Human-Robot Gait Movements Mapping using Joint Projections to Ground

Chavez-Reynoso R.; Barrios-Aranibar D.; Patino-Escarcina R.E. · 2025 IEEE International Conference on Advanced Robotics, ICAR 2025 (2025) · DOI: 10.1109/icar65334.2025.11338620

Resumen

This work presents an approach for mapping human gait movements to a humanoid robot, using ground projection of contact joints as the central stage in kinematic adaptation. Human motion data were processed to obtain spatial positions and orientations of key joints, which were then projected onto the ground plane to simplify the mapping and ensure spatial correspondence with the robot's structure. New joint angles were computed through inverse kinematics, adjusted for mechanical constraints, and applied sequentially to replicate the gait pattern. The method was evaluated on a humanoid platform using a joint-by-joint analysis, showing low RMSE as well as temporal consistency in trajectory reproduction. The lowest values were observed in the ankle joints, indicating accuracy in the replication of contact movements. The greatest differences were detected in the knees, mainly during the swing phase of gait, possibly due to the robot's reduced ability to reproduce leg lift at that stage. Results demonstrate that the proposed approach preserves the structure and timing of human movements while adapting them to robotic kinematics, offering potential applications in rehabilitation robotics and human-robot interaction.

Instituciones: Universidad Católica San Pablo