Consensus making of multi-CPU control of hyper-redundant manipulator

A modular manipulator can be reconfigured according to its required function, conditions, and/or partial faults. We propose a control algorithm for a modular redundant manipulator. In this algorithm, local control processors develop independent inverse-kinematic solutions, and determine an overall solution by consensus. In this algorithm, since decisions are not made by a single CPU, but rather through negotiation among the local processors, the control algorithm has no explicit single failure point, which was suggested to be a weak point of our previous control algorithm. Since this new algorithm is also robust against asynchronous operation of the local processors, control performance is enhanced.