Neural Networks for Cooperative Control of Multiple Robot Arms

In this chapter, we investigate the kinematic control of a single robot arm with an eye-in-hand camera for visual servoing by using neural networks. The visual servoing problem is formulated as a constrained quadratic program, which is then solved via a recurrent neural network. By this approach, the visual servoing with respect to a static point object is achieved with the feature coordinate errors in the image space converging to zero. Besides, joint angle and velocity limits of the robot arm are satisfied, which thus enhances the safety of the robot arm during the visual servoing process. The performance of the approach is guaranteed via theoretical analysis and validated via a simulative example.

This chapter studies the decentralized robot arm cooperation problem with a star control topology. The problem is formulated as a constrained quadratic program and then a recurrent neural network with independent modules is presented to solve the problem in a distributed manner. Each module in the neural network controls a single manipulator in real time without explicit communication with others and all the modules together collectively solve the common task. The global stability of the presented neural network and the optimality of the neural solution are proven in theory. Application orientated simulations demonstrate the effectiveness of the method.

This chapter studies the decentralized robot arm cooperation with a hierarchical control topology. We present in this chapter a novel strategy capable of solving the problem even though there exists some robot arms unable to access the command signal directly. The cooperative task execution problem can be formulated as a constrained quadratic programming problem. By replacing the command signal with estimations with neighbor information, the control law becomes to work in the partial command coverage situation. We then prove in theory that the system indeed also globally stabilizes to the optimal solution of the constrained quadratic optimization problem. Simulations demonstrate the effectiveness of the method presented in this chapter.

This chapter considers cooperative kinematic control of multiple robot arms with a full distributed control topology by using distributed recurrent neural networks. The problem is formulated as a constrained game, where energy consumptions for each robot arm, saturations of control input, and the topological constraints imposed by the communication graph are taken into account. An implicit form of the Nash equilibrium for the game is obtained by converting the problem into its dual space. Then, a distributed dynamic controller based on recurrent neural networks is devised to drive the system towards the desired Nash equilibrium to seek the optimal solution of the cooperative control. Global stability and solution optimality of the neural networks are proved in theory. Simulations demonstrate the effectiveness of the method presented in this chapter.