| Localization is a very important field in mobile robot navigation. Multi-robot cooperation is a challenging subject which flourished these years. In order to cooperate, robots must know their location. For the navigation technology for single mobile robot has been perfect day by day, research of cooperative localization has received more and more attention. This article is mainly in research of multi-robot cooperative localization:1 We analyse the constraints relations of relative observation in multi-robot cooperative localization and build the system model of cooperative localization in detail. Some kinds of sensors used to localize, map and measure relative observation are addressed.2 Single robot localization algorithm has been researched. The starting analysis is Fast simultaneous localization and mapping using particle filter, according to the localization environment, an improved method to calculate the importance weights of particle is put forward. The simulation results show the effectiveness of the proposed method.3 Homogeneous multi-robot cooperative localization has been researched. Single robot localization algorithm can readily be generalized to solve the multi-robot localization problem, assuming that the initial pose of all robots is known. In the more general case, the initial pose of the robots will not be known a priori, so we pay more attention to this case. We bring forward the time-reversed trajectory combination method and experimentally validated.4 Heterogeneous multi-robot cooperative localization has been researched. We bring forward the cooperative FastSLAM algorithm based on relative observation, putting the data of motion and the relative measurements into the improved FastSLAM process. This algorithm consists of an opinion that robots whose locating error is big depend on the powerful robot to localize relatively. Using heterogeneous multi-robot system which is consist of two different kind of robots, experiments verify the feasibility of the algorithm. |
PDF Full Text Request