Study On Intelligent Vehicle’s Motion Planning Under Several Typical Potential Traffic Hazard Scenarios

Self-driving vehicles in the future for a long time with the human city road trafic participants sharing,in terms of bicycle intelligence,automatic driving vehicle in the process of moving,the on-board sensors horizons often blocked by the barrier,form vision blind area,while human traffic participants from obscured area suddenly into the feasible region of self-driving vehicle,can form a kind of keep out the potential traffic danger scene view.Potentially dangerous scenarios of autonomous vehicles safety to improve the intelligent vehicle active safety is vital,and in view of the potential traffic danger situations self-driving cars safe driving motion planning,the existing problems scene of a single study,mainly in view of the intersection visibility limit,but in fact need to be aware of the potential traffic danger scene is much more than this one.Taking urban road environment as the background,this paper studies the motion planning of driverless vehicles in potential traffic danger scenarios.The main research contents and results are as follows.First,the paper analyzes the active and passive limitations of vehicle-mounted sensor perception ability,and defines the concept of sensor perception blind area.Potential traffic hazard scenes are classified according to the state of obstructions and the road form where the vehicle is located.Dynamic occlusion and static occlusion are divided according to whether the obstructions are stationary or not.Static occlusion can be divided into occlusion on the straight road and occlusion on the cross/T-intersection according to the road state,and occlusion on the straight road can be divided into occlusion caused by roadside parking,occlusion caused by roadside buildings and other occlusion.According to the classification,the potential traffic hazards of 10 typical scenarios are analyzed one by one,which provides a theoretical basis for the subsequent model building and motion planning.Second,through the comparison and analysis of several safety distance models(collision avoidance models),a safety speed model is established.Use default location and potential of other position the geometric relationship between traffic participants,set up five representative typical traffic potential dangerous scene safety velocity model,the potential of other traffic participants from sensors to the blind area suddenly appeared in the position and velocity information considered in the model,makes the model more realistic usage scenarios,a more practical application value.The results show that the model is reasonable.Third,the safety speed models under several typical scenes of static obstacle occlusion established in this paper are added to the motion planning method based on Frenet coordinate system to limit the driving speed of autonomous vehicles under potential traffic danger scenes,so as to improve safety,comfort and efficiency.The algorithm is tested under four typical potential traffic hazard scenarios,and the test results show that the motion planning algorithm is reasonable.Fourth,in the actual campus scene,the established safety speed model and the longitudinal speed planning of the improved motion planning algorithm are verified by human driving in real cars.The results show that the planned speed trend is basically consistent with the law of human driving in real cars,and the effectiveness of the model and algorithm is proved within a reasonable range.