Sensor fusion in mobile robot: some perspectives

Abstract

In this paper, techniques and theory work of multiple sensor fusion in mobile robot are reviewed. Mobile robot needs to integrate multiple sensors to accomplish tasks such as map building, object recognition, obstacle avoidance, self-localization and path planning. Our survey describes sensor fusion in three categories: 1) statistically based fusion algorithm policies need the a priori knowledge about the observation process to make inference about identity; 2) neural network and fuzzy set based fusion policies are distribution free and no prior knowledge is needed about the statistical distributions of the classes in the data source in order to apply these methods for fusion; and 3) information theoretic fusion algorithm policies make use of a transformation or mapping between parametric data and a resultant identity declaration. Techniques such as Kalman filtering, rule-based techniques, behavior based algorithms, and approaches range from Bayesian theory, Dempster-Shafer evidence theory to fuzzy logic and neural networks are reviewed. The paper concludes with further research directions.