Using attractor dynamics to generate decentralized motion control of two mobile robots transporting a long object in coordination

摘要

Dynamical systems theory is used here as a theoreticallanguage and tool to design a distributed controlarchitecture for a team of two mobile robots that musttransport a long object and simultaneously avoid obstacles.In this approach the level of modeling isat the level of behaviors. A “dynamics” of behavioris defined over a state space of behavioral variables(heading direction and path velocity). The environmentis also modeled in these terms by representingtask constraints as attractors (i.e. asymptoticallystable states) or reppelers (i.e. unstable states) ofbehavioral dynamics. For each robot attractors andrepellers are combined into a vector field that governsthe behavior. The resulting dynamical systemsthat generate the behavior of the robots may be nonlinear.By design the systems are tuned so that thebehavioral variables are always very close to one attractor.Thus the behavior of each robot is controledby a time series of asymptotically stable states. Computersimulations support the validity of our dynamicmodel architectures.