| Multi-robot system is the new aspect of robotics development. On the one hand, due to the complexity of some tasks, single robot is not competent to carry out all jobs for its limit capability, in such situations, multiple robots are employed to achieve task objectives through mutual coordination and collaboration. On the other hand, mutual coordination among robots will enhance the working efficiency of the robots system. In the research of multi-robot systems, a well-attended and essential topic is the multiple mobile robots coordination. In this dissertation, through constructing a series of novel theories and methods, the further discussions on multiple mobile robots coordination system will be presented. Firstly, to improve coordination ability of multirobot systems, a class of architecture based on agents communication mechanism is researched. Novel UDP-TCP dual communication channel design is adopted in such structure, which can reduce the channel disturbance and speed up the communication. To demonstrate results of research, a general scheme of multiple mobile robots simulation system is developed. By employing Virtual robot perception technique based on Virtual laser, Thread-based program structure and Object-oriented software design, the real-time simulation for large scale of mobile robots motion coordination is realized. Multi-robot closed-upι-φcoordination control algorithm is an effective motion coordination strategy as known. Further research is discussed in this dissertation. At first, we analyze the basic stability issue of closed-up ι-φalgorithm, compare the stability performance between serial and parallel formation and give correponding conclusion based on LFS theory. Then, the improved algorithm importing "potential field"is presented to improve dynamic performance in multirobot motion process. At last, fault tolerant control problem in multirobot motion coordiantion is presented and discussed, by using hierarchy formation graph, we research formation reconfiguration based on closed-up ι-φalgorithm initially. Flocking control is a new pattern of decentralized approach imitating animal cooperative behaviour. A leader-follower formation control mechanism is introduced based on existed theory on flocking of mobile agents, which make flocking motion be sequencable behaviour. To transform the agent-based flocking control strategy to network of mobile robots, an overall control algorithm is devised to take into account the dynamics of nonholonomic mobile robots. Then, we research a class mensurable flocking analytical method, and give some principles on flocking parameters adjustment. Several new algorithms on multi-robot obstacle and collision avoidance is introduced in this dissertation. Obstacle avoidance algorithm integrating 2-D laser information and wall-following behavior is introduced to overcome lacal minimum problem in potentail methods. Then, using virtial rigid body contact force concept and correponding algorithm, we can solve multi-robot collision avoidance problem. Next, we study a class multi-robot obstacle avoidance strategy based on unified potential method. Lastly, the universal multi-robot obstacle avoidance strategy integrating wall-following behavior and path-tracking mode are presented, which construct a class of real-time and environment-adaptive multi-robot coordination system, with closed-up l ?? algorithm or flocking algorithm. Aiming at actuality emphasizing on single task coordiantion, Multiple mobile robots Hierarchy-Agent coordination control system based on 4D/RCS which can achieve multi-task coordination is researched. The various task such as object perception, task planning and behavior control can be cooperativly completed with this scheme. Within such framework, multi-robot cooperativly prey/head-off algorithm based on "potential point"is studied. We accomplish a simulating experiment on "UGV cooperative combat", which make Hierarchy-Agent coordination control system into reality preliminarily.
|
PDF Full Text Request