Whole-Body IK vs. RL on a Wheeled Humanoid

Loading video
Loading videoDyna compares whole-body IK and reinforcement learning on the same motion for its wheeled humanoid Taku. RL cuts hand-position tracking error across much of the sequence and produces more natural whole-body coordination.
HumanoidWheeled humanoidsReinforcement LearningWhole-Body ControlMobile ManipulationDYNA RoboticsWheeled robots
Category: humanoid
Author: @YorkYang5050
Date: 2026-11-10T00:00:00
Duration: 43.68s
Reference: https://www.dyna.co/dyna-2.1





