Distributed Motion Coordination Of Mobile Robotic Network

This dissertation is focused on the distributed control of mobile robotic network (MRN) for its motion coordination.A MRN is a group of autonomous mobile robots networked by communication links and sensing.Motion coordination of MRN is a critical problem in its potential applications in environment monitoring,border security, search and rescue,etc.The basic idea of distributed control to attain motion coordination is that each robot regulates the evolution of its own states using those of its local companions in such a way that coordinated group behaviors are generated.A mathematical model of MRN is presented,including three parts:index set with elements indicating each robot,individual dynamics with velocity limits and information flow structure(IFS) in a group.The IFS is composed of communication and mutual sensing network model described by weighted digraph and proximity graph respectively. Based on the model,a distributed control architecture is designed and a simulation testbed built for algorithm verification by using tools in MATLAB/SIMULINK.Then two primitive group behaviors are investigated:parallel motion and aggregation motion.A control strategy for parallel motion is based on consensus algorithm.A proportional consensus algorithm distributed on a weighted digraph is proposed and the relationship between topologies of communication network and system equilibriums is discussed.With velocity specification considered,this linear algorithm is used to build a stop-and-go control strategy.Then this dissertation presents a class of nonlinear consensus algorithm the fact of which can make multiple single integrators reach agreement if the adjacency matrix of weighted digraph is symmetric is proven.Also a control strategy is derived from this nonlinear algorithm while satisfying velocity limits and being able to drive the system to its final state rapidly.An aggregation control law is derived from interaction forces between any two adjacent robots in the IFS.This dissertation proposes a direct method to construct forces which are distributed on the proximity graph.In the condition that a nonholonomic robot can only move forward with velocity limits,it is guided by the vector of total force in its local reference frame.In this way,the group aggregates.Then a indirect method of force definition is given based on pairwise potentials for group aggregation in crystal patterns.A graph generation algorithm is presented and applied to finding a crystal graph as communication network on edges of which Morse potentials are defined. Based on the negative gradient of total potential,a distributed switching control law is proposed to keep the distance between any two neighbors on the crystal graph being a desired value.The effectiveness of all control algorithms and effects of IFS on group behaviors are verified through computer simulations in MATLAB.