Ultrasonic and stereo vision data fusion

Recent years have seen an increasing interest in the development of multi-sensory robot systems. The reason for this interest stems from a realization that there are fundamental limitations on the reconstruction of environment descriptions using only a single source of sensor information. If robot systems are ever to achieve a degree of intelligence and autonomy, they must be capable of using many different sources of sensory information in an active and dynamic manner. A central objective of this paper is the belief that only through a quantitative description of sensor and system performance can we ever hope to understand and utilize the complexities and power provided by a multi-sensor system. A quantitative description of performance provides an a priori means of developing sensor capabilities and reasoning about the cumulative effects of observation uncertainty. The descriptions that are used are aimed to develop algorithms for the dynamic exchange of information between sensor systems and the construction of active sensor strategies