Research On The Control Of Multi-robot Coordinative Formation Under Unkown Environment

With the expansion of robot application, the demand to its ability is more and more high, and robot can cooperatively complete more complex task. Formation is a common cooperation strategy, and most research concentrate in the known environment, few in outdoor unknown complex environment. This thesis focuses on the multi-robot coordination formation in outdoor dynamic unknown environment.System structure is very important in multi-robot control. Considering robots are few and can intercommunicate, the multi-robot system uses hybrid architecture and leader-follower formation control method. Robot individual structure is based on behavior, and this thesis has designed a layered fusion architecture in order to enhance the output behavior's pertinency.Outdoor environment is unknown, and sampling-based motion planning can avoid environment modeling. This thesis has designed one kind method of Sensor-based Deterministic-sampling Tree (SDT). Considering the advantage of sensoring by group, behavior avoiding static obstacle adopted deterministic-sampling method and tangent available- territory method. Fuzzy control was used to change velocity to avoid dynamic obstacle whose track was be curve-fitted by Least-square principle.Formation wasn't incompletely kept to adapt to the complex circumstances. Formation point distribution used the farthest-minist-average-minist algorithm, moreover robot can self-determinately change the formation spot when it's occupied. Deceletaration-zone and goal–zone method was adopted to avoid concussion near the target position. During keeping formation, leader used formation conditional feedback and self-determinately changing figure due to environment.Aiming at actuality research taking formation control as the goal, the formation control unified with the work mission is researched. The different stage duty was defined as the different state, which carried on formation control separately. The Finite-State Machine(FSM) was used to transformate different states. Finally, the problem that multiple robots cooperatively hunt with unknown environment is disussed , and the validity of proposed method if proved by simulation experiment.