Controlling goal-directed motor behavior requires complicated computations by the brain. Such computations, needed for sensorimotor information processing and motion planning and control, are one of the focuses of research at the Department of Computer Science and Applied Mathematics. Work in computational motor control combines the recording and analysis of human and animal movements, brain mapping studies and the development of mathematical models, and computer algorithms. Since this field of computational neuroscience shares its goal of studying the translation of perception into action with robotics, robotics has proven a useful test bed for proposing and testing ideas about biological motor control. Likewise, since human capabilities far surpass those of artificial systems, insights from research on biological motor control provide inspiration for robotics and artificial intelligence.
A WIS team has investigated the rules governing the selection of limb motion, particularly examining the principles subserving the selection and planning of the trajectories of the hand during multi-joint arm movements such as reaching, obstacle avoidance, curved and drawing movements. A mathematical model developed for the description of the geometrical and kinematic features of observed hand trajectories suggested that multi-joint arm movements are planned in terms of the hand coordinates in external space and that the brain selects plans that maximize the smoothness and predictability of the trajectories. The principle of maximizing smoothness was mathematically expressed as minimizing the hand jerk, i.e., the rate of change of hand acceleration integrated over movement duration. This model, developed by a WIS scientist and her collaborator (from MIT), was highly successful in accounting for the geometrical and temporal features of human movements and was used to assess motor impairments in movement disorders (e.g., Parkinson's disease), and was also widely adopted in robotics. Recently a WIS scientist combined this model with studies of differential geometry to explore the notion that elementary motor building blocks are combined to generate the fluent, continuous graceful movements characterizing human motor behavior.