Automatic reconfiguration of a robotic arm using a multi-agent approach

Abstract

There are numerous methods for solving the inverse kinematic equations for a robotic arm. This paper proposes a novel, adaptive approach based on multiagent systems (MASs). An MAS employs a distributed, decentralized approach to problem solving that is not commonly employed in conventional robotic arm control. The MAS uses patterns abstracted from various configurations of the robotic arm to provide a means of solving inverse kinematic equations where there is a changing kinematic model. Such an approach is beneficial in applications such as the maintenance of power transmission lines, welding, and providing support to handicapped people. The method is demonstrated using a case study utilizing a six-degree-of-freedom Kawasaki FS02 industrial robotic arm. The results from the case study demonstrate a solution for 95 per cent of all attainable Cartesian coordinates.