This paper presents a novel map building approach for path planning purposes, which takes into account the uncertainty inherent in sensor measurements. To this end, Bayesian estimation and Dempster-Shafer evidential theory are used to fuse the sensory information and to update the occupancy and evidential grid maps, respectively. The approach is illustrated using actual measurements from a laboratory robot. The sensory information is obtained from a sonar array and the Scale Invariant Feature Transform (SIFT) algorithm. Finally, the resulting two evidential maps based on Bayes and Dempster theories are used for path planning using the potential field method. Both yield satisfying results