Parkour for Humanoids, Powered by Vision
Humanoid robots aren’t just walking anymore—they’re vaulting, dive‑rolling, and navigating messy ground like a traceur. “Deep Whole‑body Parkour” blends two worlds: smart footstep planning and full‑body skill tracking, so a single policy uses what it sees to coordinate hands, feet, and torso on uneven terrain.
- Integrates exteroceptive sensing (seeing the terrain) directly into whole‑body control.
- Trains one policy to perform multiple dynamic skills across varied obstacles.
- Enables robust multi‑contact moves—far beyond simple walking or running.
Why it matters: Most systems either walk well or mimic complex motions in lab settings. This work unifies both, unlocking high‑agility behaviors in the wild—think parkour for humanoids.
Learn more and watch results: https://project-instinct.github.io/deep-whole-body-parkour • Paper: https://arxiv.org/abs/2601.07701v1
By Ziwen Zhuang, Shaoting Zhu, Mengjie Zhao, Hang Zhao
Paper: https://arxiv.org/abs/2601.07701v1
Register: https://www.AiFeta.com
Robotics AI Humanoid Parkour ReinforcementLearning ComputerVision MotionControl
