|M.Sc Student||David Dovrat|
|Subject||Gathering & Control of Unicycle A(ge)nts with Crude|
Bearing-Only Sensing Capabilities
|Department||Department of Autonomous Systems and Robotics||Supervisor||Full Professor Bruckstein Alfred|
We present a local rule of behavior for extremely simple agents, only able to detect the presence of other agents in a visibility sector directly in front of them.
The rule causes the agents to agree upon a common center of rotation and synchronize their orientation to form dynamic patterns.
By implementing this simple, local, rule of interaction, the agents coordinate their movement without ever acquiring any information on actual bearings towards any other agent.
The simplicity of the agents allow for a cost effective implementation of the model, since the use of sophisticated equipment is rendered unnecessary.
The formation to which the agents converge, be it a rotating regular polygon, a set of rotating polygons, or some other cohesive behavior, depends on a set of simple predefined parameters: the agents' field of view, their common speed and their rotation radii.
The work presented here contains a full analysis of a beacon-agent system, as well as convergence theorems for a 2-agent system and equilibrium analysis for a N-agent system.
Gathering in the N-agent case and orientation synchronization in the 2-agent case are demonstrated in simulation.
Methods of controlling the location of such a swarm are also
discussed and were tested both in simulations and in the lab with actual