DCA: a distributed control architecture for robotics

Many contr olapplications ar e by nature distributed, not only over di erent processesbut also over several pr ocessors. Managing such a system with resp ectto the startup of pr ocesses,internal communications and state changes quickly becomesa very complex task. This pap erpr esents a distributed contr olar chite ctur e which supports a formal model of computation as described by [1]. The ar chite ctur eis primarily intended for rob ot contr olbut has a wide range of potential applications. We motivate the design and implementation of the ar chite ctur eby discussing the desired pr op ertiesof a rob ot system cap ableof doing real-time tasks like manipulation. This leads to functionality such as a pr ocessalgebra contr olling the life-cycle of the pr ocesses,grouping and distribution of processesand internal communication transpar ent to location. Our implementation do esnot in itself intr oduce any bottlenecks due to a tree structure with local contr olover pr ocesses which gives an eÆcient and scalable ar chite ctur e. At the end, an example scenario in which a fairly advanced problem like opening a door using a mobile r ob ot with a manipulator arm is demonstr ate d in the pr esented framework.