Optimal braking for impact force reduction using the dynamics of redundant manipulators

Recently, developers have created a number of manipulators designed to provide physical services to human beings. Collisions between those manipulators and humans or other objects in the environment generate massive impacts, causing injury or limiting execution speed. These collisions are unavoidable, however; therefore it is essential that the force of such impacts be reduced, thus improving safety and task performance. One of the most effective methods to reduce impact force is to decelerate by braking immediately before the collision. In this paper, we propose Dynamic Acceleration Polytope Braking (DAPB), a method which is optimal braking to decelerate impact velocity quickly and without varying the direction of motion in redundant manipulators. DAPB uses a dynamic acceleration polytope to represent the acceleration ability, while considering manipulator dynamics to determine braking torque. By adopting DAPB, the velocity at impact can be decelerated faster than by zero-braking, which decelerates all joint velocities to zero; as a result, impact force can be greatly reduced. Moreover, since the path of the impact point does not stray from the initial impact direction, additional dangerous situations could be avoided. Several simulations and experiments are executed to confirm the braking performance of DAPB.