A method for assessing the danger level of the surrounding environment of an autonomous driving vehicle

Abstract

The present invention discloses a method for evaluating the danger level of the surrounding environment of an autonomous driving vehicle, including obtaining the original information of target vehicles around the vehicle through multiple sensors, and obtaining the single target danger level based on the road information obtained based on the high-precision map; obtaining the multi-target danger level through normalization according to the single target danger level and the weight of the corresponding target vehicle; obtaining the traffic flow target type complexity, traffic flow density, traffic flow speed difference and traffic flow acceleration difference according to the original information, and then obtaining the traffic flow danger level; obtaining the current road type, lane increase or lane reduction, ramp diversion or ramp confluence according to the road information of the current road, and then obtaining the road form danger level; obtaining the longitudinal planning environmental danger level based on the multi-target danger level, traffic flow danger level and road form danger level. The evaluation method can effectively improve the environmental perception ability in complex traffic scenes and improve the safety and comfort of autonomous driving.

Description

Technical Field

[0001] The present invention belongs to the technical field of autonomous driving, and in particular relates to a method for assessing the danger level of the surrounding environment of an autonomous driving vehicle. Background Art

[0002] Autonomous driving technology mainly includes three parts: environmental perception, decision-making planning, and control execution. Environmental perception is the starting point of the technology chain and plays a key role. In the current situation where sensor capabilities are poor, detection accuracy is insufficient, and the types of detectable targets are limited, it is particularly important to fully exploit and use the available information. In order to improve environmental perception capabilities in complex traffic scenarios, it is necessary to integrate information such as road morphology, driving intentions and driving status of targets around the vehicle, and traffic flow status. The scene diagram is as follows: Figure 1 As shown in the figure, RT1 to RT6 around the vehicle are all target vehicles around the vehicle, which evaluate the danger level of the surrounding environment of the autonomous driving vehicle. This can ensure the rationality and advancement of decision-making planning, reserve sufficient processing time for the control execution part and improve driving comfort.

[0003] There are two main types of existing hazard level estimation algorithms:

[0004] (1) Danger level assessment methods based on the relative motion relationship between a single target vehicle and the vehicle, such as Chinese patents CN201910012116.X and CN201110164481.6, etc. In these methods, the impact of a single target on the safe driving of the vehicle is assessed, and decision-making planning is performed for dangerous targets without considering the environmental conditions. The assessment results are too simple and mechanical, making it impossible for planning and control to distinguish dangerous targets under different traffic conditions. There is a big gap between the driving process of human drivers and the rationality is obviously insufficient.

[0005] (2) The target danger level estimation method based on the environment state, target vehicle state, target vehicle intention state and target vehicle type, such as patent CN201910208600.X, takes into account the impact of the environment state on the target danger level. However, it does not comprehensively consider the danger levels of multiple targets in the environment, and cannot reflect the dangers caused to the vehicle by the complexity of traffic flow and driving environment. Summary of the invention

[0006] In view of the above-mentioned deficiencies in the prior art, the purpose of the present invention is to provide a method for assessing the hazard level of the surrounding environment of an autonomous driving vehicle. This assessment method can effectively improve the environmental perception capability in complex traffic scenarios, effectively improve the accuracy of the environmental hazard level, and further improve the safety and comfort of autonomous driving.

[0007] The technical solution of the present invention is achieved in this way:

[0008] A method for evaluating the danger level of the surrounding environment of an autonomous driving vehicle comprises the following steps:

[0009] S1: The original information of target vehicles around the vehicle is obtained through multiple sensors, and then the target attributes of each target vehicle are obtained based on the original information, including the cut-in probability, collision time and deceleration required for the vehicle to avoid collision with the target vehicle. At the same time, the road information of the current road is obtained based on the high-precision map, and the location of the target vehicle is integrated with the road information to obtain the regional attributes of the target vehicle, and then the single target danger level is obtained;

[0010] S2: According to the single target danger level of all target vehicles around the vehicle and the weight of the corresponding target vehicle, the multi-target danger level in the current scene is obtained by normalization processing;

[0011] S3: Obtain traffic flow attributes based on the original information of target vehicles around the vehicle obtained in S1, wherein the traffic flow attributes include traffic flow target type complexity, traffic flow density, traffic flow speed difference and traffic flow acceleration difference, and then obtain the traffic flow danger level;

[0012] S4: Obtaining road attributes according to the road information of the current road obtained in S1, wherein the road attributes include the current road type, lane addition or lane reduction, ramp divergence or ramp convergence, and then obtaining the road form hazard level;

[0013] S5: Based on the multi-objective hazard level, traffic flow hazard level, road form hazard level and corresponding weights, the environmental hazard level of longitudinal planning can be obtained by normalization processing.

[0014] Furthermore, the multiple sensors include but are not limited to cameras, millimeter wave radars, and lidars.

[0015] In the specific implementation, in S1, the numerical limits of the collision time are set to [20, 1], and then linearly converted to the interval [0, 1] to obtain DangerLevel_TTC.

[0016] Limit the deceleration to [0,10], and then convert it linearly to the interval [0,1] to get DangerLevel_TargetAccel.

[0017] The target vehicle is in the ramp state Target_Ramp_Sts: not in the ramp -0, in the diverging ramp -1, in the merging ramp -1.

[0018] The single target danger level DangerLevel_Target is:

[0019] DangerLevel_Target=tanh(DangerLevel_TTC+DangerLevel_TargetAccel+Target_Ramp_Sts*0.5)*Cutin_Potential

[0020] Among them: tanh is the normalization function;

[0021] Cutin_Potential is the cut-in probability of the target vehicle.

[0022] Furthermore, when calculating the multi-target danger level in S2, the two target vehicles located in front of the vehicle in the lane and the left and right lanes are considered, and the multi-target danger level DangerLevel_Targets is:

[0023] DangerLevel_Targets=tanh(W1*DL_RT1+W2*DL_RT2+W3*DL_RT3+W4*DL_RT4+W5*DL_RT5+W6*DL_RT6)

[0024] Where: DL_RTn is the single target danger level of target vehicle RTn.

[0025] W n is the weight of the target vehicle RTn.

[0026] Furthermore, the traffic flow target type complexity is the total number of target vehicle types in the current scene, and its value is limited to [0, 10] and then linearly converted to the interval [0, 1] to obtain DangerLevel_TFC.

[0027] The numerical limits of traffic flow density are limited to [0,5], and then linearly converted to the interval [0,1] to obtain DangerLevel_TFD.

[0028] The numerical limits of the traffic flow speed difference are limited to [0,50], and then linearly converted to the interval [0,1] to obtain DangerLevel_TFSD.

[0029] The numerical limits of the traffic flow acceleration difference are limited to [0,5], and then linearly converted to the interval [0,1] to obtain DangerLevel_TFAD.

[0030] Then the traffic flow danger level DangerLevel_TF is:

[0031] DangerLevel_TF

[0032] =tanh(DangerLevel_TFC+DangerLevel_TFD+DangerLevel_TFSD+DangerLevel_TFAD).

[0033] Furthermore, the traffic flow density is obtained according to the total number of target vehicles from 20m behind the vehicle to 100m in front of the vehicle. Specifically, when the total number of targets is 0-5, the corresponding traffic flow density is 1; when the total number of targets is 6-10, the corresponding traffic flow density is 2; when the total number of targets is 10-15, the corresponding traffic flow density is 3; when the total number of targets is 15-20, the corresponding traffic flow density is 4; when the total number of targets is 20-32, the corresponding traffic flow density is 5.

[0034] Furthermore, the traffic flow speed difference is calculated by the following method: firstly, the traffic flow speed of the left lane and the traffic flow speed of the right lane of the vehicle are calculated, and the traffic flow speeds of the left lane and the right lane of the vehicle are respectively the average values ​​of the speeds of the target vehicles located in the left lane and the right lane of the vehicle; then, the smaller one of the traffic flow speed of the left lane and the traffic flow speed of the right lane is selected and subtracted from the speed of the vehicle, so as to obtain the traffic flow speed difference.

[0035] Furthermore, the traffic flow acceleration difference is calculated by the following method: firstly, the traffic flow acceleration of the left lane and the traffic flow acceleration of the right lane of the vehicle are calculated, and the traffic flow acceleration of the left lane and the right lane of the vehicle are respectively the average values ​​of the accelerations of the target vehicles located in the left lane and the right lane of the vehicle; then, the smaller one of the traffic flow acceleration of the left lane and the traffic flow acceleration of the right lane is selected and subtracted from the acceleration of the vehicle, so as to obtain the traffic flow acceleration difference.

[0036] Furthermore, the road type hazard level coefficient DangerLevel_RoadType is determined according to the current road type. When the road type is a highway, its road type hazard level coefficient is 1; when the road type is a national highway / provincial highway, its road type hazard level coefficient is 1.3; when the road type is a city expressway, its road type hazard level coefficient is 1.5; when the road type is a city trunk road, its road type hazard level coefficient is 1.7; when the road type is a city road, its road type hazard level coefficient is 2.

[0037] When the vehicle is in the disappearance lane or adjacent to the disappearance lane, the distance between the vehicle and the lane increase or decrease point is limited to [150, 30], and then linearly converted to the interval [0, 1] to obtain DangerLevel_RoadTrans.

[0038] When the vehicle is in the merging lane or adjacent to the merging lane, the distance between the vehicle and the lane ramp merging point is limited to [150, 30] and then linearly converted to the interval [0, 1] to obtain DangerLevel_Ramp.

[0039] The road hazard level DangerLevel_Road is:

[0040] DangerLevel_Road=tanh((DangerLevel_RoadTrans+DangerLevel_Ramp)*DangerLevel_RoadType).

[0041] Furthermore, the environmental hazard level DangerLevel_Situation is:

[0042] DangerLevel_Situation=tanh(1.6*DangerLevel_Targets+0.8*DangerLevel_TF+0.6*DangerLevel_Road)*100%.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. The present invention comprehensively evaluates the danger level of the environment around the autonomous driving vehicle based on the driving intention, driving status and traffic flow status of the target vehicles around the vehicle, as well as the road form, which can effectively improve the environmental perception ability in complex traffic scenes and the accuracy of the environmental danger level.

[0045] 2. The present invention helps the longitudinal planning part of the autonomous driving vehicle to use different acceleration / deceleration planning methods in scenarios with different danger levels, making the behavior of the autonomous driving vehicle more reasonable and more in line with human expectations, thereby improving driving safety and riding comfort. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 -Scene graph.

[0047] Figure 2 - Flowchart of the present invention.

[0048] Figure 3 - The curve of the normalized function tanh. DETAILED DESCRIPTION

[0049] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0050] A method for evaluating the danger level of the surrounding environment of an autonomous driving vehicle comprises the following steps:

[0051] S1: The original information of target vehicles around the vehicle is obtained through multiple sensors, and then the target attributes of each target vehicle are obtained based on the original information, including the cut-in probability, collision time and deceleration required for the vehicle to avoid collision with the target vehicle. At the same time, the road information of the current road is obtained based on the high-precision map, and the location of the target vehicle is integrated with the road information to obtain the regional attributes of the target vehicle, and then the single target danger level is obtained;

[0052] S2: According to the single target danger level of all target vehicles around the vehicle and the weight of the corresponding target vehicle, the multi-target danger level in the current scene is obtained by normalization processing;

[0053] S3: Obtain traffic flow attributes based on the original information of target vehicles around the vehicle obtained in S1, wherein the traffic flow attributes include traffic flow target type complexity, traffic flow density, traffic flow speed difference and traffic flow acceleration difference, and then obtain the traffic flow danger level;

[0054] S4: Obtaining road attributes according to the road information of the current road obtained in S1, wherein the road attributes include the current road type, lane addition or lane reduction, ramp divergence or ramp convergence, and then obtaining the road form hazard level;

[0055] S5: Based on the multi-objective hazard level, traffic flow hazard level, road form hazard level and corresponding weights, the environmental hazard level of longitudinal planning can be obtained by normalization processing.

[0056] In a specific implementation, the multiple sensors include but are not limited to cameras, millimeter wave radars, and lidars.

[0057] In the specific implementation, in S1, the numerical limits of the collision time (s) are limited to [20, 1], and then linearly converted to the interval [0, 1] to obtain DangerLevel_TTC.

[0058] Deceleration (m / s 2 ) to [0,10], and then linearly converted to the interval [0,1] to obtain DangerLevel_TargetAccel.

[0059] The target vehicle is in the ramp state Target_Ramp_Sts: not in the ramp -0, in the diverging ramp -1, in the merging ramp -1.

[0060] The single target danger level DangerLevel_Target is:

[0061] DangerLevel_Target=tanh(DangerLevel_TTC+DangerLevel_TargetAccel+Target_Ramp_Sts*0.5)*Cutin_Potential

[0062] Wherein: tanh is a normalization function; the curve of the normalization function tanh used in this implementation is as follows Figure 3 Cutin_Potential is the cut-in probability of the target vehicle, and its value range is 0 to 100%.

[0063] During implementation, if the collision time for target vehicle A is 10s, the deceleration required for the vehicle to avoid collision with target vehicle A is 5, and the cut-in probability is 30%, the data listed here is only to illustrate the evaluation method of the single target danger level, then the obtained DangerLevel_TTC is equal to 0.53, and DangerLevel_TargetAccel is equal to 0.5. Then according to the state of the target vehicle being in the ramp, if the target vehicle is in the diversion ramp, then Target_Ramp_Sts is equal to -1, so the obtained single target danger level DangerLevel_Targe = tanh (0.53 + 0.5-1) * 30% = tanh0.03 * 30%, and then according to Figure 3 It is obtained that tanh0.03 is equal to 0.23, so the danger level of the target vehicle A is 0.069.

[0064] In the specific implementation, when calculating the multi-target danger level, consider the two target vehicles in front of the vehicle, which are located in the lane and the left and right lanes respectively. The target vehicle distribution is as follows: Figure 1 As shown, the multi-target danger levels DangerLevel_Targets are:

[0065] DangerLevel_Targets=tanh(W1*DL_RT1+W2*DL_RT2+W3*DL_RT3+W4*DL_RT4+W5*DL_RT5+W6*DL_RT6)

[0066] Where: DL_RTn is the single target danger level of target vehicle RTn;

[0067] W n is the weight of the target vehicle RTn, which is calibrated according to engineering experience. In specific implementation, the weights of the six target vehicles are shown in Table 1.

[0068] Table 1. Weights of the six target vehicles

[0069]

[0070] In specific implementation, the traffic flow target type complexity is the total number of target vehicle types in the current scene, and its value is limited to [0, 10] and then linearly converted to the interval [0, 1] to obtain DangerLevel_TFC.

[0071] The numerical limits of traffic flow density are limited to [0,5], and then linearly converted to the interval [0,1] to obtain DangerLevel_TFD.

[0072] The numerical limits of the traffic flow speed difference are limited to [0,50], and then linearly converted to the interval [0,1] to obtain DangerLevel_TFSD.

[0073] The numerical limits of the traffic flow acceleration difference are limited to [0,5], and then linearly converted to the interval [0,1] to obtain DangerLevel_TFAD.

[0074] Then the traffic flow danger level DangerLevel_TF is:

[0075] DangerLevel_TF

[0076] =tanh(DangerLevel_TFC+DangerLevel_TFD+DangerLevel_TFSD+DangerLevel_TFAD).

[0077] The traffic flow target type complexity is the total number of target vehicle types detected by multiple sensors in the current scene, and the target vehicle types include bicycles, electric vehicles, vans, motorcycles, trucks, trucks, trucks, tricycles, etc. The number of target vehicle types equals the number of traffic flow target type complexity.

[0078] In specific implementation, the traffic flow density is obtained according to the total number of target vehicles from 20m behind the vehicle to 100m in front of the vehicle. Specifically, when the total number of targets is 0-5, the corresponding traffic flow density is 1; when the total number of targets is 6-10, the corresponding traffic flow density is 2; when the total number of targets is 10-15, the corresponding traffic flow density is 3; when the total number of targets is 15-20, the corresponding traffic flow density is 4; when the total number of targets is 20-32, the corresponding traffic flow density is 5. As shown in Table 2.

[0079] Table 2. Traffic flow density table

[0080]

[0081] During specific implementation, the traffic flow speed difference is calculated by the following method: first calculate the traffic flow speed of the left lane and the traffic flow speed of the right lane of the vehicle (Km / h), where the traffic flow speeds of the left lane and the right lane of the vehicle are the average values ​​of the speeds of the target vehicles located in the left lane and the right lane of the vehicle, respectively; then select the smaller of the traffic flow speed of the left lane and the traffic flow speed of the right lane and subtract it from the speed of the vehicle to obtain the traffic flow speed difference.

[0082] In specific implementation, the traffic flow acceleration difference is calculated as follows: First, calculate the traffic flow acceleration of the left lane and the traffic flow acceleration of the right lane of the vehicle (m / s 2 ), the traffic flow accelerations of the left lane and the right lane of the vehicle are respectively the average values ​​of the accelerations of the target vehicles located in the left lane and the right lane of the vehicle; then the smaller one of the left lane traffic flow acceleration and the right lane traffic flow acceleration is selected and subtracted from the vehicle acceleration to obtain the traffic flow acceleration difference.

[0083] In specific implementation, the road type hazard level coefficient DangerLevel_RoadType is determined according to the current road type. When the road type is a highway, its road type hazard level coefficient is 1; when the road type is a national highway / provincial highway, its road type hazard level coefficient is 1.3; when the road type is an urban expressway, its road type hazard level coefficient is 1.5; when the road type is a city trunk road, its road type hazard level coefficient is 1.7; when the road type is an urban road, its road type hazard level coefficient is 2. The detailed table of road type hazard level coefficients is shown in Table 3.

[0084] Table 3. Detailed list of road type hazard level coefficients

[0085]

[0086] When the vehicle is in the disappearance lane or adjacent to the disappearance lane, the distance between the vehicle and the lane increase or decrease point is limited to [150, 30], and then linearly converted to the interval [0, 1] to obtain DangerLevel_RoadTrans.

[0087] When the vehicle is in the merging lane or adjacent to the merging lane, the distance between the vehicle and the lane ramp merging point is limited to [150, 30] and then linearly converted to the interval [0, 1] to obtain DangerLevel_Ramp.

[0088] The road hazard level DangerLevel_Road is:

[0089] DangerLevel_Road=tanh((DangerLevel_RoadTrans+DangerLevel_Ramp)*DangerLevel_RoadType).

[0090] In specific implementation, the environmental hazard level DangerLevel_Situation is:

[0091] DangerLevel_Situation=tanh(1.6*DangerLevel_Targets+0.8*DangerLevel_TF+0.6*DangerLevel_Road)*100%.

[0092] During implementation, when the normalized function tanh is involved in the calculation process, the curve of the normalized function tanh is used to read the value, and then the corresponding danger level is calculated. The single-target danger level is calculated according to the original information of the target vehicles around the vehicle and the road information, and then the multi-target danger level is calculated through the single-target danger level; the traffic flow danger level is calculated according to the original information of the target vehicles around the vehicle; the road shape danger level is calculated according to the road information; finally, the environmental danger level is obtained based on the multi-target danger level, the traffic flow danger level and the road shape danger level, and the calculation process of the multi-target danger level, the traffic flow danger level and the road shape danger level is carried out simultaneously.

[0093] Finally, it should be noted that the above embodiments of the present invention are only examples for illustrating the present invention, and are not intended to limit the implementation methods of the present invention. For those skilled in the art, other different forms of changes and modifications can be made based on the above description. It is impossible to list all implementation methods here. Any obvious changes or modifications derived from the technical solution of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for assessing the danger level of the surrounding environment of an autonomous driving vehicle, characterized in that: The following steps are involved: S1: The original information of target vehicles around the vehicle is obtained through multiple sensors, and then the target attributes of each target vehicle are obtained based on the original information, including the cut-in probability, collision time and deceleration required for the vehicle to avoid collision with the target vehicle. At the same time, the road information of the current road is obtained based on the high-precision map, and the location of the target vehicle is integrated with the road information to obtain the regional attributes of the target vehicle, and then the single target danger level is obtained; S2: According to the single target danger level of all target vehicles around the vehicle and the weight of the corresponding target vehicle, the multi-target danger level in the current scene is obtained by normalization processing; S3: Obtain traffic flow attributes based on the original information of target vehicles around the vehicle obtained in S1, wherein the traffic flow attributes include traffic flow target type complexity, traffic flow density, traffic flow speed difference and traffic flow acceleration difference, and then obtain the traffic flow danger level; S4: Obtaining road attributes according to the road information of the current road obtained in S1, wherein the road attributes include the current road type, lane addition or lane reduction, ramp divergence or ramp convergence, and then obtaining the road form hazard level; S5: Based on the multi-objective hazard level, traffic flow hazard level, road form hazard level and corresponding weights, the environmental hazard level of longitudinal planning can be obtained by normalization processing; In S1, the collision time value is limited to [20, 1], and then linearly converted to the interval [0, 1] to obtain DangerLevel_TTC; Set the deceleration limits to [0,10], and then convert them linearly to the interval [0,1] to get DangerLevel_TargetAccel; The target vehicle is in the ramp state Target_Ramp_Sts: not in the ramp - 0, in the diverging ramp - -1, in the merging ramp - 1; The single target danger level DangerLevel_Target is: DangerLevel_Target=tanh(DangerLevel_TTC+DangerLevel_TargetAccel+Target_Ramp_Sts*0.5)*Cutin_Potential Among them: tanh is the normalization function; Cutin_Potential is the cut-in probability of the target vehicle; When calculating the multi-target danger level in S2, the two target vehicles in front of the vehicle, which are located in the lane and the left and right lanes, are considered. The multi-target danger level DangerLevel_Targets is: DangerLevel_Targets=tanh(W1*DL_RT1+ W2*DL_RT2+ W3*DL_RT3+ W4*DL_RT4+ W5*DL_RT5+ W6*DL_RT6) Where: DL_RTn is the single target danger level of target vehicle RTn; W n is the weight of the target vehicle RTn; The traffic flow target type complexity is the total number of target vehicle types in the current scene, and its value is limited to [0, 10], and then linearly converted to the interval [0, 1] to obtain DangerLevel_TFC; Limit the value of traffic density to [0,5], and then linearly convert it to the interval [0,1] to obtain DangerLevel_TFD; The numerical limits of the traffic flow speed difference are limited to [0,50], and then linearly converted to the interval [0,1] to obtain DangerLevel_TFSD; Limit the value of traffic flow acceleration difference to [0,5], and then linearly convert it to the interval [0,1] to obtain DangerLevel_TFAD; Then the traffic flow danger level DangerLevel_TF is: DangerLevel_TF =tanh(DangerLevel_TFC+DangerLevel_TFD+DangerLevel_TFSD+DangerLevel_TFAD); The road type hazard level coefficient DangerLevel_RoadType is determined according to the current road type. When the road type is a highway, its road type hazard level coefficient is 1; when the road type is a national highway / provincial highway, its road type hazard level coefficient is 1.3; when the road type is an urban expressway, its road type hazard level coefficient is 1.5; when the road type is a city trunk road, its road type hazard level coefficient is 1.7; when the road type is an urban road, its road type hazard level coefficient is 2; When the vehicle is in the vanishing lane or adjacent to the vanishing lane, the distance between the vehicle and the lane increase or decrease point is limited to [150, 30], and then linearly converted to the interval [0, 1] to obtain DangerLevel_RoadTrans; When the vehicle is in the merging lane or adjacent to the merging lane, the distance between the vehicle and the lane ramp merging point is limited to [150, 30], and then linearly converted to the interval [0, 1] to obtain DangerLevel_Ramp; The road hazard level DangerLevel_Road is: DangerLevel_Road=tanh((DangerLevel_RoadTrans + DangerLevel_Ramp)*DangerLevel_RoadType); Environmental hazard level DangerLevel_Situation is: DangerLevel_Situation=tanh(1.6*DangerLevel_Targets + 0.8*DangerLevel_TF +0.6*DangerLevel_Road)*100%.

2. The method for evaluating the danger level of the surrounding environment of an autonomous driving vehicle according to claim 1, characterized in that: The multiple sensors include cameras, millimeter wave radars, and lidars.

3. The method for evaluating the danger level of the surrounding environment of an autonomous driving vehicle according to claim 1, characterized in that: The traffic flow density is obtained according to the total number of target vehicles from 20 m behind the vehicle to 100 m in front of the vehicle. Specifically, when the total number of targets is 0-5, the corresponding traffic flow density is 1; when the total number of targets is 6-10, the corresponding traffic flow density is 2; when the total number of targets is 10-15, the corresponding traffic flow density is 3; when the total number of targets is 15-20, the corresponding traffic flow density is 4; when the total number of targets is 20-32, the corresponding traffic flow density is 5.

4. The method for evaluating the danger level of the surrounding environment of an autonomous driving vehicle according to claim 1, characterized in that: The traffic flow speed difference is calculated by the following method: first calculate the traffic flow speed of the left lane and the traffic flow speed of the right lane of the vehicle, wherein the traffic flow speeds of the left lane and the right lane of the vehicle are the average values ​​of the speeds of the target vehicles located in the left lane and the right lane of the vehicle, respectively; then select the smaller one of the traffic flow speed of the left lane and the traffic flow speed of the right lane and subtract it from the speed of the vehicle, thereby obtaining the traffic flow speed difference.

5. The method for evaluating the danger level of the surrounding environment of an autonomous driving vehicle according to claim 1, characterized in that: The traffic flow acceleration difference is calculated as follows: first calculate the traffic flow acceleration of the left lane and the traffic flow acceleration of the right lane of the vehicle, where the traffic flow accelerations of the left lane and the right lane of the vehicle are the average values ​​of the accelerations of the target vehicles located in the left lane and the right lane of the vehicle, respectively; then select the smaller of the traffic flow acceleration of the left lane and the traffic flow acceleration of the right lane and subtract it from the acceleration of the vehicle, thereby obtaining the traffic flow acceleration difference.

Images