Research For Unmanned Vehicles Navigation Method In Urban Environment

This paper focus on navigation solution of unmanned vehicle in urban environment.Fully autonomous vehicle is the solution of congestion and traffic safety.Nowadays,the autonomous vehicle has the problem of high cost,low reliability and poor adaptability.We put forword to give an autonomous vehicle navigation solution for productive vehicle.Firstly,this paper proposes a system architechture and give a sensor placement which has a low cost and a high degree of integration.At the same time,it can also meet all kinds of scene perception needed in urban environment.For multiple sensors and multiple computers,a distributed architecture network transmission is set up to ensure that the communication is stable and reliable.Secondly,this paper studies the method of local environment perception based on laser.And how to utilize multi-line laser to form grid map and ground detection.Besides,ILP feature is used to speed up road curb detection and road model is fitted.Radar is used for dynamic vehicle detection and they are tracked in grid map.We speed up the operation efficiency of particle filter.Then we combine the all perception result by calibration the sensors.Thirdly,global path planning is done on digital map.A* algorithm is used to find the best way and the path is optimized by B spline curve.What's more,A* algorithm is also used in high resolution digital map to find a lanelevel path.Finally,local path planning is done based on the results of local environment perception and global path planning.Bezier curve is directly used in structured environment and an improved RRT algorithm is used in unstructured envirnomnet.We have done a lot of test in Zizhu and campus of Shanghai Jiaotong University.We also have a good result in Chinese Future Challenge of Intellegent Vehicle.