Motion planning of multi body robots using feedback control simulation (In the case of multiple physical constrains)

Abstract

Recently, robots with high degree-of-freedom (DOF) have been used widely in various fields. These robots are able to conduct multiple purposes such as obstacle avoidance, postural control, trajectory tracking, etc. simultaneously using their redundancy. Hence, various motion generation methods are proposed for such robots. However, many conventional methods use the pseudo-inverse and null-space of the Jacobian matrix. Then the dedicated evaluation indexes are needed to solve the redundancy. These processes generally require the advanced expertise. Meanwhile, methods using configuration space have been utilized for a long time. However, these methods require enormous computation since the methods use iterative calculation. Recently, “virtual spring-damper hypothesis” was proposed as the motion generation method without using the expertise and iterative calculation for high redundant systems. In this method, the end-effector of the robot arm and reference point is connected by the virtual spring and damper in simulation space. The dynamic motion simulation is performed, then, from the simulation result, reference motion of the real robot arm is obtained. In this paper, a method that extends the virtual spring-damper hypothesis is proposed. In the proposed method, the virtual external force and torque are used to satisfy multiple constrains. Then, the effectiveness of the method is confirmed through several numerical simulations.