Integrated binocular and motion stereo in a inertial navigation sensor-based mobile vehicle

Range measurements to real-world objects from mobile platforms such as ground or air vehicles are critical for visually aided navigation and obstacle detection/avoidance. The authors present an approach that consists of a synergistic combination of two types of passive ranging methods: binocular stereo and motion stereo. They describe a new way to model the errors in binocular and motion stereo in conjunction with an inertial navigation system and derive the appropriate Kalman filter for refining the estimates from these two stereo ranging techniques. Results for laboratory images are presented, showing that refined estimates can be optimally combined to give range values which are more accurate than any one of the individual estimates from binocular or motion stereo. By incorporating a blending filter, the approach has the potential of providing accurate, dense range measurements for all the pixels in the field of view