Fuzzy logic inference for occupancy state modeling and data fusion

Autonomous robotic systems require a detailed model of space occupancy to be built from sensory information in order to navigate safely in their environment. Probabilistic occupancy models have been proposed that use conditional probabilities evaluation to merge redundant measurements. These approaches provide meaningful representation of space but require important approximations to remain computationally tractable for high dimensionality. As a result, the strict definition of probability is denatured. The present paper proposes an exploration of the fuzzy logic paradigm as a modeling tool for occupancy mapping in the context of workspace representation for robotic applications. A computationally tractable fuzzy logic inference engine is introduced that allows data fusion to construct a robot workspace representation in a more intuitive way while preserving desirable characteristics achieved by probabilistic modeling schemes.