Multi-agent Optimal Control of Ball Balancing on a Mobile Robot

Multi-agent systems have origin in computer engineering. However, they have found applications in different fields. One of the newly emerged problems in multi-agent systems is multi-agent control. In multi-agent control it is desired that the control is done in distributed manner. That is the controller of each agent should be implemented based on local feedback. In this paper a mechanism is introduced as a test bed for multi-agent control systems. The introduced mechanism is balancing of a ball on link located on a planar mobile robot. Dynamic equations of the mechanism are derived and the control task is distributed among two agents. For each agent a two loop controller is designed wherein external loop is a LQR controller and inner loop is a simple proportional controller. Regulation and fault tolerance performance of controller scheme is evaluated by simulations.