Cooperative Circumnavigation and Tracking of an Irregular Shape Target

Collective behaviour happens not only among humans but also across several species of the animal kingdom. It depends on local interactions among individuals, which coordination is driven by the spread of a general information. By taking inspiration from biological systems, researchers and engineers are trying to reproduce the cooperative behaviour of humans and animals like ants and birds, giving birth to multi-vehicle systems, with the aim of solving several problems. A multi-vehicle system is composed of interconnected vehicles coordinated to act collectively towards global objective. When controlling such systems, the goal is to obtain a coordinated behaviour through local interactions among vehicles and the surrounding environment. The aim of the degree project is to perform a multi-vehicle circumnavigation and tracking of an irregular shape target, by implementing estimation and control algorithms. The degree project is based on the licentiate thesis performed by PhD student Joana Filipa Gouveia Fonseca, concerning the multi-vehicle target tracking. The mentioned thesis has been developed in particular to monitor and study colonies of algae which grow out of control and produce toxic or harmful effects. These are known as harmful algal blooms, or HABs. They appear frequently and have a substantial negative effect, mainly on the environment but also on humans, causing large-scale mortality of fish, mammals and birds. Through this master thesis, we aim at the achievement of a solution to the HABs problem, by applying the collective behaviour concept to a multi-agent system. Starting from the literature, a preliminary work has been developed in Matlab/Simulink to get familiar with the algorithms presented in the licentiate thesis. Afterwards, it is introduced the estimation and control's algorithm, developed in a simulating environment, provided by the Gazebo simulator. The approach presented in this thesis consists of three important steps: the estimation of the target, the agent's arrangement in a regular configuration, the circumnavigation of the target by the system. This method involves the estimation of an irregular shape target, on the basis of the agents local measurements. Then, after the determination of a leader agent, to which the remaining robots are subordinate, each agent is positioned around the target. Finally, the whole system assumes a regular configuration, and circumnavigate the target. Computer vision and estimation algorithms have been developed by using the computer vision library OpenCV in the ROS framework. Circumnavigation has been achieved thanks to the development of Python code. Final results are discussed, presenting pros and weakness of this work.