| The challengeable research of AUVs formation controlling has received considerable attention, which has become one of the classic and representative problems in Autonomous Underwater Vehicle (AUV) field. The contents of this thesis are discussed as follows: AUVs system architecture, formation path planning based on Fuzzy Q-Learning algorithm, hierarchical formation technology based on leader.At the first place, according to the classic architecture of MAS, AUVs formation system architecture based on blackboard communication is proposed, which includes communication module, motion control module, interaction with environment module. Three modules are described in detail respectively. The architecture can strengthen the capabilities of coordination and cooperation among AUVs.In addition, introduces reinforcement learning theory and fuzzy controlling technology. To solve the problem of AUVs collision avoidance with the environmental objects (including the static obstacles and the other AUVs) while AUVs are moving forward in the expected formation, the thesis makes some research on path planning and collision avoiding methods, puts forward collision avoiding policy which combines Fuzzy Inference and Q-Learning algorithm, designs Fuzzy Membership Functions and Fuzzy Inference Rules through MATLAB tools.Thirdly, according to the problems of formation controlling field, the thesis presents a hierarchical formation system based on leader-follower strategy, including task planning level, pattern designing level, behavior controlling level, formation evaluating level. Formation fault tolerant controlling is discussed primarily, formation exchanging policy based on sensor information and controlling matrix is brought forth.Finally, the simulation platform of AUVs formation is developed through VC++ software, different formation cases are designed and simulated under different environments, the results show that the algorithms are verified reasonably and effectively. |
PDF Full Text Request