Collision avoidance and multiple criteria optimization for a redundant manipulator

A control strategy that allows a manipulator with realistic hardware limitations to operate effectively in a real-world environment is presented. A redundancy-based collision-avoidance criterion, capable of real-time implementation, is developed to avoid work-space obstacles. Practical issues concerning the imposed computational burden of manipulator modelling, obstacle modelling, and distance computation are discussed. Since real-world applications simultaneously expose the manipulator to hardware limitations and collision risk, a method to simultaneously optimize multiple performance criteria in real time is developed. Formulation requirements for the individual criteria are established and previous normalization methods are examined. A method to extend redundancy to incorporate the multiple criteria is proposed and a statistically based variable weighting method is developed. The seven-degree-of-freedom Robotics Research Corporation (RRC) manipulator is used as a test bed for both the collision-avoidance and multiple criteria optimization methods.