Controlling the cooperative behavior of a system of automous mobile robots

Abstract

A novel material transport system is presented that uses 'swarms' of small autonomous mobile robots to collectively lift and move palletized loads. The robots are relatively unsophisticated in design and have no advanced sensory or communications capability. There is no central or supervisory controller directing the robots. Each robot must be able react to its environment autonomously, yet cooperate within a team of similarly designed robots. Reactive and behavior-based principles are the basis of the system design. Although the entire material transport scenario is presented in this thesis in the context of a reactive behavior-based control architecture, emphasis is placed on developing a single behavior that allows the robots to move a loaded pallet once it has been lifted by a team of robots. Toward this goal, a centralized control scheme is derived that directs the actions of the robots while underneath a loaded pallet. It is shown that this approach produces an 'optimal' distribution of work among the robots. Alternatively, a distributed control scheme is derived that allows each robot to autonomously assist in moving a palletized load. This approach assumes a team of robots is capable of electing a leader and uses the dynamics of a caster as the basis for the development of a control law. Both the central and distributed control schemes are verified via computer simulation.