Human-Like Redundancy Resolution for Real-Time Bimanual Humanoid Teleoperation

This research proposes a human-like redundancy-resolution framework for real-time bimanual humanoid teleoperation, combining adaptive elbow guidance with singularity-robust wrist tracking. The method uses shoulder-wrist geometric retargeting to generate robot-scale wrist targets and an adaptive elbow-offset criterion to progressively release redundant degrees of freedom. Results show a 54.4% reduction in mean elbow-direction error and a 97.9% reduction in peak joint-velocity demand in stress tests.