Development And Experimental Verification Of A Multi-Robot Cooperative Control Algorithm With Containment And Group Dispersion

Distributedcooperativecontrolofmulti-agentsystemshasbeenanactiveresearchtopicduetoitswiderangeofpotentialapplicationsinbothmilitaryandciviliansectors.Thedesignofapropercontrolstrategysuchthatmorethanoneagentswillworktogeth-er to accomplish a given task is one of the important control problems in distributedcooperative control of multi-agent systems.Containment control considers a group of leaders and a group of followers. Ina containment control problem, multiple followers will move into the convex hull s-panned by the multiple leaders and maintain the same fnal state as the leaders. Therehave been many results on the topic of containment control. However, few works havebeen reported on the experimental validation of theoretical developments. This the-sis focuses on the development of a multi-agent system experimental platform, whichincludes the design and construction of both the hardware system and the softwaresystem. A containment algorithm with group dispersion is then adapted to the experi-mental platform for validation of its efectiveness.In the development of the experimental platform, we frst give an overall design ofthe system and divide the system into four parts. The high-level computer is to designthe software control program that is responsible for the construction of the winelessnetwork, initialization of the hardware units, communications access and data trans-mission, as well as the local system information processing and sharing. The low-levelmotion microcomputer is used for the development of the motion control protocol and real-time control commands parsing and execution. The agents, the wheeled robot sys-tems and ultrasonic sensors form, respectively the executive modules and the feedbackmodules of the overall closed-loop control system.We use the double integrator model to represent the dynamics relating the agents’motion and their control inputs, which include the agents’ positions and speeds. Toachieve the containment and group dispersion, we propose the control laws for thedouble integrator dynamics in the presence of both stationary and dynamic leaders. Inthe case of stationary leaders, we propose a distributed containment control algorithmand study conditions on the network topology and the control gains that guarantee thegroup dispersion. The proposed control algorithm drives all the follower agents intotheconvexhullformedbythestationaryleaderswhileguaranteeingcollisionavoidanceduringthemovement. Thefnalstateofthefollowersarestationaryaswell. Forthecaseof the dynamic leaders, we consider the scenario where the velocities of the leaders areconstant and the convex hull of the leaders are fxed. We propose a control law underwhich the followers will move into the convex hull of the leaders and keep velocitiesmatching with the leaders. In the mean time, collision is avoided all the time.Simulation is performed on Matlab before the proposed containment algorithmsare tested on the experimental platform. Both the simulation and experimental resultsdemonstrate the efectiveness of the experimental platform and the correctness of the-oretical conclusions.