Genetic optimisation of manipulation forces for co-operating robots

Abstract

Force planning for co-operating manipulators requires the solution of an underdetermined set of linear equations, which yields infinitely many solutions. The task then becomes that of locating the best solution from the infinite set. Various methods have been proposed for this optimisation problem, including pseudo inverses and search algorithms. This paper adopts a non-linear optimisation approach using genetic algorithms (GA). GAs have well known capabilities for searching complex function spaces. It is shown in this work that real number GAs provide a powerful force-planning method for co-operating manipulators. The effectiveness of the algorithm is illustrated with a numerical example.