Brain-inspired obstacle detection based on the biological visual pathway

Abstract

Obstacle detection is crucial for intelligent systems (e.g. robots, unmanned ariel vehicle) that interact with the real world. This paper proposes a brain-inspired rasterization algorithm for obstacle detection. Rasterization algorithm is inspired by the information processing mechanism of the biological brain (including arthropod brain and human brain). Obstacle detection relies on feed forward and feed backward information processing mechanism. Receptive fields in every level of abstraction transmit different sizes of image regions to higher levels. Feedback is related to modulating attention about the position and size of target receptive field. Inspired by the circuit in human vision system, this paper provides a computational model for obstacle detection. Good performance on the experiments supports the proposed theoretical model. The major contribution of the proposed brain-inspired rasterization algorithm is that it can detect obstacle in any size from any direction without any preprocessing.