Lane-changing Risk Assessment Method for Autonomous Vehicles Considering the Heterogeneity of Surrounding Vehicles

Abstract

The present invention is applicable to the field of autonomous driving safety technology, and provides a method for evaluating the lane-changing risk of autonomous driving vehicles considering the heterogeneity of surrounding vehicles, including: the autonomous driving vehicle uses a perception system to collect real-time information about itself and the surrounding environment, and introduces the heterogeneity of vehicles by considering the reaction time and the maximum braking deceleration to calculate the threat level of each surrounding vehicle; divides the road space into grids, constructs risk energy fields for the surrounding vehicles and lane lines respectively, and obtains the comprehensive risk energy field of the road area after superposition; the autonomous driving vehicle calculates the risk force it receives in the comprehensive risk energy field of the road area in combination with its own motion state, and compares it with the set risk level threshold to evaluate the risk during the lane-changing process. The present invention solves the problems that the existing risk assessment methods cannot cope with complex road environments and do not comprehensively consider the characteristics and motion uncertainties of surrounding vehicles, and realizes dynamic early warning of lane-changing risks under the conditions of considering the characteristics of surrounding vehicles and autonomous driving vehicles.

Description

Technical Field

[0001] The present invention belongs to the technical field of autonomous driving safety, and particularly relates to a method for evaluating the lane-changing risk of autonomous vehicles considering the heterogeneity of surrounding vehicles. Background Art

[0002] The core system of autonomous vehicles can be divided into modules such as environmental perception, risk analysis, decision-making, control, and execution. Among them, the risk analysis module plays a bridging role. It evaluates risks based on the results of environmental perception and the motion prediction of dynamic objects, and uses the evaluation results as the input of the decision-making module. There are mainly two driving behaviors of autonomous vehicles during driving: following and lane-changing. Compared with following behavior, lane-changing behavior is more complex and dangerous. The evaluation and prediction of its risk are crucial for ensuring the driving safety of autonomous vehicles.

[0003] Existing risk assessment methods are mainly divided into methods based on time and kinematic indicators, statistical methods, potential field methods, and abnormal behavior detection methods, etc. The method based on time indicators evaluates driving risks from a time perspective, including time headway (THW), time to collision (TTC), and reaction time (TTR), etc. The method based on kinematic indicators uses kinematic parameters to evaluate driving risks, such as the minimum distance and the acceleration required for collision avoidance. Such methods perform well in longitudinal collision scenarios but have limitations in dealing with lateral collision risks. Statistical methods use machine learning based on data to achieve the quantification and evaluation of risks. Although they can model uncertainties, they are computationally complex and have low real-time performance. At the same time, the model training process requires a large amount of high-quality data. The potential field method uses field theory to integrate the risks generated by different factors in the traffic environment on the host vehicle, but usually ignores the influence of the vehicle's own characteristics and motion uncertainties, and its adaptability in complex scenarios such as vehicle lane-changing still needs to be improved; the method based on abnormal behavior detection can evaluate driving risks related to non-collision, such as unexpected behaviors such as driver fatigue and running red lights, and quantify risks based on the degree of deviation from normal behavior, but the difficulty of its detection and risk quantification is relatively large.

[0004] Based on the above analysis, although existing risk assessment methods can quantify risks to a certain extent, they still have deficiencies in dealing with complex road environments and considering the heterogeneity and motion uncertainties of surrounding vehicles, and further research and improvement are urgently needed. Therefore, a method for evaluating the lane-changing risk of autonomous vehicles considering the heterogeneity of surrounding vehicles is proposed. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for evaluating the lane-changing risk of autonomous vehicles considering the heterogeneity of surrounding vehicles, aiming to solve the problems raised in the above background art.

[0006] To achieve the above purpose, the present invention provides the following technical solutions:

[0007] Method for evaluating lane-changing risk of autonomous vehicle considering heterogeneity of surrounding vehicles, comprising the following steps:

[0008] Step S1: Evaluate threat degree of surrounding vehicles considering vehicle heterogeneity: The autonomous vehicle collects information of itself and the surrounding environment in real time through a perception system, and calculates the threat degree of each surrounding vehicle by introducing vehicle heterogeneity considering reaction time and maximum braking deceleration;

[0009] Step S2: Construct comprehensive risk energy field of road area based on threat degree of surrounding vehicles: Divide the road space into grids, construct risk energy fields for surrounding vehicles and lane lines respectively, and obtain the comprehensive risk energy field of the road area after superposition;

[0010] Step S3: Evaluate lane-changing risk of autonomous vehicle using comprehensive risk energy field of road area: The autonomous vehicle calculates the risk force received in the comprehensive risk energy field of the road area in combination with its own motion state, and compares the received risk force with a set risk level threshold to evaluate the risk during the lane-changing process.

[0011] Further, the specific process of the step of evaluating threat degree of surrounding vehicles considering vehicle heterogeneity is as follows:

[0012] Step S11: The autonomous vehicle intending to change lanes or in the process of changing lanes uses a perception system to obtain information of the environment and surrounding vehicles, and obtains the trajectories and motion states of surrounding vehicles based on a prediction algorithm; specifically including: obtaining the length l box and width w box of surrounding vehicles; obtaining the first type of surrounding vehicles to determine the reaction time, and obtaining the second type of surrounding vehicles to determine the maximum braking deceleration; obtaining the intention probability predicted at time T + t of surrounding vehicles, and the motion state information corresponding to the intention probability p k , including T + t the trajectory point coordinates at time( x k , y k ), speed v k and steering angle ; obtaining the road surface material to determine the road surface adhesion coefficient; obtaining the position and type of lane lines; obtaining the position coordinates of the autonomous vehicle itself( x T , y T ), speed vT and the front-wheel steering angle ;

[0013] The first type includes CAV (Connected and Autonomous Vehicle), AV (Autonomous Vehicle) and HDV (Human-driven Vehicle); the second type includes light vehicles and heavy vehicles;

[0014] Step S12, calculate the threat level of each surrounding vehicle :

[0015] Equation 1: ;

[0016] Equation 2: ;

[0017] In the formula: represents the threat level of the surrounding vehicle i ; represents the reaction time of the surrounding vehicle i , and the reaction times of CAV, AV and HD vehicles are preset values, ; represents the maximum braking deceleration of the surrounding vehicle i , and the maximum braking decelerations of light and heavy vehicles are preset values, ; k t represents the reaction time adjustment coefficient of the surrounding vehicle; c represents the road adhesion coefficient; X s represents a 0-1 variable used to determine whether the vehicle is speeding. When speeding X s = 1; k t0 , k ts both represent undetermined coefficients.

[0018] Furthermore, the specific process of constructing the risk energy field of the surrounding vehicles based on the threat level of the surrounding vehicles is as follows:

[0019] Step 21, divide the road space into meshes in the road area D. The formula is as follows:

[0020] Equation 3: ;

[0021] Equation 4: ;

[0022] Equation 5: ;

[0023] Equation 6: ;

[0024] In the formula: L D represents the length of the road area D along the lane line direction; s var represents the longitudinal visual range of the vehicle; W D represents the width of the road area perpendicular to the lane line direction; n represents the number of lanes included in the road area D; l lane represents the lane width; a 0 represents the length of the road grid cell; b 0 represents the width of the road grid cell; f represents that the road area along the lane line direction is divided into the number of grid cells with a length of a 0 ; represents that the road area along the direction perpendicular to the lane line is divided into the number of grid cells with a width of b 0 ;

[0025] Step 22, calculate the risk radiation energy of surrounding vehicles to other grids, and construct the risk energy field of surrounding vehicles. The formula is as follows:

[0026] ;

[0027] In the formula: represents the longitudinal risk correction factor; represents the lateral risk correction factor; are all undetermined coefficients, respectively representing the influence degrees of distance on longitudinal and lateral risks; are all undetermined coefficients, respectively representing the influence degrees of speed on longitudinal and lateral risks; represents the length of the surrounding vehicle; represents the width of the surrounding vehicle; represents the grid coordinates of the predicted trajectory point of the surrounding vehicle at the T+1 moment under the kth type of behavior intention; represents the original coordinates of the grid point; represents the corresponding deflection value after the original coordinates of the grid point are deflected considering the vehicle steering angle; represents the speed of the surrounding vehicle under the kth type of behavior intention; represents the steering angle when the vehicle turns; represents the included angle between the vehicle speed direction and the positive direction of the x-axis; represents the surrounding vehicle i the risk energy value generated by the grid where the predicted trajectory point is located under the kth type of behavior intention to other grids; Indicates the threat level of surrounding vehicles i ; Indicates the field strength of the risk energy field generated by surrounding vehicles i ; Indicates the probabilities of different driving intentions of surrounding vehicles i , k = {1 (left turn), 2 (going straight), 3 (right turn)}.

[0028] Furthermore, a lane - line risk energy field is constructed based on the threat level of the surrounding vehicles, specifically as follows:

[0029] Equation 12: ;

[0030] In the formula: Indicates the field strength of the risk energy field generated by the lane - line j at other grids; Indicates different risk radiation coefficients generated by different lane - line types; Indicates a coefficient to be determined; Indicates the y - axis coordinate of other grids; Indicates the y - axis coordinate of the grid where the lane - line is located; l lane Indicates the lane width.

[0031] Furthermore, a comprehensive risk energy field of the road area at time t is established, specifically as follows:

[0032] Equation 13: ;

[0033] In the formula: Indicates the field strength of the risk energy field generated by surrounding vehicles i at other grids; Indicates the field strength of the risk energy field generated by the lane - line j at other grids; m represents the number of surrounding vehicles in the road area; s represents the number of lane - lines in the road area; Indicates the field strength of the comprehensive risk energy field generated by surrounding vehicles and lane - lines.

[0034] Furthermore, calculate the risk force received by the autonomous vehicle at time t from surrounding vehicles, specifically as follows:

[0035] Equation 14: ;

[0036] In the formula: Indicates the field strength of the risk energy field generated by surrounding vehicles in the same lane as the autonomous vehicle at the location of the autonomous vehicle; Surrounding vehicles in different lanes from the autonomous vehicle The field strength of the risk energy field generated at the location of the autonomous vehicle; Represents the number of surrounding vehicles in the same lane as the autonomous vehicle; Represents the number of surrounding vehicles in different lanes from the autonomous vehicle; Represents the field strength of the risk energy field generated at the location of the autonomous vehicle by the surrounding vehicles in the same lane as the autonomous vehicle; Represents the field strength of the risk energy field generated at the location of the autonomous vehicle by the surrounding vehicles in different lanes from the autonomous vehicle; Represents the speed of the autonomous vehicle; Are all undetermined coefficients, respectively representing the sensitivity of the longitudinal motion state of the autonomous vehicle to the risks of surrounding vehicles in the same lane and the sensitivity of the lateral motion state of the autonomous vehicle to the risks of surrounding vehicles in different lanes; Represents the steering angle of the autonomous vehicle.

[0037] Furthermore, calculate the risk force on the autonomous vehicle at a certain moment from the lane lines, specifically as follows:

[0038] Equation 15: ;

[0039] In the formula: Represents the lane line j The field strength of the risk energy field generated at the location of the autonomous vehicle; Represents the number of lane lines in the road area; Is an undetermined coefficient, representing the sensitivity of the lateral motion state of the autonomous vehicle to the risks of lane lines; Represents the speed of the autonomous vehicle; Represents the steering angle of the autonomous vehicle.

[0040] Furthermore, judge the risk level during the lane change of the autonomous vehicle based on the risk force Risk Level , specifically as follows:

[0041] Equation 16: ;

[0042] Equation 17: ;

[0043] In the formula: Represents the risk force on the autonomous vehicle during the lane change in the risk energy field; Are all preset parameters, respectively representing the minimum and maximum risk thresholds.

[0044] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0045] 1. Considering vehicle heterogeneity: The present invention analyzes the physical attributes and motion characteristics of different types of vehicles, and at the same time takes into account the uncertainty caused by changes in the behavior of surrounding vehicles. The risk energy fields generated by vehicles with different characteristics and intention probabilities are heterogeneous. Through this refined analysis, the present invention can more accurately evaluate the potential threats posed by surrounding vehicles to autonomous vehicles, thus significantly improving the safety of autonomous vehicles during lane-changing.

[0046] 2. Construction of dynamic risk energy field: The present invention constructs a risk energy field in real time based on the motion prediction results of surrounding vehicles to predict future risk changes, providing a more comprehensive risk assessment for autonomous vehicles.

[0047] 3. Easy to implement and expand: The implementation of the present invention does not rely on a large amount of training data or complex machine learning models. Its rule-based method makes the implementation relatively simple and can be adjusted and expanded according to different road conditions or traffic rules.

[0048] In summary, based on the concept of risk energy field, the present invention proposes a method for evaluating the lane-changing risk of autonomous vehicles considering the heterogeneity of surrounding vehicles. This method quantifies the impact of surrounding vehicles with different characteristics on the host vehicle, provides a theoretical basis for accurately evaluating the risk during autonomous lane-changing, and helps to solve the problem that the existing risk assessment technology insufficiently considers the heterogeneity of surrounding vehicles and motion uncertainty in the lane-changing condition of autonomous vehicles, thus making it difficult to ensure lane-changing safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 is a flowchart of the method of the present invention.

[0050] Figure 2 is a schematic diagram of the lane-changing scenario of an autonomous vehicle.

[0051] Figure 3 is a schematic diagram of the risk energy field of a heavy HDV (FV).

[0052] Figure 4 is a schematic diagram of the risk energy field of a light CAV (RV).

[0053] Figure 5 is a schematic diagram of the risk energy field generated by lane lines.

[0054] Figure 6 is a schematic diagram of the change in the comprehensive risk energy field of the road area during the lane-changing process of an autonomous vehicle; where (a) is the schematic diagram of the comprehensive risk field at the initial moment of lane-changing, (b) is the schematic diagram of the comprehensive risk field during the lane-changing process, and (c) is the schematic diagram of the comprehensive risk field at the end of lane-changing.

[0055] Figure 7 Schematic diagram of the change in the risk force level during the lane-changing process of an autonomous vehicle. Specific implementation mode

[0056] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0057] The following describes in detail the specific implementation of the present invention with reference to specific embodiments.

[0058] As Figure 1 shown, a lane-changing risk assessment method for an autonomous vehicle considering the heterogeneity of surrounding vehicles provided by an embodiment of the present invention includes the following steps:

[0059] Step S1: Evaluate the threat level of surrounding vehicles considering vehicle heterogeneity: The autonomous vehicle collects information on itself and the surrounding environment in real time through a perception system, and calculates the threat level of each surrounding vehicle by introducing the heterogeneity of the vehicle by considering the reaction time and the maximum braking deceleration. Specifically as follows:

[0060] (1) An autonomous vehicle intending to change lanes or in the process of changing lanes uses a perception system to obtain information on the environment and surrounding vehicles, and obtains the trajectories and motion states of the surrounding vehicles based on a prediction algorithm; specifically including: obtaining the length l box and width w box of the surrounding vehicle; obtaining the first type (CAV / AV / HDV) of the surrounding vehicle to determine the reaction time, and obtaining the second type (light vehicle / heavy vehicle) of the surrounding vehicle to determine the maximum braking deceleration; obtaining the T + t intention probability predicted at time , and the motion state information corresponding to the intention probability p k (including the trajectory point coordinates at time T + t ( x k , y k ), speed v k and steering angle ); obtaining the road surface material to determine the road surface adhesion coefficient; obtaining the position and type of the lane lines; obtaining the position coordinates of the autonomous vehicle itself ( x T , y T ), speedv T and the front wheel steering angle ;

[0061] (2)Calculate the threat level of each surrounding vehicle ;

[0062] Equation 1: ;

[0063] Equation 2: ;

[0064] Wherein: represents the threat level of the surrounding vehicle i ; represents the reaction time of the surrounding vehicle i , and the reaction times of CAV, AV, and HD vehicles are preset values, ; represents the maximum braking deceleration of the surrounding vehicle i , and the maximum braking decelerations of light and heavy vehicles are preset values, ; k t represents the reaction time adjustment coefficient of the surrounding vehicle; c represents the road surface adhesion coefficient; X s represents a 0-1 variable used to determine whether the vehicle is speeding. When speeding X s = 1; k t0 , k ts both represent undetermined coefficients.

[0065] Step S2. Construct a comprehensive risk energy field for the road area based on the threat levels of surrounding vehicles: Divide the road space into grids, construct risk energy fields for surrounding vehicles and lane lines respectively, and obtain the comprehensive risk energy field for the road area after superposition. Specifically as follows:

[0066] (1)Divide the road space into grids in a certain rule within the road area D, and the formula is as follows:

[0067] Equation 3: ;

[0068] Equation 4: ;

[0069] Equation 5: ;

[0070] Equation 6: ;

[0071] In the formula: L D represents the length of the road area D along the lane line direction; s var represents the longitudinal visual range of the vehicle; W D represents the width of the road area perpendicular to the lane line direction; n represents the number of lanes included in the road area D, taking n = 3, and defining n 1 as the target lane, n 2 as the current lane, n 3 as the adjacent lane on the other side; l lane represents the lane width; a 0 represents the length of the road grid cell; b 0 represents the width of the road grid cell; f represents that the road area along the lane line direction is divided into the number of grid cells with a length of a 0 ; represents that the road area along the direction perpendicular to the lane line is divided into the number of grid cells with a width of b 0 ;

[0072] (2) Calculate the risk radiation energy of surrounding vehicles to other grids and construct the risk energy field of surrounding vehicles;

[0073] Calculate the longitudinal and lateral risk correction factors:

[0074] ;

[0075] Calculate the surrounding vehicle i at the th type of behavior intention of the risk radiation energy:

[0076] Formula 10: ;

[0077] Calculate the surrounding vehicle i all behavior intentions of the risk energy field:

[0078] Formula 11: ;

[0079] In Formulas 7 - 11: represents the longitudinal risk correction factor; represents the lateral risk correction factor; are all undetermined coefficients, respectively representing the influence degrees of distance on longitudinal and lateral risks; are all undetermined coefficients, respectively representing the influence degrees of speed on longitudinal and lateral risks; Represents the length of the surrounding vehicle; Represents the width of the surrounding vehicle; Represents the grid coordinates of the predicted trajectory point at the (T + 1)th moment under the kth type of behavior intention of the surrounding vehicle; Represents the original coordinates of the grid point; Represents the corresponding deflection value after the original coordinates of the grid point are deflected considering the steering angle of the vehicle; Represents the speed of the surrounding vehicle under the kth type of behavior intention; Represents the steering angle when the vehicle turns; (the counterclockwise direction of the steering angle is specified as the positive direction); Represents the angle between the vehicle speed direction and the positive direction of the x-axis; Represents the surrounding vehicle i The risk energy value generated by the grid where the predicted trajectory point of the surrounding vehicle is located under the kth type of behavior intention to other grids; Represents the surrounding vehicle i Threat level; Represents the surrounding vehicle i The field strength of the risk energy field generated; Represents the probabilities of different driving intentions of the surrounding vehicle, k = {1 (left turn), 2 (going straight), 3 (right turn)}.

[0080] (3)Construct a lane line risk energy field, and use a model in the form of a quasi-Gaussian to describe the risk energy of the lane line. The risk energy field generated by a lane line of type at other grids is as follows:

[0081] Equation 12: ;

[0082] In the formula: Represents the field strength of the risk energy field generated by the lane line j at other grids; Represents different risk radiation coefficients generated by different lane line types. Let the white dotted line A 1 = 1, and the white solid line A 2 = 2; Represents a coefficient to be determined; Represents the y-axis coordinate of other grids; Represents the y-axis coordinate of the grid where the lane line is located; l lane Represents the lane width.

[0083] (4)Establish a comprehensive risk energy field for the road area. Taking each grid as the basic research unit, the risk distribution of the road area within the vehicle's visible range is characterized by studying the concept of the risk energy generated by different risk factors for each grid, so as to construct the comprehensive risk energy field of the road area at time t, reflecting the spatial distribution of the comprehensive risk in the traffic environment. The calculation method of the comprehensive risk energy field of the road area is as follows:

[0084] Equation 13: ;

[0085] In the formula: represents the risk energy field of surrounding vehicles; represents the lane line j The field strength of the risk energy field generated at other grids; m represents the number of surrounding vehicles in the road area; s represents the number of lane lines in the road area; represents the field strength of the comprehensive risk energy field generated by surrounding vehicles and lane lines.

[0086] Step S3. Evaluate the lane-changing risk of the autonomous vehicle using the comprehensive risk energy field of the road area: The autonomous vehicle calculates the risk force it receives in the comprehensive risk energy field of the road area in combination with its own motion state, and compares the received risk force with the set risk level threshold to evaluate the risk during the lane-changing process. Specifically as follows:

[0087] (1)Calculate the risk force received by the autonomous vehicle at time from surrounding vehicles, which characterizes the interaction risk between the autonomous vehicle and surrounding vehicles. Specifically as follows:

[0088] Equation 14: ;

[0089] In the formula: represents the surrounding vehicles in the same lane as the autonomous vehicle The field strength of the risk energy field generated at the position of the autonomous vehicle; represents the surrounding vehicles in a different lane from the autonomous vehicle The field strength of the risk energy field generated at the position of the autonomous vehicle; represents the number of surrounding vehicles in the same lane as the autonomous vehicle; represents the number of surrounding vehicles in a different lane from the autonomous vehicle; represents the field strength of the risk energy field generated by the surrounding vehicles in the same lane as the autonomous vehicle at the position of the autonomous vehicle; represents the field strength of the risk energy field generated by the surrounding vehicles in a different lane from the autonomous vehicle at the position of the autonomous vehicle; represents the speed of the autonomous vehicle; are undetermined coefficients, representing the sensitivity of the longitudinal motion state of the autonomous vehicle to the risks of surrounding vehicles in the same lane and the sensitivity of the lateral motion state of the autonomous vehicle to the risks of surrounding vehicles in different lanes, respectively; represents the steering angle of the autonomous vehicle.

[0090] (2) Calculate the risk force from the lane lines on the autonomous vehicle at a specific moment, as follows:

[0091] Equation 15: ;

[0092] In the formula: represents the field strength of the risk energy field generated by the lane line j at the location of the autonomous vehicle; represents the number of lane lines in the road area; is an undetermined coefficient, representing the sensitivity of the lateral motion state of the autonomous vehicle to the risks of lane lines; represents the speed of the autonomous vehicle; represents the steering angle of the autonomous vehicle.

[0093] (3) Judge the risk level during the lane-changing process of the autonomous vehicle based on the risk force Risk Level , as follows:

[0094] Equation 16: ;

[0095] Equation 17: ;

[0096] In the formula: represents the risk force suffered by the autonomous vehicle in the risk energy field during the lane-changing process; are all preset parameters, representing the minimum and maximum risk thresholds respectively.

[0097] Example 1. Apply the method of the present invention to the following scenario:

[0098] On a three-lane asphalt road (the middle lane line is a dotted line, allowing lane changes), on a sunny day, an autonomous vehicle (Ego vehicle) intends to change from the middle lane 2 to the right lane 3 to avoid a vehicle in front that is performing a deceleration operation and to select a safer driving route. The vehicle in front (FV) in the current lane of the ego vehicle is a heavy-duty HDV that is decelerating, and the vehicle behind (RV) in the right target lane is a light-duty CAV with a relatively fast driving speed. The scenario is shown as Figure 2 and the vehicle information is shown in Table 1:

[0099] Table 1 Vehicle Information around the Autonomous Vehicle

[0100]

[0101] (1) The autonomous vehicle obtains information about the environment and surrounding vehicles based on the perception system, and predicts the future intention probabilities and corresponding positions at time T+t of the truck ahead and the car in the target lane through the prediction algorithm. At a certain moment, the intention probability of the truck is [0.6, 0.3, 0.1], and the intention probability of the car is [0, 1, 0]. Calculate the threat levels for surrounding vehicle 1 and surrounding vehicle 2 respectively. The reaction times and maximum braking decelerations of different vehicle types are shown in Tables 3 and 4, and the adjustment coefficient is calibrated through simulation experiments. The adhesion coefficient is obtained according to the road surface material in Table 2.

[0102] Table 2 Adhesion Coefficients of Different Road Surface Materials

[0103]

[0104] Table 3 Reaction Times of Different First-Type Vehicles

[0105]

[0106] Table 4 Maximum Braking Decelerations of Different Second-Type Vehicles

[0107]

[0108] (2) Divide the road space into grids with a granularity of , calculate the risk radiation energy of surrounding vehicle 1 and surrounding vehicle 2 to other grids to construct the vehicle risk energy field at time T+1, as shown in Figure 3 and Figure 4 respectively; calculate the risk radiation energy of the lane lines to other grids to construct the lane line risk energy field, as shown in Figure 5 ; then establish the comprehensive risk energy field of the road area.

[0109]

[0109] (3) Calculate the risk forces exerted on the autonomous vehicle from the surrounding vehicles and lane lines at each moment during the lane change duration t of the autonomous vehicle, characterize the risk of the autonomous vehicle lane change process and its interaction with them, and obtain the corresponding comprehensive risk level. The comprehensive risk energy field of the road area at different moments during the lane change process is shown in Figure 6 , and the change in the risk force level during the autonomous vehicle lane change process is shown in Figure 7 .

[0110] The above is only the preferred implementation mode of the present invention. It should be noted that for those skilled in the art, without departing from the concept of the present invention, several deformations and improvements can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the implementation effect of the present invention and the practicability of the patent.

Claims

1. A lane-changing risk assessment method for an autonomous driving vehicle considering the heterogeneity of surrounding vehicles, characterized in that: The following steps are involved: Step S1, considering vehicle heterogeneity to evaluate the threat level of surrounding vehicles: the autonomous driving vehicle collects information about itself and its surrounding environment in real time through a perception system, and calculates the threat level of each surrounding vehicle by introducing vehicle heterogeneity by considering reaction time and maximum braking deceleration; Step S2, constructing a comprehensive risk energy field of the road area based on the threat degree of surrounding vehicles: dividing the road space into grids, constructing risk energy fields for surrounding vehicles and lane lines respectively, and obtaining a comprehensive risk energy field of the road area after superposition; Step S3, using the comprehensive risk energy field of the road area to evaluate the lane changing risk of the autonomous driving vehicle: the autonomous driving vehicle calculates the risk force it is subjected to in the comprehensive risk energy field of the road area based on its own motion state, and compares the risk force it is subjected to with the set risk level threshold to evaluate the risk in the lane changing process; The specific process of the step of considering vehicle heterogeneity to assess the threat level of surrounding vehicles is as follows: Step S11: The autonomous driving vehicle intending to change lanes or in the process of changing lanes uses the perception system to obtain information about the environment and surrounding vehicles, and predict the trajectory and motion state of surrounding vehicles; specifically, it includes: obtaining the length of surrounding vehicles l box ,width w box ; Get the first type of surrounding vehicles to determine the reaction time, and get the second type of surrounding vehicles to determine the maximum braking deceleration; Get the surrounding vehicles T + t Moment-by-moment predicted probability of intent , and the probability of intention p k The corresponding motion status information includes T + t The coordinates of the trajectory point at time ( x k , y k ),speed v k and steering angle ; Obtain the road surface material to determine the road adhesion coefficient; Obtain the lane line position and type; Obtain the position coordinates of the autonomous driving vehicle itself ( x T , y T ),speed v T and the front wheel steering angle ; The first type includes CAVs, AVs, and HDVs; the second type includes light vehicles and heavy vehicles; Step S12: Calculate the threat level of each surrounding vehicle : Formula 1: ; Formula 2: ; Where: Indicates surrounding vehicles i degree of threat; Indicates surrounding vehicles i Reaction time of CAV, AV and HD vehicles is the default value, ; Indicates surrounding vehicles i Maximum braking deceleration for light and heavy vehicles is the default value, ; k t Indicates the reaction time adjustment coefficient of surrounding vehicles; c It represents the road adhesion coefficient; X s Represents a 0-1 variable, used to determine whether the vehicle is speeding. X s =1; k t0 , k ts All represent undetermined coefficients.

2. The method for assessing lane change risk of an autonomous driving vehicle considering the heterogeneity of surrounding vehicles according to claim 1, characterized in that: The specific process of constructing the surrounding vehicle risk energy field based on the threat level of the surrounding vehicles is as follows: Step 21: Divide the road space in the road area D into The grid is as follows: Formula 3: ; Formula 4: ; Formula 5: ; Formula 6: ; Where: L D Indicates the length of the road area D along the lane line; s var Indicates the longitudinal visual range of the vehicle; W D It represents the width of the road area perpendicular to the lane line; n represents the number of lanes contained in the road area D; l lane Indicates lane width; a 0 indicates the road grid unit length; b 0 indicates the road grid unit width; f Indicates that the road area along the lane line is divided into sections of length a The number of grid cells that are 0; Indicates that the road area perpendicular to the lane line is divided into two areas with a width of b The number of grid cells that are 0; Step 22: Calculate the risk energy of surrounding vehicles to other grids and construct the risk energy field of surrounding vehicles. The formula is as follows: ； Where: represents the longitudinal risk modification factor; represents the horizontal risk correction factor; Both are undetermined coefficients, representing the influence of distance on longitudinal and transverse risks respectively; Both are undetermined coefficients, representing the influence of speed on longitudinal and transverse risks respectively; Indicates the length of surrounding vehicles; Indicates the width of surrounding vehicles; Represents the grid coordinates of the predicted trajectory point at time T+1 under the k-th behavior intention of the surrounding vehicles; Represents the original coordinates of the grid points; It indicates the corresponding deflection value after the vehicle steering angle turns to the original coordinates of the grid point; represents the speed of the surrounding vehicles under the kth behavior intention; Indicates the steering angle when the vehicle turns; Indicates the angle between the vehicle speed direction and the positive direction of the x-axis; Indicates surrounding vehicles i The risk energy value generated by the grid where the predicted trajectory point is located to other grids under the k-th type of behavior intention; Indicates surrounding vehicles i degree of threat; Indicates surrounding vehicles i The strength of the risk energy field generated; Indicates surrounding vehicles i The probabilities of different driving intentions are k={1(turn left), 2(go straight), 3(turn right)}.

3. The method for assessing lane change risk of an autonomous driving vehicle considering the heterogeneity of surrounding vehicles according to claim 2, characterized in that: The lane line risk energy field is constructed based on the threat level of the surrounding vehicles, as follows: Formula 12: ; Where: Indicates lane lines j The strength of the risk energy field generated at other grids; Indicates the different risk radiation coefficients generated by different lane line types; represents the undetermined coefficient; Represents the y-axis coordinates of other grids; Indicates the y-axis coordinate of the grid where the lane line is located; l lane Indicates lane width.

4. The method for assessing lane change risk of an autonomous driving vehicle considering the heterogeneity of surrounding vehicles according to claim 3, characterized in that: The comprehensive risk energy field of the road area at time t is established as follows: Formula 13: ; Where: Indicates surrounding vehicles i The strength of the risk energy field generated at other grids; Indicates lane lines j The intensity of the risk energy field generated at other grids; m represents the number of surrounding vehicles in the road area; s represents the number of lane lines in the road area; Indicates the comprehensive risk energy field strength generated by surrounding vehicles and lane lines.

5. The method for assessing lane change risk of an autonomous driving vehicle considering the heterogeneity of surrounding vehicles according to claim 3, characterized in that: Calculating the autonomous vehicle The risk from surrounding vehicles at all times , as follows: Formula 14: ; Where: Indicates surrounding vehicles in the same lane as the autonomous vehicle The strength of the risk energy field generated at the location of the autonomous vehicle; Indicates surrounding vehicles that are in different lanes than the autonomous vehicle The strength of the risk energy field generated at the location of the autonomous vehicle; Indicates the number of surrounding vehicles in the same lane as the autonomous vehicle; Indicates the number of surrounding vehicles that are in different lanes from the autonomous vehicle; represents the speed of the autonomous vehicle; are all undetermined coefficients, representing the sensitivity of the longitudinal motion state of the autonomous driving vehicle to the risk of surrounding vehicles in the same lane and the sensitivity of the lateral motion state of the autonomous driving vehicle to the risk of surrounding vehicles in different lanes; Represents the steering angle of the autonomous vehicle.

6. The method for assessing lane change risk of an autonomous driving vehicle considering the heterogeneity of surrounding vehicles according to claim 5, characterized in that: Calculating the autonomous vehicle The risk forces from lane lines at all times are as follows: Formula 15: ; Where: Indicates lane lines j The strength of the risk energy field generated at the location of the autonomous vehicle; Indicates the number of lane lines in the road area; is an undetermined coefficient, which indicates the sensitivity of the lateral motion state of the autonomous driving vehicle to the lane line risk; represents the speed of the autonomous vehicle; Represents the steering angle of the autonomous vehicle.

7. The method for assessing lane change risk of an autonomous driving vehicle considering the heterogeneity of surrounding vehicles according to claim 6, characterized in that: Determining the risk level of the autonomous driving vehicle during lane change based on risk force Risk Level , as follows: Formula 16: ; Formula 17: ; Where: It represents the risk force in the risk energy field during the lane changing process of the autonomous driving vehicle; Both are preset parameters, representing the minimum and maximum risk thresholds respectively.

Images