Design And Implementation Of Automobile Simulated Driving System Fusing Video Images And Driving Control

As the number of applicants for driving schools increases year by year,people have put forward higher requirements for traditional driving training methods.However,the current traditional training methods still have shortcomings such as long learning cycle,high cost,and inability to guarantee absolute learning to drive safety.Compared with traditional training methods,training on simulated driving systems has the advantages of being unaffected by the weather environment,low energy consumption,low cost,and absolute safety.However,there is a big gap between the simulated driving system and the real driving,which is mainly due to the obvious gap between the picture display effect of simulated driving and the real driving scene.With the gradual maturity of virtual reality technology,the introduction of virtual reality technology into the simulated driving system can well improve the shortcomings of traditional simulated driving.In order to solve the above problems,this paper designs an automobile driving simulation system which integrates video images and driving control.The virtual scene of this system is built by full-motion video,and the driving effect can be closer to that of real vehicle driving with VR display equipment.Firstly,according to the requirements of system driving control performance,the vehicle dynamics model of this system is established,and the improved Euler method is selected to solve the model to realize the driving control of the system;secondly,in order to meet the real-time detection requirements of the driving collision system,the real-time detection of the virtual scene image is carried out by using the yolov3 algorithm to obtain the pixel coordinates of the detection target(person and vehicle).At the same time,because the system uses panoramic video to build the virtual scene,which makes the target in the picture lose the depth information.Therefore,this paper uses the calibration experiment and the data regression method to realize the distance modeling of the scene target,and obtains the position mapping relationship between the scene target and the real target.Finally,the distance model is used to map the detected pixel coordinates in real time to realize the collision function of the system.In addition,this article describes in detail the specific implementation of the simulated driving system in the Unity3 D engine,including the construction of virtual scenes,the realization of the driving control process,and the realization of the collision function,and carries out experimental test and analysis on the whole system.In the test,each function module of the system works normally,the driving picture is realistic,and the driving control is more precise.Although this system is still in the experimental development stage,but it still has certain practical significance and application value.