Virtual Obstacles Based Intellgent Vehicle Navigatton Method Research

With the increase in the number of cars,Urban traffic problems are getting worse.Intelligent vehicle technology is drawing attention, for it is an effective solution to solve the urban traffic problems, such as the safety and efficiency problems. In this paper, with the intelligent vehicle applications under the urban environment as the background, the smart car navigation as the research object. In the field of path planning and control, for only on one kind of path planning methods has the inherent problems (such as global path planning method is unadapted to dynamic environment and local path planning maybe have the plight of local minimum points, etc.). And the adaptability of existing integrated path planning for multi-task and different environments is poor. It is often necessary to be adjusted with changes in application environment and task. Therefore, this paper proposes an integrated path planning method based on the virtual obstacle. This method can fully utilize the map information to get a global optimal path and avoid the slow obstacles on the road in real-time. On this basis, the corresponding localization method is selected through the analysis of the existing technology and application environment, then the navigation methods based on virtual obstacles is formed. This navigation method can complete various tasks under the urban environment such as the cruise lane, obstacle avoidance and vertical automatic parking, etc. The details of this article as follows:Firstly, this paper briefly introduces the sensor configuration and calibration. Then based on the analysis of the existing technology and application environment, the preliminary design of the navigation program is presented. In the field of perception and positioning, high-precision GPS (RTK-GPS) and dead reckoning method is used. By using the method of EKF(Extended Kalman Filter) for pose estimation of vehicles, it not only has the advantage of high precision with GPS ,but also could make up the shortcomings of RTK-GPS in the low data frame rate and the signals easily interference by obstacles. Laser radar is selected for obstacles detection sensor. In the field of path planning and control, for only on one kind of path planning methods has the inherent problems, And the adaptability of existing integrated path planning for multi-task and different environments is poor. This paper proposes an integrated path planning method based on the virtual obstacle. First, network of the road is seen as a set of nodes and lines. And the intersection point of the node as a candidate, the A* algorithm is used to gain a global optimal path. Second, virtual obstacles are generated according to the global optimal path. Finally, the local path is planned by using the VFH+ (improved Vector Field Histogram)method according to the virtual obstacles and actual obstacles from laser radar. Our method can fully utilize the map information to get a global optimal path and avoid the slow obstacles on the road in real-time.Then, used in road cruise and obstacle avoidance of intelligent vehicle, Navigation method based on virtual obstacle is described in detail. The concept, generation method and selection method of virtual obstacles are fully elaborated. The memory method for the obstacle on both sides of the vehicle is presented, etc. Aiming at the problem that vector field histogram (VFH+) method is sensitive to threshold and the parameters of evaluation function are difficult to be determined, VFH+ is improved in this paper. At the same time, a large number of navigation experiments are done in campus environment of Shanghai Jiao tong University. The results of such experiments prove the effectiveness, real time and reliability of the navigation method in this paper.Finally, the navigation method is used in vertical automatic parking of intelligent vehicle. It uses arc - straight line - circular arc method as a global optimal path planning method. Aiming at the difficulty for VFH+ in the situation of the car in the garage, the improvements have been given. The results of experiments under the real environment prove the effectiveness of the method in the vertical automatic parking. And the above applications of the method are fully illustrated that it has the adaptability for multi-task and different environments.