Sequential estimation of sensor orientation for stereo image sequences

Abstract

Among a wide variety of sensing technologies vision systems are capable to derive a mobile robot's relationship and orientation with respect to the surrounding environment.In this paper, we present an algorithmic approach to derive sensor orientation from image sequences acquired by a binocular vision-based mobile "robot", which moves in an office, sensing its environment.The data acquisition, the extraction of characteristic features in the image frames, and the tracking and matching of the landmarks along the sequence is described.The determination of the sensor orientation by sequential estimation based on Givens transformations is demonstrated using the tracked data of the test sequence.The results of the vision system's calibration and of the sensor orientation are presented.Finally, a comparison of the sensor orientation of the last stereo pair with reference data indicates the achieved accuracy of the algorithmic approach.