Dynamical Behavior of Multi-Robot Systems Using Lattice Gas Automata

摘要

Recent attention has been given to the deployment of an adaptable sensor array211u001erealized by multi-robotic systems. Our group has been studying the collective 211u001ebehavior of autonomous, multi-agent systems and their applications in the area of 211u001eremote-sensing and emerging threats. To accomplish such tasks, an 211u001einterdisciplinary research effort at Sandia National Laboratories are conducting 211u001etests in the fields of sensor technology, robotics, and multi-robotic and multi-211u001eagents architectures. Our goal is to coordinate a constellation of point sensors 211u001ethat optimizes spatial coverage and multivariate signal analysis using unmanned 211u001erobotic vehicles (e.g., RATLERs, Robotic All-ten-sin Lunar Exploration Rover-211u001eclass vehicles). Overall design methodology is to evolve complex collective 211u001ebehaviors realized through simple interaction (kinetic) physics and artificial 211u001eintelligence to enable real-time operational responses to emerging threats. This 211u001epaper focuses on our recent work understanding the dynamics of many-body systems 211u001eusing the physics-based hydrodynamic model of lattice gas automata. Three design 211u001efeatures are investigated. One, for single-speed robots, a hexagonal nearest-211u001eneighbor interaction topology is necessary to preserve standard hydrodynamic 211u001eflow. Two, adaptability, defined by the swarm's deformation rate, can be 211u001econtrolled through the hydrodynamic viscosity term, which, in turn, is defined by 211u001ethe local robotic interaction rules. Three, due to the inherent non-linearity of 211u001ethe dynamical equations describing large ensembles, development of stability 211u001ecriteria ensuring convergence to equilibrium states is developed by scaling 211u001einformation flow rates relative to a swarm's hydrodynamic flow rate. An initial 211u001etest case simulates a swarm of twenty-five robots that maneuvers past an obstacle 211u001ewhile following a moving target. A genetic algorithm optimizes applied nearest-211u001eneighbor forces in each of five spatial regions distributed over the simulation 211u001edomain. Armed with knowledge, the swarm adapts by changing state in order to 211u001eavoid the obstacle. Simulation results are qualitatively similar to lattice gas.