

Choi, S., Ji, G., Park, J., Kim, H., Mun, J., Lee, J. H., & Hwangbo, J. (2023). Learning quadrupedal locomotion on deformable terrain. Science Robotics, 8(74). https://doi.org/10.1126/scirobotics.ade2256

Vancouver

Choi S, Ji G, Park J, Kim H, Mun J, Lee JH, et al. Learning quadrupedal locomotion on deformable terrain. Science Robotics. 2023;8(74). doi:10.1126/scirobotics.ade2256

BibTeX
@article{Choi2023,
  author = {Choi, Suyoung and Ji, Gwanghyeon and Park, Jeongsoo and Kim, Hyeongjun and Mun, Juhyeok and Lee, Jeong Hyun and Hwangbo, Jemin},
  title = {Learning quadrupedal locomotion on deformable terrain},
  journal = {Science Robotics},
  year = {2023},
  volume = {8},
  number = {74},
  doi = {10.1126/scirobotics.ade2256}
}
Kostenlos starten →