Efficient Fusion Technique for Disparate Sensory Data
Hòng Xu · 1991
Multisensor system has received increasing attention in recent years. Not all sensors can provide complete estimates of robot's location 27rk paper proposes a more general methodology for multisensor data furion. i'le sensory data may be partial or indirect. This method introduces definiwn of three primitive sensory data types and provides general, sensor independent and practical solutions for fusion of differenl types of data. For a multisensor system, thk method can deal with dkparate senrory measurements for position or relatwnshb, implement integration of derived information in an efficient manner and give more accurate estimates. A case shuiy as well as Monte Carlo simulations illustrate application of the presented method to greatly improve the position and orientation estimation for a mobile robot and show the statistical effect of two dimenswnal data integration. infomation handling and no general method to integrate the partial and indirect sensory measurements. Actually, there are a number of sensors whose measurements either are partial estimate or provide implicit information. A relocalization system using laser scanning sensor and position reference beacons, for example, cannot produce a locational estimate but only constraints to the current location if there is not enough beacons perceived on the spot. This paper proposes a general methodology for disparate sensor data fusion. Definitions of three primitive sensory data types are introduced in the following section, the general solutions for data fusion of different types are presented in section 3. Finally, a mobile robot is taken as example to illustrate application of presented method to improve location estimation by handling incomplete sensory information and Monte Carlo simulation shows the performance.