A Novel Human-robot Three Arm Collaboration Method for Multiaim reach-grasp-place task via Hand-eye Coordination

Reach-grasp-place tasks are crucial for human-environment interaction, enabling everyday actions like arranging items and assembling parts. However, when multiple targets are distributed randomly, human efficiency drops sharply due to limited arm reach and workspace. Supernumerary robotic limb (SRL) can overcome these constraints by allowing simultaneous manipulation of more targets across a wider range. Nonetheless, coordinating humans and SRL efficiently in multi-goal tasks remains challenging. We propose a three-arm collaboration method for multi-goal reach-grasp-place tasks that integrates a gaze-based control strategy for the SRL with optimized timing for motion commands. Our results show that gaze-based control imposes a low cognitive load and maintains high task fluency in parallel multi-goal scenarios. Notably, issuing commands during the task planning phase and the latter half of the reaching phase minimally interferes with natural limb movements, preserving overall efficiency. Using this collaboration method, participants completed multi-goal tasks faster and quickly mastered SRL control.