Sensing is a fundamental operation for almost any motion-based system. We have chosen electrosensory systems as a platform to explore sensing and control in both artificial and biological systems. Electrosense is a convenient sensing modality because artificial electrosensory systems are relatively simple to implement, and weakly electric fish have some of the most well-documented sensorimotor pathways. In particular, active electrolocation is investigated, where the task is to estimate the location of a target using measurements from a self-generated electric field. The fundamentals of electrolocation are described first with a finite-element numerical approximation of the governing equations, and then simple models are used to predict electrosensory observations. Several belief maintenance schemes are employed to fuse sensor data and explicitly account for uncertainties in the position of the target. In the biological realm, a protocol for simulating the sensory acquisition and belief maintenance during prey-capture behavior in the weakly electric fish was developed. Using these simulations optimal sensing was investigated, and results provide insight into the interdependencies and co-evolution of sensing and motion systems of the weakly electric fish. In the artificial realm, an electrosensory robot capable of actively locating underwater targets by measuring perturbations in a self-generated electric field was built. Using seven different control algorithms, the robot can successfully locate nearby targets, as well as localize itself when placed in a pre-mapped environment, in both fresh and saltwater.

Last modified

• 08/27/2018

Creator

• James R Solberg

DOI

• https://doi.org/10.21985/N2S42S

Subject

• Mechanical Enginnering

Keyword

• Weakly Electric Fish
• Electrolocation
• Electrosense
• Active Sensing
• Bayes Filter

Date created

• 2007-11-30

Resource type

• Dissertation

Rights statement

• In Copyright