A Study On Coordinated Control Of Multiple Autonomous Underwater Vehicles

The recent years have witnessed people's growing interests of applying of autonomous underwater vehicles (AUVs) in ocean explorations and exploitations, e.g. oceanographic data collection, underwater object inspection and pipes tracking. With its onboard power and instruments, an AUV can be used to explore oceans without physical link to the support ship supply vehicle and intervention of human beings.However, due to the limitations from related technologies such as power, control and sensing, a single AUV can only conduct simple tasks within a small area. To enhance AUV's ability for complex and long duration missions, there are two options: 1) to develop a single AUV with high intelligence, strong capacities and multiple functions; 2) to deploy multiple small AUVs and make them to achieve a cooperated group behavior.The key idea behind the second option is that the coordinated multiple simple AUVs can work in a coordinated manner and act as a group during missions. By employing related control strategies, the complex missions can be reduced to many simple sub-missions which are allocated to corresponding AUVs, each of which is responsible for its own sub-mission and exhibits the team work spirit. Through cooperation and coordination of multiple vehicles, many complex tasks can be conducted. Moreover, coordinated multiple AUVs have much more advantages than single AUV in many aspects, such as spatial distribution, function distribution and higher reliability.Due to the special undersea environment, the existing research on coordination and cooperation of multiple AUVs is much more immature compared with those works for land and air vehicles. In this dissertation, the coordinated operation of multiple AUVs has been studied through theory analysis and numerical simulations, the following efforts have been made in this work, including:1. The modeling and controlling of a single AUV in six degrees-of-free (DOFs). To guarantee the reliability and safety of the system, an overall analysis of the performance of a single AUV is necessary since the coordinated system is based on single vehicles. The existing modeling of marine vehicles in six-DOFs is mainly based on Euler angles which has representation singularities for a pitch angle of positive/negative ninety degrees. To avoid singularities inherently in Euler angle representation, quaternion representation is adopted in dynamic modeling.2. Cooperative path planning for multiple vehicles. During the exploration, the path of an AUV consists of finite sequence of waypoints which are defined a prior. It is desirable to obtain an optimal route due to the limited onboard power. Enumeration or permutation can be applied to find optimal paths for limited waypoints, but they are computationally expensive for large scale of waypoints. The efficient intelligent methods, e.g. genetic algorithm and ant colony algorithm, have been utilized to solve the combinational optimization problem in this work. The proposed path planning algorithm consists of three phases: 1) waypoint assignment: allocating the waypoints to individual AUVs; 2) route optimization: minimizing the total journey of all vehicles by heuristic methods involved in the Traveling Salesman Problem (TSP); 3) route validation: checking if there exists static or moving potential collisions.3. Formation control of multiple AUVs. It is significant for a group of vehicles to maintain certain formation from initial location to a specified destination within an unknown environment. It is believed that such a formation may improve the safety and efficiency of the group. Since there exists possible failure of Leader in the conventional Leader-Follower method, an alternative, Virtual Leader method, is presented in this paper to improve the Leader-Follower method. In this method, the interactions between vehicles are based on artificial potentials. Since the potential force is not continuous for conventional function-based artificial potentials, a fuzzy-logic-based artificial potential, which deals with input and output variables of fuzzy logic controller through membership functions and fuzzy rules, is proposed to generate continuous potential forces.4. Coverage control of multiple AUVs. Suppose a group of initially clustered AUVs, which equipped with sensors for measuring and communicating, are exploring a certain ocean region. In order to optimize the exploring performance, it is desirable to distribute the vehicles according to certain density function. The control law minimizes the locational optimization function, which defines coverage quality, and ultimately drives a group of vehicles to form centroidal Voronoi tessellations (CVT) over the given area with a prescribed geometric pattern, e.g. uniform, circular and Gaussian styles. It is of great important to design a proper penalty coefficient for non-uniform density distribution areas since an improper coefficient would result in unexpected vehicle trajectories. Therefore, a continuous linear penalty coefficient, i.e. penalty function, is utilized in density function to achive the desired distribution.