Optimal angular back-projection error for vehicle motion estimation using omnidirectional vision

Nowadays, motion estimation systems have been applied in many researches on robotics, autonomous navigation, intelligent transportation systems, and industry applications. This paper presents a method for motion estimation based on an optimal approach using the angular back-projection error as a criterion for evaluating. The advantage of an omnidirectional camera is that allows tracking landmarks over the long-distance of a travel. Especially, it results in the high accurate estimation under large rotation. The method avoids the disadvantage of bundle adjustment (BA), incremental structure from motion, which is often diverging in long distance of the vehicle motion. The global motion of the omnidirectional camera is estimated in high accuracy based on utility of the error minimization in L_∞-norm method, which contrast with BA using L₂-norm that can gain local minima of back-projection errors. The experimental results demonstrate the effectiveness of this method under the large view scene in the outdoor environment.