A fuzzy approach to map building

In this paper, we present test results for a fuzzy map building method (introduced in [2]). We also introduce modifications to the sensor's model and map-update scheme and compare results. Both sensor models and map updates gave encouraging results manifested by recognising cluttered objects as separate entities, detecting small objects and discovering relatively small gaps between objects. The proposed methods can update the map in real-time. The fuzzy map update fusion functions provided a means of tweaking the map adaptability in dynamic environments. Thus, the proposed methods are suited to real-world applications. It has been shown that the direct update scheme suits real-time obstacle avoidance applications, while the fuzzy fusion functions are more appropriate for environment mapping and path-planning applications.