Evolutionary Purposive or Behavioral Vision for Camera Trajectory Estimation

摘要

Active, animate, purposive or behavioral vision are all understood as a research area where a seeing system interacts with the world in such a way of creating a balance between perception and action. In particular, it is said that a selective perception process in combination with a specific motion-action works as a unique complex system that accomplishes a visuomotor task. In the present work, this is understood as a visual behavior. This work describes a real-working system composed of a camera mounted on a robotic manipulator that is used as a research platform for evolving a visual routine specially designed in the estimation of specific motion-actions. The idea is to evolve an interest point detector with the goal of simplifying a well-known simultaneous localization and map building system. Experimental results shows as a proof-of-concept that the proposed system is able to design a specific interest point detector for the case of a straight-line displacement with the advantage of eliminating a number of heuristics.