Towards a biologically-inspired model for underwater localization based on sensory-motor coupling

Abstract

In recent years there has been a growing interest in studying biologically-inspired models for localization that can cope with the stochasticity of natural environments. Most of the works available have considered 2D navigation tasks in static and structured environments, such as the lab maze or the corridors of a building. Reactive navigation is usually employed, so the robot avoids colliding the walls or maintains itself in the center of the road. Thereby, the localization error is bounded by the physical structure of the environment. A different situation is encountered in underwater scenarios, where the agent has to move through the open 3D space, the perceptual cues are scarce, and the scene is dynamic. In this work, we present the design of a biologically-inspired model for operation in underwater missions. The main novelty of our proposal consists in considering the information redundancy provided by the sensory-motor coupling. Thus, perceptive information is anticipated as a means to improve the processing efficiency for large areas, to handle the dynamic aspect of the scene, and to detect abnormal situations. A preliminary evaluation conducted in simulation is reported in this paper, showing the plausibility of the model.