The Trajectory Estimation For A Mobile Robot And A Mobile Robot Team

The mobile robot, which integrates automatic control, electronic, computer science and other latest research results, is one of hot research topics in control and artificial intelligence areas. The studies in mobile robot not only have important theoretical value, but also have broad application prospects in the fields such as national defense, production, aerospace, military and so on. In recent years, the trajectory estimation of mobile robot/mobile robot team with wireless networks has become one of the key research subjects. Mobile robots estimate locations or postures of their neighbors and/or themselves, such as, coordinates, heading, speed and so on with measurements from local sensors and remote sensors through a network channel. Based these estimated parameters, mobile robots make decisions and generate control signals to accomplish given tasks.Extended Kalman filter is a commonly used and effective tools in the trajectory estimation of the robot. It is a nonlinear state estimation scheme developed from Kalman filter for linear systems. The key idea of this scheme is to estimate states of nonlinear systems based on local linearization models of these systems. Extended Kalman filter has been applied to solve many nonlinear state estimation problems.This work studies the trajectory estimation for mobile robot/robot team over network channels by using extend Kalman filter. The research work includes three parts:The first part of this work studies state estimation of moving target in network environment where the state of a moving target is estimated and predicted by using limited measurements, such as coordinates. Based extend Kalman filter, a state estimation algorithm is developed for this problem and it is turned out that the mean-square estimation error is exponentially bounded. And then random transmission delay in the network channel is considered. A modified state estimate algorithm is presented based on extended Kalman filter.The second part of this work studies the state estimation and localization problem for a mobile multi-robot team with one leader and two followers. Each robot is equipped some local sensors which measure a part of relative positions with team members. A two stage localization algorithm is developed for the team's localization.In the first stage, two followers estimate their relative positions and headings regarding to a landmark based on the measurements from local sensors on these robots. In the second stage, the team leader estimates its position and heading based on measurements from local sensors and follower's state estimation. Compared with centralized state estimation algorithm, the two stage algorithm has less computation and communication load in each robots.A distributed cooperative localization and state estimation algorithm is studied for a multi-robot team in the third part of this work. Each of team members is equipped with sensors which measure local variables and the relative position about its neighbor. And two robots can exchange their local estimation information at every sampling time. A distributed cooperative localization state estimation algorithm is developed. The robots in the team estimate their own state according to measurements from their own sensors and the estimation information from their neighbor. The structure of this collaborative extended Kalman filter makes it possible to extend this algorithm for a large number robot team.At last, the simulation and experiment results are presented to verify the effectiveness and implementability of the three algorithms.