MOTION CONTROL AND LOCALISATION OF ROBOTIC FISH

Abstract

Biological fishes are efficient swimmers, exhibiting many desirable features such as high power efficiency while minimal noise to the surroundings. Such characteristics are highly desirable in underwater robots and hence, learning from nature, robotic fishes have been invented. Prior to mimicking the required features to the robots, some of the essential aspects of robot motion control system must be addressed. The aim of this dissertation is to study and establish some technical approaches towards robust motion control and precise on-board localization of multi-link robotic fishes. First, both angular and linear motion of the robotic fish is controlled by different approaches, namely, model parameter adaptation, data-assisted modelling and analytical model derivations. Second, on-board localization is attained via vision feedback system and a particle filter technique. Collectively, this dissertation assists a robotic fish localization and motion control - completing the closed-loop motion control system of the robotic fish.