英文摘要 | The research on control and coordination of multiple bio-mimetic robot fishes is studied for essential theories and key technologies for cooperative task of multiple robots under a complex and dynamic underwater environment, which will provide valuable applications such as oceanic investigation and military mission. In this paper, started with a simplified kinematic model for fish, the robot fish's motion control problem is solved. Proceeded with real-time visual information processing, information of the fishes' position and orientation is obtained for positioning in the following coordinate experiment. By utilizing the behavior-based strategies, coordinate tasks are achieved. Finally, all algorithms are all integrated into Multiple Robot Fishes coordinate System (MRFS). Firstly, the conception, advantages and applications of multiple robot fishes' coordination are introduced. The research development and main research directions of both fish's propulsive mechanism and robot fish's control at home and abroad are reviewed. The background and structure of this thesis are also addressed. Secondly, according to characteristics of fish's swimming, both body and movement parameters of bio-mimetic robot fish are extracted. After that, a kinematic propulsive model for carangiform propulsion is simplified and improved. Considered the limitations of the electrical and mechanical apparatus, the robot fish design is optimized. In the mean time, a robot fish prototype is developed, which validates the effectiveness of the design method. Thirdly, the fish's motion control task is decomposed into on-line speed control and orientation control. As for the speed control, a piecewise control strategy and a steady-swimming speed control method integrated RBF neural network with PID controller are proposed; as for the orientation control, a fuzzy logic method is employed in the implementation. Combined the speed control with the orientation control, a point-to-point control algorithm is realized. In the meantime, the path planning of the robot fish is achieved by improved Distance Transform method. Fourthly, an adaptive segmentation algorithm based on hue histogram and saturation histogram is presented. At the same time, on the basis of computer parallel processing technology, the overall visual tracking algorithms are optimally parallelized using assembler-based MMX and SSE instructions. Fifthly, six primitive behaviors are imported, and reasonable conditions of behavior selection are designed. The coordinate motions of the fishes are then accomplished by means of motion strategy based on role allocation, behavior selection and role transition. Sixthly, the control architecture of both hardware and software facing multiple robot fishes' coordination is proposed. Based on the above, a design scheme for the multiple robot fishes experimental system is also laid out. The overall physical experimental |
修改评论