Control of multiple robots

Mohamed Zribi, Purdue University

Abstract

In this thesis, we address the problem of controlling multiple robots manipulating a common load cooperatively. Several controllers are designed for rigid robots, flexible joints robots and redundant robots. These controllers enable us to control both the load position and internal grasping forces. The controllers take into account both the dynamics of the object and that of the manipulators. Properties such as the linearity of the dynamics with respect to the parameters are exploited during the derivation of the controllers. The proposed controllers do not require feedback of joint accelerations. First, a controller that ensures the asymptotic stabilization of the position of the load and the internal forces of the multi-robot system is designed. To account for the uncertainty in the mass of the load, an adaptive version of the controller is introduced. Then, a controller which is based on Lyapunov theory is developed; this controller assumes full knowledge of the parameters of the multi-robot system. However the parameters might not be known exactly, thus an adaptive version of this controller is proposed. This controller is based on the passivity property, and it leads to an error equation where the regressor is independent of the joint acceleration. We also address the issue of robustness in the presence of bounded disturbances. Another way of eliminating the joint acceleration is by filtering the dynamics of the load and the robots. A predictive adaptive controller for rigid robots is thus obtained. This controller is based on the cancellation of the nonlinearities. The least-squares estimation method is used to estimate the parameters of the multi-robot system from the prediction errors. The previous controllers are derived with the assumption that the robots are rigid, but in real life such assumption may not be valid. Thus we address the problem of controlling multiple flexible joint robots manipulating a common load cooperatively. Asymptotic stability of the system is insured regardless of the joint flexibility and despite the fact that force sensors are not required. Some tasks such as obstacle avoidance require the use redundant robots because they offer means to overcome the limitations of non-redundant robots. Thus, we propose an adaptive control scheme for kinematically redundant multiple robots in cooperative motion. The proposed controller guarantees exponential tracking of the load position and also ensures the convergence of the internal force trajectories.

Degree

Ph.D.

Advisors

Ahmad, Purdue University.

Subject Area

Electrical engineering|Mechanical engineering

COinS