Extended kalman filter based moving object tracking by mobile robot in unknown environment

Abstract

In order to solve the problem of moving object tracking by robot in unknown environment, an estimation algorithm based on extended Kalman filter (EKF) is proposed. The states of robot, environment feature and object are used to form system state as a whole in the algorithm, such that sufficient relation is established gradually among states of different objects in iteration process, which improves accuracy of object state estimation. Moreover, a method of moving object detection based on occupancy grid map is combined with our algorithm to obtain the measurements of moving object and environment landmarks, so that the final algorithm can be used in actual environment. Furthermore, the step of data association proposed in algorithm can deal with the system state estimation disturbance caused by false object observations. Simulation experiment and real robot experiment results prove the effectiveness and accuracy of the presented approach.