Observation consistency for moving object tracking with mobile robot in dynamic environments

In order to solve the problem of spatial observation consistency from heterogeneous multi-sensor in the process of simultaneous localization, mapping and object tracking (SLAMOT), a calibration optimization method of camera and laser range measuring sensor based on information fusion is proposed. Uncertain arera of laser scanning point image plane projection is determined based on error propagation formula, and a covariance intersection based method which fuses informations come from moving object detection and Camshift method to object state estimation is designed. On this basis, the objective function is constructed with bearing error of object image projection, and calibration parameters of camera and laser range finder are optimized using nonlinear optimization method. Experiments show that the designed method improves accuracy of both object tracking and multi-sensors calibration. The method offers measurements which support further research of SLAMOT filter based on multi-sensor information fusion. ©, 2015, Chinese Optical Society. All right reserved.