EUVIMEDEUVIMED entdecken
Crossref

Agile perceptive multiskill locomotion for quadrupedal robots in the wild

Jun-Gill Kang iD, Jaehyun Park iD, Tae-Gyu Song iD, Joon-Ha Kim iD, Seungwoo Hong iD, Hae-Won Park iD

Science Robotics · 15.07.2026 · 11 (116)

DOI: 10.1126/scirobotics.adz7397 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Science Robotics, erschienen am 15.07.2026. Verfasst von Jun-Gill Kang und 5 weiteren Autor:innen. Beteiligte Einrichtung: Agency for Defense Development, Daejeon 34186, Republic of Korea. Hauptthemen laut MEDLINE: Locomotion, Robotics. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 82 Literaturangaben. Gefördert von: Ministry of Trade, Industry and Energy, Agency for Defense Development.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Enabling quadrupedal robots to traverse complex terrains, from rugged outdoor environments to urban landscapes, requires seamless integration of multiple motor skills, smooth transitions between gaits, and high-speed perceptive locomotion using only onboard sensors. We present APT-RL (action pretrained transformer–based reinforcement learning), a unified framework that enables multiskill locomotion to achieve high-speed traversal in complex environments through autonomous skill transitions using only onboard perception and computation. Our approach generates large-scale, feature-rich two-dimensional (2D) motion datasets through trajectory optimization with simplified dynamics. These datasets enable training of diverse, reusable locomotion skills that transfer effectively to a real quadruped robot operating on complex uneven terrains. The resulting high-quality skills serve as strong priors for efficient learning of complex downstream tasks and extend naturally to 3D environments, enabling smooth, high-speed multiskill locomotion in deployed policy. Real-world experiments demonstrate the framework’s capabilities: The robot performed agile maneuvers through complex indoor obstacles and outdoor wild environments, including dynamic drop-down maneuvers that reached instantaneous peak speeds of up to 6 meters per second. A single onboard policy enabled robust traversal of diverse obstacles, including stairs, hurdles, stepping stones, gaps, and fallen branches, demonstrating the versatility and effectiveness of our approach.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

LocomotionRoboticsAutonomous RobotsReinforcement Machine LearningEquipment DesignHumansAlgorithmsAnimals

Autor:innen und Institutionen

  • Agency for Defense Development, Daejeon 34186, Republic of Korea.
  • Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology, Daejeon 34141, Republic of Korea.
  • DIDEN Robotics, Seoul 04799, Republic of Korea.
  • School of Mechanical Engineering, Korea University, Seoul 02841, Republic of Korea.
  • School of Smart Mobility, Korea University, Seoul 02841, Republic of Korea.

ORCID-Profile: Jun-Gill Kang, Jaehyun Park, Tae-Gyu Song, Joon-Ha Kim, Seungwoo Hong, Hae-Won Park

Förderung

  • Ministry of Trade, Industry and Energy
  • Agency for Defense Development

Zitieren

APA

Kang, J. G., Park, J., Song, T. G., Kim, J. H., Hong, S., & Park, H. W. (2026). Agile perceptive multiskill locomotion for quadrupedal robots in the wild. Science Robotics, 11(116). https://doi.org/10.1126/scirobotics.adz7397

Vancouver

Kang JG, Park J, Song TG, Kim JH, Hong S, Park HW. Agile perceptive multiskill locomotion for quadrupedal robots in the wild. Science Robotics. 2026;11(116). doi:10.1126/scirobotics.adz7397

BibTeX
@article{Kang2026,
  author = {Kang, Jun-Gill and Park, Jaehyun and Song, Tae-Gyu and Kim, Joon-Ha and Hong, Seungwoo and Park, Hae-Won},
  title = {Agile perceptive multiskill locomotion for quadrupedal robots in the wild},
  journal = {Science Robotics},
  year = {2026},
  volume = {11},
  number = {116},
  doi = {10.1126/scirobotics.adz7397}
}
Kostenlos starten →