Abstract:The computation of the time-to-collision (TTC) in collision course is a critical issue for an animal to response quickly and accurately to avoid a looming object. The physical formula of TTC was not feasible for neural computation realization. We deduced a new approximation of TTC and constructed a neural computational model (NCM). The simulated results showed that NCM with neurons of sigmoid response can accurately estimate TTC and well reconstruct the practical animal collision experiments. NCM might be a potential and realizable model to imitate the physiological neural mechanism of the visual information processing in the looming collision course.

Collision Detection Using a Neurual Computational Model