| With the development of China’s Fourteenth Five-Year Plan,the new development pattern to accelerate the construction of China’s transportation industry will be more vigorous upward development.As the expressway network becomes more and more mature and complicated,the functions of the dedicated lanes for autonomous driving need to be relatively perfect.However,there are few researches on the design of the traffic signs for the dedicated lanes for autonomous driving in China.In the process of vehicle identification,there are too many external interference factors,which may easily lead to many safety risks.Therefore,the adaptability and matching between expressways and autonomous driving vehicles have also attracted wide attention from all walks of life.Scientific and reasonable setting of traffic signs for autonomous driving lanes can effectively avoid the influence of adverse environmental factors,correctly guide traffic flow,and provide guarantee for the safety of autonomous driving vehicles on expressways.The main theoretical research contents of this paper are as follows: firstly,the paper studies the traffic sign image technology of automatic driving vehicle detection and recognition,and analyzes the various factors that affect the traffic sign recognition.Secondly,the different working conditions that affect vehicle identification of traffic signs are divided into different working conditions,and the scene modeling of different working conditions is carried out in the Prescan software.Thirdly,Prescan software is used to carry out automatic driving simulation tests in different scenarios.By co-simulation of Prescan and Simulink,indicators such as reaction speed and reaction time of automatic driving vehicles are obtained.Besides,horizontal comparison and analysis of the changes among the indicators are made,and it is found that with the decrease of weather visibility and rainfall.With the increase of snowfall,the recognition accuracy will decrease.When the layout size of traffic signs is set to120 cm and the tilt Angle is 15,the recognition accuracy will be improved.Finally,the random forest algorithm is used to make decisions on the RGB and HOG eigenvalue texture and size of traffic sign images.The results show that the random forest evaluation model can identify the RGB and HOG feature and texture of traffic signs under different weather conditions,and it has a high classification accuracy when the number of decision trees in the random forest is about 40.Under different traffic sign Settings,the number of decision trees in the random forest is obtained with a high classification accuracy between 50-70 by classifying and recognizing indicators such as the size and texture of the RGB and HOG feature value of the sign position Combined with the actual application of the project,in the Beijing-Taiwan Expressway section from K516+360 to K518+360,according to the experimental conclusions in this paper,the setting of the speed limit traffic signs is simulated,and the feasibility of the design scheme is verified by the random forest evaluation model,which provides a reference method for the design of traffic signs in the special lane for automatic driving. |
