An analysis of constrained motions in manipulation

The manipulation of objects is a task which humans perform with great ease and success. Robots can have difficulty in manipulation tasks, especially if, as a result of the manipulation, the motion of the robot is constrained through contact with the environment. A general understanding of the effects of constraints on manipulation tasks will benefit designers and users of robots and prostheses. The aim of this thesis is to understand the effects of certain kinematic constraints on manipulation by either intact humans, amputees using prostheses, or robots.

Three hypotheses are proposed and developed to explain the strategy used by humans to deal with constraints. The first suggests that the manipulator is moved precisely along the required path and never comes into conflict with the constraint. The second proposes that the manipulator takes advantage of the constraint to minimize the force or energy needed to perform the task. The final hypothesis is that the manipulator is moved as if the constraint were not present and allows naturally present compliance in the arm to accommodate the constraint.

The specific task chosen for study in this research is turning a crank. A model of the arm is developed in order to predict the motions that would result from the application of the above hypotheses to the crank turning task. The results of simulations illustrate the implications of each of the theories. The model is used to determine which aspects of the motion are significant in the analysis of the task and to identify aspects of the motion which may cause the subjects difficulty in executing the task.

Crank turning experiments on intact human subjects were performed to clarify the techniques used by humans in dealing with constraints. The data collected during the execution of the crank turning experiments were the kinematics of the crank and the arm, the forces on the crank and EMG signals from some flexors and extensors of the elbow. The results indicate the path following hypothesis is incorrect. The effort minimization hypotheses does not satisfactorily explain the data. The constraint accommodation hypothesis is very promising but further experiments need to be performed to test it completely.

These studies clarify the effects of kinematic constraints on manipulation tasks, be they robot or prosthetic manipulators. This is a necessary step toward understanding tool use. (Copies available exclusively from MIT Libraries, Rm. 14-0551, Cambridge, MA 02139-4307. Ph. 617-253-5668; Fax 617-253-1690.)