Stable on-line neural control of systems with closed kinematic chains

Artificial neural networks have been used extensively in control research. In industrial systems, however, it is crucial to adopt neural control struc tures which have a guaranteed proof of stability, especially if control sys tem failure were to endanger life (e.g. in fast moving manipulators or tran sportation). In the paper, the neural control of robotic systems with close d kinematic chains is discussed and theorems guaranteeing the control stabi lity of such systems are developed. The first class of systems have a singl e serial chain with a prescribed contact force when moving across a surface , i.e. the problem of hybrid position/force neural control. The second clas s of systems considered includes hexapod walking machines, which have a var ying topology of closed kinematic chains during walking. The equations of m otion can be solved by optimising contact forces according to a predefined cost function, and so the hybrid/position neural controller is extended to this class. A novel control structure which makes no initial assumptions ab out the system is also presented, using the concept of virtual neural netwo rks': a projection of the neural controllers into the underconstrained spac e of the generalised co-ordinates of the equations of motion. This approach can be applied to a large number of different systems, including parallel manipulators and Stewart platforms, and it is also extended to include neur al networks implemented on digital microprocessors.