Research On Intelligent Vehicle Path Planning Based On Fusion Of Image And Lidar Information

The emergence of intelligent vehicles has greatly alleviated the pressure environmental pollution caused by traditional vehicles,as well as the traffic pressure caused by the sharp increase in vehicle ownership.Research on intelligent vehicles is an important direction for the future development of automobiles.And the accurate perception of the external environment is the key to realize the autonomous obstacle avoidance of the vehicle.In order to give full play to the advantages of each sensor,obtain more comprehensive information of the external environment,and complete the real-time obstacle avoidance of the vehicle,this paper is based on the camera and Lidar sensor,and uses the ROS intelligent vehicle as the experimental platform.Research on the path planning algorithm of,the specific content is as follows:(1)Completed the registration of sensor data in time and space.Firstly,the camera calibration is completed based on the Matlab Camera Calibrator module according to the camera imaging model.Secondly,the conversion of the Lidar coordinate system to the world coordinate system is completed through rotation and translation,and the Lidar calibration is completed according to the experimental data.Finally,using the world coordinate system as the conversion coordinates,the unity of the two sensors in space is completed;the unity of the two sensors in time is completed by the interpolation and extrapolation method.(2)An information fusion algorithm based on improved Adaptive Weighting algorithm is proposed.Firstly,the data filtered by the Kalman Filter algorithm is used to replace the average value in the Adaptive Weighted Average algorithm as the estimated value,and an improved Adaptive Weighting algorithm is proposed.Secondly,the improved Adaptive Weighting algorithm and Adaptive Weighted Average algorithm were simulated and verified.Finally,two experimental environments are designed to verify the fusion results of the improved Adaptive Weighting algorithm and Kalman Filter algorithm.The simulation results show that the fusion result of the improved Adaptive Weighting algorithm proposed in this paper is closer to the real measured value,and the data fluctuation is small.The experimental results show that the improved Adaptive Weighting algorithm proposed in this paper is better than the Kalman Filter algorithm in detecting obstacles.(3)A path planning algorithm combining improved Ant Colony algorithm and Dynamic Window method is designed.Firstly,the heuristic function and pheromone update strategy of the traditional Ant Colony algorithm are improved to plan the global optimal path,and the path is smoothed by using cubic B-spline.Secondly,combined with the global path information,the Dynamic Window algorithm is used to set the local target points,and then the preview tracking method is used to track the local target points,which realizes the real-time obstacle avoidance of the vehicle.Finally,by setting up three sets of simulation experiments in different environments,the robustness and real-time performance of the path planning fusion algorithm are verified.(4)A model car test under different obstacle conditions is designed to verify and analyze the feasibility of the algorithm in this paper.Firstly,the intelligent vehicle based on the ROS system is used as the experimental platform,and the SLAM map of the labyrinth-type environment is constructed using the improved Adaptive Weighting algorithm.Secondly,in the constructed SLAM map,the improved Ant Colony algorithm was used to plan the global optimal path.Finally,by adding obstacles of different states to the environment,the model car was tested and analyzed for local obstacle avoidance.The test results show that the vehicle can avoid static and dynamic obstacles smoothly and in real time,and reach the target point safely.It provides application reference value for the subsequent automatic driving of vehicles.