Research And Implementation Of Behavior-based Reactive AUV Decision-making Control System

Autonomous Underwater Vehicle（Referred to as AUV) is the main tool ofinvestigating the marine resources. Its presence makes the exploration of marineresources has opened a new one. AUV decision-making control system is the keytechnology of studying AUV，and it matters if the AUV can be successfully toaccomplish the task smoothly. This thesis mainly discusses Behavior-based reactiveAUV decision-making control system in order to make AUV complete autonomousnavigation.Firstly, the thesis have described the history of AUV and C-RANGER testingprototype, then made a detail presentation of Behavior-based reactive AUVdecision-making control system. Behavior-based reactive AUV decision-makingcontrol system consists of rational behavior system and perceptual behavior system.Rational behavior system consists of complex algorithms and rules high-levelbehavior, and it makes indirectly control of behavior of the AUV by the underlyingperceptual behavior system. As opposed to rational behavior system, the algorithms ofperceptual behavior system are relatively simple, so it can implement real-timecontrol to AUV. Those two systems are complementary and indispensable.Secondly, in order to solve the local minimum problem of traditional artificialpotential field approach, the relative distance between the current position of AUVwith the target point has been introduced to the repulsive potential field function. Inthe same time, the thesis has introduced control technology of Fuzzy Neural Network.By hybrid learning, this system have optimized the parameters of membershipfunction, come over the static parameters shortcomings of membership function infuzzy control system and dealed with the coordination of the weighting coefficients ofvarious acts. Finally, AUV can output reliable behavior and complete the task even incomplex environment.Finally, we selected several representative behaviors in the decision-making control system and built an effective simulation platform to simulatetending-to-targets behavior and the avoiding-obstacles behavior, by using the controlfeatures of the C-RANGER、MATLAB’s Simulink and adaptive neural network fuzzyinference system tools (ANFIS). Through the analysis of simulation results, we haveobtained that Behavior-based reactive AUV decision-making control system wasfeasible. This system could make AUV complete the task efficiently and reliably evenin the complex environment.