Research On Formation Control And Cooperation Mechanism For Multiple Autonomous Underwater Vehicles With Limited Communication

The autonomous underwater vehicle (AUV) can perform complex, arduous and evendangerous tasks in marine environments instead of human beings, such as the seabedexploration, underwater operation, and offshore battle. Due to the advantages of multipleAUVs and the development of underwater vehicles, multiple AUV cooperation becomes anew application style. Compared to the autonomous mobile robots on land, the severeunderwater environments bring in more difficulties to the research and design of multipleautonomous underwater vehicle system (MAUVS). The drawbacks of underwater acousticcommunication, such as the unelectable communication delays, communication failures andhigh bit error rate, are great challenges to the coordination and cooperation of MAUVS. Thecore of the dissertation is the cooperation control of MAUVS under conditions of limitedcommunication, respectively, focusing on the formation and its stability with communicationdelays and communication failures, the coordination and cooperation mechanism based on thelocal sensing and limited communication range. This main research work is as follows:Firstly, the stability of the group formation system with a double integral dynamics isstudied. For the large delayed information exchange through underwater acousticcommunication, a delay-dependent distributed control law is adopted to generate the controlinput in order to enable the formation converge to the specified speed. Then the properties ofthe Laplacian matrix are used to take transformations on the system. The stability conditionsand convergence are investigated according to the Nyquist stability criterion with thefrequency domain analysis method. The relationship between the upper-bound delay andcontrol parameters, the communication topology is also obtained. The analysis results verifythe necessary of prediction item in the control law. For the problem of underwater acousticcommunication failure or accidental interruption, the formation system is taken as a switchingtopology with Markov chain nature. The properties of the irreducible and aperiodic stochasticmatrix are used to analyze the conditions that ensure the group with consensus algorithm canconverge to the stable specified formation. Experimental results show the correctness of thecontrol law and the stability conclusions which solve the group formation problem withcommunication delays and communication failures. The work is of practical value for MAUVS under conditions of limited communication.Secondly, combining with the dynamics of AUV, the stability of multiple AUV formationsystem is studied on the basis of the above research with communication delays andcommunication failures. Taking a type of actuated underwater vehicle as example, thedynamic model of the underwater vehicle is expressed as a double integrator by feedbacklinearization. Then the formation control law and the stability conclusion derived by thedouble integrator can be used to produce the control input for underwater vehicles in case ofcommunication delays and communication failures. The control laws are justified to enablemultiple AUVs converge to the specified formation under certain conditions. For the problemof communication failures, the effect of DR algorithm to convergence process is analyzed.The experimental results show the validity of formation control strategy and the conclusions.Thirdly, the coordination and cooperation mechanism is studied with the target searchtask. The task process is divided into three stages: searching for target clues, tracking thetarget clues and locating the accurate position of the target. Taking seafloor hydrothermaldetection for example, multiple AUVs move in formation pattern in the global range to searchfor clues within larger water area at the beginning stages, and then track clues in local rangeuntil they find the accurate position of the target. For the restrictions of limitedcommunication range and poor communication quality of underwater acousticcommunication, trophallaxis communication is introduced to emerge the direction of thegradient of the "virtual food" within the AUV group. The swarming behaviors of fish areimitated to track clues for AUVs. The experimental results show that crowding factor withinthe group and the threshold of the "virtual food" impact on the system performance. Also, theintroduction of fish behaviors and the increase of crowding factor help to improve theefficiency of the system.Finally, the application of biological trophallaxis behavior for multiple AUV cooperativecollection task is considered with the application background of marine mineral specimen orseabed sulfide sample collection mission. An internal variable is defined for each AUV tomeasure its own working state. The possibility and the direction of trophallaxis are judged bycomparing internal variables of different AUVs. The finite state machine is proposed todescribe the switching process of AUVs between different states. Differential equations arealso used to describe the distribution of AUVs on each state over time from a macro perspective. The effect of parameters, state transition probability, and other factors on systemperformance is analyzed on the basis of mathematical model built as above. Comparing theproposed method with other approaches, experimental results show the introduction oftrophallaxis behavior improves system performance effectively.