Fuzzy logic rules for mapping sensor data to robot control

Abstract

We use fuzzy logic rules to directly map sensor data to robot control outputs by classifying a set of typical subtasks, such as path tracking, local collision avoidance, contour tracking, situation evaluation, etc. With the help of existing heuristics, the decision-making process for each subtask can be modelled and represented with IF-THEN rules. The underlying concepts of mapping with fuzzy logic rules are briefly explained by considering the proximity sensors, the control of speed and steering angle of a mobile robot. The development of these fuzzy rules is explained, typical rules for dealing with various motion situations are listed. The modularly developed fuzzy rule bases can be integrated to realise task-level programming and the exploration task. Experiments with the mobile robot validate this concept.