Survey on robot multi-sensor information fusion technology

Abstract

Multi-sensor information fusion technology is an emerging technology, which is the foundation of robot intelligent control. This paper introduces the definition of multi-sensor information fusion technology from bionic, mathematic and engineering aspects respectively. Multi-sensor information fusion technology’s working process is explained. The principles, advantages and shortcomings of some common multi-sensor information algorithms such as weighted average method, Kalman filtering, Bayes estimation, Dempster-Shafer evidential reasoning, fuzzy logic and neural networks are proposed. This survey overviews and analyses the study status of multi-sensor information fusion technology applied in robotics field. Some typical applications are given to illustrate the advantages of robot multi-sensor information fusion technology. The authors forecast the developing trend of robot multi-sensor information fusion technology from the following four aspects: sensor manufacture, sampling rate, information fusion algorithm and bio-inspired information.