End-effector Motion and Force Control for Mobile Redundant Manipulators

Abstract

Manipulators extend the application of mobile platforms, such as unmanned aerial vehicles, underwater vehicles, and satellites. The dynamic coupling between the manipulators and mobile base brings great challenges to the motion control of this complex model. The existence of external disturbances and systematic uncertainties requires high robust control strategies. This dissertation focuses on the trajectory tracking control and force control in joint space and in task space. Theoretic analysis and simulation work are given to show the effectiveness of the proposed controllers. Three types of control strategy are proposed to follow desired joint trajectories: 1. adaptive backstepping with fuzzy logic 2. neural-adaptive control 3. adaptive dual integral sliding mode control These controllers are explored by computer simulations. A controller designed in task space is proposed to follow desired end-effector trajectory. The mapping relationship between joint space and task space is modified to guarantee system stabilities and a neural network is used to approximate system uncertainties. Simulation results show the stability when applied to trajectory and force control tasks. Since neural networks are used in several of the proposed controllers, a simplified robot arm is built to verify the effectiveness of neural network. Two types of controllers, torque control and position control, are tested in this platform and reasons are given to explain the performance of the two controllers. In the experiments, the neural network compensator is able to reduce the Integral of Squared Error (ISE) by more than 15x that achieved by the uncompensated (“open-loop”) commercial controller, and 8x-20x better than the PID compensated system.