Formation tracking control problem for multiple nonholonomic chained form systems are considered in this paper. In the framework of virtual structure, the formation keeping problem between chained-form agents are treated as tracking problem between the virtual agent and the actual agent. Theories from nonautonomous cascaded systems are introduced to simplify the design of formation controllers after carefully studying the structure of the error dynamic systems and global exponential controllers are constructed for the the above cooperative control problems in the end. Simulation results using Matlab show the feasibility of solving the formation tracking problems using methodology presented in this paper.
I. INTRODUCTIONThe advent of powerful embedded systems and communication networks have caused intensive interests in the cooperative motion control of multi robotic systems on the land [15], in the air [26] [27] or even in the water [10]. Due to multi-agents' potentiality of accomplishing complex tasks and single agent's simplicity and inexpensiveness, multiagent robotic systems have great potential to be used in rescue mission, large object moving, troop hunting, formation control and satellites clustering which is difficult or impossible for one single robot to accomplish.Previous work mainly focused on decentralized control of integrator systems and nominal linear system. In [11], cooperative control laws based on nearest neighbor rules were presented for single integrator systems in discrete form. Consensus problem for double-integrator systems were considered in [23] where conditions of both undirected graph and directed graph were given. Previous consensus results of single or double integrator systems were generalized in [25] to high order integrator system with consensus to desired reference model. Decentralized dynamic output feedback were introduced in [33], [34] to deal with the consensus problem of nominal linear systems. Cooperative consensus and formation control problems for nominal linear systems were also studied in [22] utilizing results of irreducible and reducible matrices. While cooperative target tracking control of single integrator robots are considered in [30] using distributed Kalman filter. In the view of input and output, the powerful tool of passivity was adopted in [17] to solve the cooperative control of multiple passive systems. But the passivity-based method may not applied to nonholonomic system as shown in [2]. Compared with linear