Research On Multi-Robot Collaboration And Environment Modeling

The cooperation of multi-robot is becoming one of the hot topics in robotics. And the cooperative exploration and mapping of multi-robot is considered as the important mark of autonomous navigation and task cooperatively. This can be applied to the fields of planet exploration, opening field exploration, military detecting, disaster rescue and so on.The research of this thesis aims to improve the precision and effiecincy in cooperative exploration of multi-robot. The fast high precision map building, task allocation for improving the system efficiency, and the coordination for complete coverage of multi-robot are researched here.Firstly, the formation control of the multi-robot based on constrain equation is implemented. In order to improve the possibility in cooperation, coverage in exploration and coordination in moving, the formation contol in one clustering is needed. After analyzing the feature of constrain equations, the static obstacle avoidance and path tracking is also designed based similar equations. This reduces the difficulty in controlling robots. In addition, in order to reduce the collision between different clusterings, the dynamic avoiding strategy based on speed space is presented here, which improves the dynamic avoiding capability of robots.Second, the multi-robot task allocation strategy based on clustering is researched. Inspired by swarm cooperation behavior, the multi-robot cooperation strategy is designed. By modeling and solving the system behaviors, the discipline of how the system behaviors evolve with the task executing is analyzed to support the task assignment of multi-robot system. The results show that this method makes robot complete the task adaptively, efficiently and flexibility.Thirdly, the inter-cluster cooperation method based on specialized robot is presented. Inspired by the specialization phenomenon in swarm animals, such as ants, bees, and so on, the inter-cluster robots task switch method based specialized role is designed. The environment exploration is implemented by combining initiative and passive robot clusters. Through this method, the system can adjust the cooperative swarm size in real-time according to the environment task changes. Thus it can improve the efficiency and flexibility of the system through this mode, and also with the ability to adapt different tasks.Finally, the relative localization with each other based on local map is presented. In order to obtain exact environment information, the map building is needed. While the map is built, the relative localization among robots is also implemented. This is called as Cooperative Localization and Mapping. The continuous self-localization and relative localization based on local map is researched. This method reduces the calculation space from global field to local field, which improves the speed of multi-robot localization.