Human motion provides rich priors for training general-purpose humanoid control policies, but raw demonstrations are often incompatible with a robot's kinematics and dynamics, limiting their direct use. We present a two-stage pipeline for generating natural and dynamically feasible motion references from task-space human data. First, we convert human motion into a unified robot description format (URDF)-based skeleton representation and calibrate it to the target humanoid's dimensions. By aligning the underlying skeleton structure rather than heuristically modifying task-space targets, this step significantly reduces inverse kinematics error and tuning effort. Second, we refine the retargeted trajectories through progressive kinodynamic trajectory optimization (TO), solved in three stages: kinematic TO, inverse dynamics, and full kinodynamic TO, each warm-started from the previous solution. The final result yields dynamically consistent state trajectories and joint torque profiles, providing high-quality references for learning-based controllers. Together, skeleton calibration and kinodynamic TO enable the generation of natural, physically consistent motion references across diverse humanoid platforms.
Figures & tables
Fig. 1 : Experimental results of highly dynamic side flip tracking on Unitree G1.
Fig. 2 : Retargeting framework.
Fig. 3 : Coordinate frame definitions of human skeleton and the corresponding human URDF in rest pose. While the human bone frames Bi are often aligned to the specific bones, we align the URDF link frames Li to a user-defined aligned frame A with the same orientation as the robot URDF root link to facilitate URDF calibration.
Fig. 4 : Rest pose key frame calibration of human URDF to humanoid robot dimensions. We match the link lengths for arms regardless of their absolute positions, and the absolute positions for the remaining key frames. We also calibrate the orientation offsets for end effectors if the robot has enough degree of freedom to track end effector orientations.
Fig. 5 : IK Comparison between GMR (left) and URDF Calibration (right).
Retarget Method
G1
H1
T1
PM01
Kuavo 4Pro
GMR
9.37
12.53
6.59
9.51
19.50
URDF Calibration
1.60
4.40
1.85
2.47
4.71
TABLE I : IK Empbpe [cm] Comparison between GMR and Our URDF Calibration on AMASS ACCAD Dataset over Multiple Robots.
Fig. 6 : Unitree G1 performing a forward jump. The left case is the baseline, and the right one is edited to jump higher.
Fig. 7 : Comparison of RL Empbpe learning curves of jumping motion editing when the reference motions are provided by raw edit, KTO, and KDTO.
Fig. 8 : Comparison of RL Empbpe learning curves of side flip tracking when the reference motions are provided by KTO, KDTO, KDTO+T.
Institute of Humanoid Robots, Department of Precision Machinery and Precision Instrumentation, University of Science and Technology of China, Hefei, Anhui 230026, China