Research On Intelligent Driving Control Strategy Integrating Visual Information Of Traffic Cones Placing And Picking Up Vehicle

With the continuous development of China’s economy,the mileage of China’s highway construction continues to increase,and the maintenance work is heavy.As an important highway maintenance vehicle,traffic cones placing and picking up vehicle can effectively improve the efficiency and safety of traffic cones laying and recycling,and have a great market demand.The vehicle’s driving environment is relatively closed and its speed is low.It is feasible to realize automatic driving in the specific scene and it is of great significance to further improve the efficiency of the vehicle operation and reduce the difficulty of operation.This paper integrates visual information to research the vehicle’s automatic driving control strategy.The specific work is as follows:Firstly,the characteristics of automatic cone machine driving environment and operation requirements are analyzed.According to the vehicle control constraints and requirements,the comparison and analysis of the characteristics of existing environmental sensing technologies are carried out.And the overall scheme of automatic cone machine control system which combines vision and GPS information is determined.Secondly,the demands of vehicle perception system for visual environment are analyzed.Based on deep learning technology,the fusion scheme of visual recognition technology is determined.On the basis of this,a detailed tagging database of seven kinds of objects and scenes,including lane lines,traffic cones,etc.,is made.In this paper,the structure of different depth convolutional neural networks is compared and analyzed seek the optimal network structure for image semantic segmentation.The final accuracy of the deep model is 84%,with a forward speed of 12fps,which can meet the needs of the vehicle.Thirdly,the mapping model between the image visual coordinate system and the physical world coordinate system is established and calibrated.And the curve model of the lane line is obtained accordingly.Based on the influence of different starting position,front wheel angle,the location of traffic cones and the distribution of road lines,the driving path planning strategy and the driving path updating strategy under various operating conditions are analyzed.Combining the action state of the traffic cone manipulator,the planned route and the requirements for the speed of the vehicle under various operating conditions,the vehicle control strategy was analyzed based on the goal of high efficiency and safety.Finally,the co-simulation of this vehicle is established based on MATLAB/Simulink and CarSim.The ideal driving trajectory is obtained by the path planning model and control strategy.The driving speed and lateral displacement are controlled by the preview follow control method based on PID.The simulation results show that the longitudinal velocity error of the vehicle is 0.078 m/s,the longitudinal displacement error is 0.068 m,and the transverse error is 0.0048 m.This verifies the rationality of the vehicle destination path planning result and the effectiveness of the control method.