Weakly electric fish use electric organ discharge (EOD) and their electroreceptors to identify prey, explore their surroundings, and communicate with other members of the same species. They are specialized in active electrolocation using a self-generated electric field, and they can sense distortion of their self-generated electric field caused by a target object. Electric fish have many electrosensors on the surface of their body, and the sensor readings from the electroreceptors form an electric image. A correlation exists between features of the electric images and characteristics of a target object. In estimating the location of a target object, the intensity, width, and slope of the electric image must be considered. In this article, we suggest that periodic EOD signals are helpful to extract localization features from noisy electrosensory signals. Cross-correlation between an efference copy signal and sensory signals in the waveform can produce filtered signals in the temporal domain. For a biomimetic fish robot, we can use two-phase filtering: noise-filtering with cross-correlation in the temporal axis and additional filtering in the rostrocaudal spatial axis. This spatiotemporal filtering can effectively remove noise, thus making it possible to obtain accurate localization features of a target object in an underwater environment.
All Science Journal Classification (ASJC) codes
- Experimental and Cognitive Psychology
- Artificial Intelligence