Autonomous Control of Reaching Movement by 'Mobility' Measure

Humans achieve reaching movements even when environmental conditions change unexpectedly, and because the human arm is redundant, a controller must incorporate constraints, and these constraints should adapt to changes in environmental conditions. In this paper, we propose new constraints that emerge in real-time from the sensory information acquired by the system during movement. In our model, the constraints needed to determine the control commands for each joint are implemented as autonomously decentralized interactions among the joints, based on a 'mobility' measure that evaluates instantaneous dynamic and kinematic properties of each joint. Simulated results indicate that these interactions can yield suitable motor commands in real-time, in such a way that the more mobile joints work dominantly and the less mobile joints work cooperatively, even when dynamic and kinematic perturbations are applied to the system during movement. The proposed constraints are thus adaptive to environmental and dynamic changes, and our controller is capable of functioning in the real world.