Implementation Using a MATLAB-Based Rapid Prototyping System

Abstract

he main goal of research on reactive navigationstrategies is to allow autonomous units, equippedwith relatively low-cost sensors and actuators, toperform complex tasks in uncertain or unknownenvironments [7]. These technologies have a widerange of potential application fields, which include theexploration of inaccessible or hazardous environments, indus-trial automation, and also biomedicine. In this research area,the development of the decision and control strategies neces-sary for autonomous operation plays a central role [4], [7].Many studies focus on behavior-based approaches, in whichthe reactivity to unforeseeable circumstances is achieved withcomputationally simple algorithms that process sensory infor-mation in real time by means of high-level inference strate-gies [7]. In this context, fuzzy logic (FL) is often adopted toovercome the difficulties of modeling the unstructured,dynamically changing environment, which is difficult toexpress using mathematical equations [4], [8], [9]. Recentexamples include the coordination of robot soccer teams [9]and the navigation on rugged terrain [8]. To cope withuncertainties and enhance the robustness of navigation, manyresearchers have adopted automatic learning techniques,which allow the FLC to exploit sensory data about theexplored environment not only for autonomous navigationbut also for the adaptation of decision and control algo-rithms. In particular, computational intelligence methodssuch as genetic algorithms, reinforcement learning, and neur-al learning are extremely promising [4].