Search

In addition to their visual sense, weakly electric fish use active electrolocation to detect and analyse objects in their nearby environment. Their ability to generate and sense electric fields combined with scanning-like swimming movements are intended to extract further parameters like the size, shape and material properties of objects. Inspired by this biological example, this work introduces an application for active electrolocation based on reduced sensor movement sequences as presented in Wolf-Homeyer et al (2016 Bioinspir. Biomim. 11 055002). Initially, the application is conducted with a simulated receptor-system consisting of an emitter-dipole and an orthogonally arranged pair of sensor-electrodes. Close inspection of a minimal set of scanning movements allows the exclusion of sectors of the general search area early in the proposed localization algorithm (search area partitioning). Furthermore, the proposed algorithm is based on an analytical representation of the electric field and of the so-called EEV (ensemble of electrosensory viewpoints) (Solberg et al 2008 Int. J. Robot. Res. 27 529-48) rather than using computationally expensive FEM simulations, rendering it suitable for embedded computer systems. Two-dimensional discrete EEV contour-ring points (CRPs) of desired accuracy are extracted. In the core of the localization algorithm, fragments of the EEV are selected from valid sectors of the search area, which generates sets of CRPs, one for each sensor-emitter position/orientation. These sets are investigated by means of a nearness metric to find points in different sets which correspond to each other in order to estimate the object position. Two resultant scanning strategies/localization algorithms are introduced.

Weakly electric fish use self-generated electric fields for communication and for active electrolocation. The sensor part of the biological system consists of a vast amount of electroreceptors which are distributed across the skin of the electric fish. Fish utilise changes of their position and body geometry to aid in the extraction of sensory information. Inspired by the biological model, this study looks for a fixed, minimal scanning strategy compiled of active receptor-system movements that allows unique identification of the positions of objects in the vicinity. The localisation method is based on the superposition of numerical extracted contour-rings of rotated and/or linearly shifted EEVs (Solberg et al 2008 Int. J. Rob. Res. 27 529-48), simulated by means of FEM. For the evaluation of a movement sequence, matrices of unique intersection points and respective contrast functions are introduced. The resultant optimal scanning strategy consists of a combination of a linear shift and a rotation of the original EEV.