A Brain-Inspired Visual Fear Responses Model for UAV Emergent Obstacle Dodging

Dodging emergent dangers is an innate cognitive ability for animals, which helps them to survive in the natural environment. The retina-superior colliculus (SC)-pulvinar- amygdala-periaqueductal gray pathway is responsible for the visual fear responses, and it is able to quickly detect the looming obstacles for innate dodging. Inspired by the mechanism of the visual fear responses pathway, we propose a brain-inspired emergent obstacle dodging method to model the functions of the related brain regions. This method first detects the moving direction and speed of the salient point of moving objects (retina). Then, we detect the looming obstacles (SC). Third, we modulate attention to the most dangerous area (pulvinar). Fourth, if the degree of danger exceeds the threshold (amygdala), the unmanned ariel vehicle (UAV) moves back to dodge it (periaqueductal gray). Two types of experiments are conducted to validate the effectiveness of the proposed model. In a simulated scene, we simulate the process of mice's fear responses by putting looming dark lights shining on them. In a natural scene, we apply the proposed model to the UAV emergent obstacles dodging. Compared to the stereo vision model, the proposed model is not only more biologically realistic from the mechanisms perspective, but also more accurate and faster for computation.