Research Of Unmanned Vehicle Avoidance And Navigation Based On Multi-sensor Data Fusion

With the increasing development of science and technology and sensor performance, the application of multi-sensor data fusion technology is the inevitable trend in all areas. Along with the rapid growth of cars and the rapid improvement of auto-electronics industry, intelligent unmanned vehicles emerge as the times require. Though the navigation and obstacle avoidance of unmanned vehicle is a challenging research, it is one of the heat topics of intelligent artificial and intelligent control areas.In consideration of the pratical significance of multi-sensor data fusion technology, this paper focuses on the research of data fusion technology and obstacle avoidance algorithm of unmanned vehicle. In order to meet the demands of unmanned vehicle navigation and obstacle avoidance system, the multi-sensor data fusion technology is combined applied to unmanned vehicle navigation control system. An advanced A* VFF avoidance algorithm based on fuzzy neural network is proposed. In the end, by building a simulation platform we complete the navigation and obstacle avoidance simulation of the unmanned vehicle. Experimental results show that the unmanned vehicle can plan out a perfect route even in the complex environment. The unmanned vehicle autonomous navigation and obstacle avoidance capabilities have been realized.Firstly, we do research about the background, significance, status quo and development trends on multi-sensor data fusion algorithms both at home and abroad.There are detailed descriptions and explanations of the algorithms’ application in obstacle avoidance on unmanned vehicle navigation. We paid our attention on the analysis of the advantages of the navigation technology based on multi-sensor data fusion.Secondly, in this thesis, the mathematical model of obstacle avoidance navigation sensor, obstacle avoidance navigation sensor coordinate conversion and multi-sensor data fusion theories and multi-sensor data fusion algorithms are studied and analyzed in depth. Also, we focus on the joint Kalman data fusion and artificial neural network data fusion these two data fusion algorithms.Thirdly, through making analysis of unmanned vehicle obstacle avoidance andnavigation system we define the concept of an unmanned vehicle positioning. We learn the navigation obstacle avoidance algorithms on two different aspects which are the global and local obstacle avoidance navigation algorithm. Those algorithms include Dijkstra’s algorithm, A* heuristic search algorithm, genetic algorithm and virtual force field method. The simulation of A* heuristic search algorithm results are given. About the local obstacle avoidance navigation algorithm, we focus on the genetic algorithms and virtual force field method. We pointed out several problems and limitations of the virtual force field method.Finally, a new algorithm called A* VFF is proposed. It is based on fuzzy neural network and combined the multi-sensor fusion technology and obstacle avoidance together. And by building a simulation platform for unmanned vehicle, automous navigation and obstacle avoidance has been realized and we make appropriate analysis from the experiment results.