Research On Robot Tracking Based On Wireless Sensor Network

As a new information capturing and proeessing technology, Wireless Sensor Network (WSN) is one of hot research field in the world currently. Mobile robots which have the function of environment interaction are expected to do some routine or dangerous tasks automatically for human in many situations. Research on the combination of WSN and robots has received a great deal of attentions from the society. Tracking technology, as a research on mobile robot introduced into WSN, is one of the most important and indispensable technologies.In this thesis, the tracking technology of mobile robot combined with WSN is researched.The range-based localization algorithm of WSN is investigated. The maximum likelihood-weighted centroid algorithm is proposed, which is applied to localization of robot in WSN. TDOA localization mechanism is adopted to establish the localization system model of robot, and localization of mobile robot is realized with maximum likelihood-weighted centroid algorithm with sample data through nodes of WSN. Simulation results show that the performance of maximum likelihood-weighted centroid algorithm exceeds other algorithms with better estimation accuracy, lower computation complexity and more stable.Based on WSN, the mobile robot tracking is studied with Sigma point Kalman filter algorithm. The non-linear state model of robot tracking is established with WSN and dead reckoning system. The extended Kalman filter and Sigma point Kalman filter are presented to use in robot trajectory tracking based on WSN. Simulation results indicate that tracking precision and stability of Sigma point Kalman fiter algorithm are better than that of extended Kalman filter in tracking when maneuvering of robot is strong.According to the uncertainty of form of movement in the complex environment, Interacting Multiple Model (IMM) algorithm is proposed to track trajectory of mobile robot that input and output of each filter are weighted using information of multiple models. The extended Kalman filter and Sigma point Kalman filter based on IMM algorithm are researched and applied to tracking of robot respectively. Simulation results show that the performance of filtering algorithm combined with IMM algorithm is better than that with single model.