An Environmental-Adaptability-Improved RatSLAM Method Based on a Biological Vision Model

Inspired by rodents’ free navigation through a specific space, RatSLAM mimics the function of rat hippocampus to establish an environmental model within which agent localizes itself. However, suffers from deficiencies erroneous loop-closure detection, low reliability on experience map, and weak adaptability changes, such as lighting variation. To enhance adaptability, this paper proposes improved algorithm based HSI (hue, saturation, intensity) color is superior in handling characteristics image brightness saturation perspective biological visual model. The proposed first converts raw data RGB (red, green, blue) space into using geometry derivation method. Then, homomorphic filter adopted act I (intensity) channel weaken influence light intensity. Finally, guided filtering used process S (saturation) improve significance details. experimental results reveal that original method terms accuracy template matching robustness.