Interactive prediction risk field construction method and system based on dynamic environment characteristics

Abstract

The invention provides an interactive prediction risk field construction method and system based on dynamic environment characteristics, and the method comprises the steps: constructing a barrier field according to the attributes and prediction behaviors of surrounding vehicles; constructing a road field according to road conditions; configuring driving factors according to characteristics of a driver; and fusing and interacting the obstacle field, the road field and the driving factors to construct an interactive prediction risk field. According to potential factors in a proposed traffic scene, corresponding risk field models are respectively constructed, and the risk field models comprise obstacle fields caused by traffic participants, road fields caused by lane lines and boundaries and driver factors determined by characteristics of drivers. Due to the fact that track prediction and relative motion calculation of surrounding vehicles are integrated, the model can perceive upcoming risks earlier, and the potential of the method applied to decision making and planning subsequently is shown.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving scenario modeling, and in particular to a method and system for constructing an interactive prediction risk field based on dynamic environmental characteristics. Background Art

[0002] Smart cars integrate multiple technologies, including sensors, computer vision, and artificial intelligence, to achieve autonomous driving, thereby greatly improving driving safety and comfort. The lateral decision-making of smart cars refers to the vehicle making appropriate lateral control actions, such as lane keeping, lane changing, and obstacle avoidance, based on the environmental information obtained during driving. The decisions of these actions directly affect the driving safety and comfort of the vehicle.

[0003] The lateral decision-making of intelligent vehicles relies on technologies such as environmental perception and scene modeling during vehicle driving, while motion planning is based on accurate assessment of road environmental risks. How to achieve accurate assessment of road environmental risks has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] One of the purposes of the present invention is to provide a method and system for constructing an interactive prediction risk field based on dynamic environmental characteristics, so as to accurately assess the risks of the road environment.

[0005] An embodiment of the present invention provides a method for constructing an interactive prediction risk field based on dynamic environmental characteristics, comprising:

[0006] Construct an obstacle field based on the attributes and predicted behaviors of surrounding vehicles;

[0007] Construct the road field according to the road conditions;

[0008] Configure driving factors based on the driver's own characteristics;

[0009] The obstacle field, road field and driving factors are integrated and interacted to construct an interactive prediction risk field.

[0010] Preferably, the steps for constructing the obstacle field are as follows:

[0011] Perform static and dynamic identification of surrounding vehicles and divide them into static and dynamic groups;

[0012] Generate the risk field strength of the static group of vehicles based on the center of mass and the posture of the vehicle;

[0013] The trajectory of the vehicles in the dynamic group is predicted and combined with the relative speed between the vehicles to calculate the relative motion trajectory;

[0014] Perform point sampling on the relative motion trajectory to determine the field strength point;

[0015] The risk field strength of each field strength point is generated based on the trajectory length and vehicle width corresponding to the field strength point.

[0016] Preferably, the steps of constructing the road field are as follows:

[0017] Analyze road conditions to determine the number of lanes, lane width, lane line type, and road curvature;

[0018] The field strength value of the road field is determined based on the number of lanes, lane width, lane line type, and road curvature.

[0019] Preferably, the driving factor configuration steps are as follows:

[0020] Assess the driver's ability and determine the driver's ability factor;

[0021] Evaluate the driver's driving style and determine the driver's style factor;

[0022] Assess the driver's mental state and determine the driver's mental state factors;

[0023] The driving factor is determined based on the driver ability factor, the driver style factor and the driver status factor.

[0024] Preferably, the steps of integrating and interacting the obstacle field, the road field and the driving factor are as follows:

[0025] The obstacle field, road field and driving factor are normalized to obtain the interactive prediction risk field. The normalization formula is as follows:

[0026]

[0027] Where m represents the number of obstacles in the current environment, n is the number of lane boundaries or lines; σ dr represents the driving factor; E road(j) (x, y) represents the road field strength at the coordinate (x, y); E k(i) (x, y) represents the dynamic field strength of the obstacle field at coordinates (x, y); E s(i) (x, y) represents the static field strength of the obstacle field at coordinates (x, y).

[0028] The present invention also provides an interactive prediction risk field construction system based on dynamic environmental characteristics, comprising: an obstacle field analysis module, a road field analysis module, a driving factor analysis module and an interactive fusion module;

[0029] Among them, the obstacle field analysis module constructs the obstacle field based on the attributes and predicted behaviors of surrounding vehicles; the road field analysis module constructs the road field based on road conditions; the driving factor analysis module configures the driving factor based on the driver's own characteristics; the interactive fusion module integrates the obstacle field, road field and driving factors to construct an interactive prediction risk field.

[0030] Preferably, the obstacle field analysis module includes: a grouping unit, a static analysis unit and a dynamic analysis unit;

[0031] Among them, the grouping unit performs static and dynamic identification of surrounding vehicles and divides them into static group and dynamic group; the static analysis unit generates the risk field strength of the vehicles in the static group based on the center of mass and posture of the vehicle; the dynamic analysis unit predicts the trajectory of the vehicles in the dynamic group, and combines it with the relative speed of the vehicle itself to calculate the relative motion trajectory, samples the point of the relative motion trajectory, determines the field strength point, and generates the risk field strength of each field strength point according to the trajectory length and vehicle width corresponding to the field strength point.

[0032] Preferably, the road field analysis module includes: a data analysis unit and a road field determination unit; wherein the data analysis unit analyzes the road conditions to determine the number of lanes, lane width, lane line type, and road curvature; the road field determination unit determines the field strength value of the road field based on the number of lanes, lane width, lane line type, and road curvature.

[0033] Preferably, the driving factor analysis module includes: an ability assessment unit, a style assessment unit, a mental state assessment unit and a driving factor determination unit;

[0034] Among them, the ability assessment unit assesses the driver's ability and determines the driver's ability factor; the style assessment unit assesses the driver's driving style and determines the driver's style factor; the mental state assessment unit assesses the driver's mental state and determines the driver's mental state factor; the driving factor determination unit determines the driving factor based on the driver's ability factor, the driver's style factor and the driver's state factor.

[0035] Preferably, the interactive fusion module performs the following operations:

[0036] The obstacle field, road field and driving factor are normalized to obtain the interactive prediction risk field. The normalization formula is as follows:

[0037]

[0038] Where m represents the number of obstacles in the current environment, n is the number of lane boundaries or lines; σ dr represents the driving factor; E road(j) (x, y) represents the road field strength at the coordinate (x, y); E k(i)(x, y) represents the dynamic field strength of the obstacle field at coordinates (x, y); E s(i) (x, y) represents the static field strength of the obstacle field at coordinates (x, y).

[0039] The present invention has the following advantages:

[0040] According to the potential factors in the proposed traffic scenario, corresponding risk field models are constructed respectively, including the obstacle field caused by traffic participants, the road field caused by lane lines and boundaries, and the driver factor determined by the driver's own characteristics. In addition, in the process of constructing the obstacle field, in order to further improve the prediction ability and accuracy of the field, the prediction trajectory is introduced in the dynamic prediction field, and the relative trajectory method is used to calculate the relative motion state between vehicles. The established interactive prediction risk field can intuitively and comprehensively describe the risks generated by different traffic elements in the current environment, providing a new basis for situation assessment in complex environments. Due to the integration of the trajectory prediction and relative motion calculation of the surrounding vehicles, the model can perceive the upcoming risks earlier, showing the potential of the method for subsequent application in decision-making and planning.

[0041] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0042] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0044] Figure 1 A schematic diagram of a method for constructing an interactive prediction risk field based on dynamic environmental characteristics in an embodiment of the present invention;

[0045] Figure 2 is a schematic diagram of a static attribute field;

[0046] Figure 3 Design schematics for dynamic prediction of field shapes;

[0047] Figure 4 It is a schematic diagram of the dynamic prediction field;

[0048] Figure 5 Schematic diagram of lane changing scenario;

[0049] Figure 6 is a schematic diagram of the road field;

[0050] Figure 7 Schematic diagram of the risk field boundary in an embodiment of the present invention;

[0051] Figure 8 Schematic diagram of a system for constructing an interactive prediction risk field based on dynamic environmental characteristics in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0053] The embodiment of the present invention provides a method for constructing an interactive prediction risk field based on dynamic environmental characteristics, such as Figure 1 As shown, including:

[0054] Step 1: Construct an obstacle field based on the attributes and predicted behaviors of surrounding vehicles;

[0055] Surrounding vehicles are the factor that has the greatest impact on the driving of the ego vehicle and are also the main source of risk. The model built for this is the obstacle field. It can be divided into static attribute field and dynamic prediction field according to the motion state of the obstacle vehicle. The static field is affected by the objective attributes of the vehicle, which is equivalent to the impact of a stationary vehicle on the ego vehicle. The magnitude of the field strength mainly depends on the size, position and posture of the obstacle. For a moving obstacle vehicle, the dynamic field not only describes the risk caused by its relative motion state with the ego vehicle, but also integrates the prediction results into it, diffuses the risk and projects it onto the predicted trajectory. The magnitude and direction of the field strength are mainly determined by the relative speed and acceleration of the obstacle vehicle and the ego vehicle, as well as the predicted future trajectory;

[0056] Step 2: Construct a road field based on road conditions;

[0057] When a vehicle is driving on the road, it will inevitably be restricted by the lanes and lane lines it is traveling on, and will drive in an orderly manner according to the established constraints and rules. The road field reflects the impact of these road conditions on driving safety. In order to make the constructed road field model more universal, the field strength value is mainly related to parameters such as the number of lanes, lane width, lane line type, and road curvature;

[0058] Step 3: Configure driving factors according to the driver's own characteristics;

[0059] Drivers have completely different behavioral characteristics and driving abilities, which leads to different risk judgments and personalized decision-making behaviors even in the same driving scenario. For example, in terms of driving style, conservative drivers tend to overestimate some risks in the environment to a certain extent because they are timid compared to aggressive drivers; in terms of driving ability, skilled drivers have a relatively lower risk factor than novices because they can calmly deal with various driving conditions; in terms of mental state, fatigued drivers will mistakenly underestimate environmental risks to a certain extent compared to when they are full of energy. Therefore, the driver factor is to describe the impact of the above three types of driver factors on risk;

[0060] Step 4: Integrate and interact the obstacle field, road field and driving factors to construct an interactive prediction risk field.

[0061] The steps of fusion interaction of obstacle field, road field and driving factor are as follows: normalize the obstacle field, road field and driving factor to obtain the interactive prediction risk field. The normalization formula is as follows:

[0062]

[0063] Where m represents the number of obstacles in the current environment, n is the number of lane boundaries or lines; σ dr represents the driving factor; E road(j) (x, y) represents the road field strength at the coordinate (x, y); E k(i) (x, y) represents the dynamic field strength of the obstacle field at coordinates (x, y); E s(i) (x, y) represents the static field strength of the obstacle field at coordinates (x, y);

[0064] For the surrounding vehicles in the environment, when it is stationary on the road, the risk it generates is equivalent to a static obstacle, and the field strength is only related to its own objective attributes such as size and posture. When it is driving on the road, in addition to the risks generated by static objective attributes, its relative motion relationship with the vehicle itself will also generate risks, and its future driving intention and trajectory will change the distribution mode and shape of the field strength. Therefore, the obstacle field is further refined into a static attribute field related to objective attributes and a dynamic prediction field related to the motion state. That is, in one embodiment, the steps for constructing the obstacle field are as follows: static and dynamic identification of surrounding vehicles is performed, and they are divided into static groups and dynamic groups; based on the center of mass of the vehicle and the posture of the vehicle, the risk field strength of the vehicles in the static group is generated; the trajectory of the vehicles in the dynamic group is predicted and combined with the relative speed of the vehicle to calculate the relative motion trajectory; the relative motion trajectory is sampled at points to determine the field strength points; based on the trajectory length and vehicle width corresponding to the field strength points, the risk field strength of each field strength point is generated.

[0065] The design of the static attribute field function follows the following rules: the overall risk distribution is around the vehicle's outer dimensions. As the distance from the center of the vehicle increases, the risk gradually decreases, with the field intensity peak being on the entire surface of the vehicle. Affected by the road and traffic regulations, there are certain restrictions on the direction of the vehicle's travel, that is, the lateral speed is usually much smaller than the longitudinal speed, so that the longitudinal risk range of the obstacle vehicle should be much larger than the lateral one. The location of the risk field changes with the center of mass of the vehicle, and the direction changes with the posture of the vehicle. Based on the above analysis, the risk field intensity of the vehicle labeled i at rest can be expressed as:

[0066]

[0067] Among them, x obs(i) ,y obs(i) is the horizontal and vertical coordinates of the center of mass of the obstacle vehicle; (x, y) is the position of any point in space, A obs is the field strength coefficient; β is the high-order coefficient, which is used to adjust the shape of the potential field peak; σ x and σ y The obstacle shape coefficient is used to adjust the relationship between the field intensity distribution and the lateral and longitudinal dimensions of the vehicle. In addition, considering the different positions of the obstacles, the field intensity distribution is rotated and translated through projection transformation based on the current heading angle ψ of the vehicle. The schematic diagram of the static attribute field is shown in Figure 2 As shown in the figure, the field strength increases as the distance from the center of the obstacle decreases. For points close to the edge of the obstacle, the field strength increases rapidly, and the field strength at the spatial position occupied by the entire obstacle is equal and is the maximum value. On the other hand, the overall shape of the risk field is determined by the size of the obstacle vehicle. The obstacle shape coefficient can be determined by a pre-configured shape coefficient determination library, that is, according to the brand, model, etc. of the vehicle, the corresponding σ is determined from the shape coefficient determination library x and σ y ; A obs , β are pre-configured.

[0068] The dynamic prediction field is mainly used to describe the relative state between the moving traffic participants and the ego vehicle and the potential risks generated. It has the following characteristics: the greater the relative speed and the smaller the relative distance, the greater the possibility of causing a traffic accident and the greater the field strength of the dynamic prediction field. The direction of the relative speed affects the direction of the field. When the obstacle vehicle is faster than the ego vehicle, the obstacle vehicle is equivalent to "moving away" from the ego vehicle, and the field is facing forward, and vice versa. In addition to the relative motion state of the obstacle vehicle and the ego vehicle, the dynamic prediction field also incorporates the predicted trajectory, so that the shape of the risk field can change and expand along the direction of its predicted trajectory. Based on the trajectory characteristics of the obstacle vehicle, the present invention adopts the following four steps to construct the shape of the dynamic prediction field: Figure 3 First, as Figure 3As shown in (a), the risk distribution is bent along the predicted trajectory to form a torus using the same method as the static attribute field. Secondly, the risk should gradually decay with the distance perpendicular to the trajectory, so the cross section of the field is designed to be a Gaussian cross section, as shown in Figure 3 (b); Next, considering that the risk should gradually decay along the direction of the predicted trajectory, and the decay speed in the direction perpendicular to the trajectory should also gradually slow down with the increase of the arc length, this paper designs the height and width coefficients and associates them with the length of the predicted trajectory, so that the risk field gradually expands and decreases in the horizontal and vertical directions along the trajectory, as shown in Figure 3 (c)(d) effect. Based on the above analysis, if Figure 3 As shown in (a), for a point (x, y) in the space to be calculated, it can be projected onto the predicted trajectory, that is, the obtained projection point (x n ,y n )satisfy:

[0069]

[0070] In the formula, All points in the obstacle vehicle’s predicted trajectory sequence p at time t, i.e., the predicted trajectory of the obstacle vehicle within 5s;

[0071] At the same time, the trajectory length s corresponding to this point n It can be expressed as:

[0072]

[0073] Based on the trajectory length s n , defines the height coefficient A d , which determines the steepness and attenuation characteristics of the risk field along the predicted trajectory, is expressed as:

[0074] A d =k s (s n -d pre )(s n -s total );

[0075] Among them, k s is the steepness coefficient, d pre =t p v obs(i) is the forward-looking preview distance; t p Forward-looking preview time, configured as 2s; s total is the total length of the predicted trajectory at the current moment; v obs(i) is the speed of the obstacle vehicle;

[0076] For the field width design, the width factor σ is defined v, which determines the horizontal expansion characteristics of the field, is expressed as follows:

[0077] σ v =(k l +k ψ |ψ|)s n +c obs ;

[0078] Among them, c obs Indicates the basic lateral range of the obstacle vehicle's location field, which is related to the vehicle width; k l k is the field diffusion slope when the obstacle vehicle is traveling in a straight line; ψ is the steering gain coefficient, which makes the width of the field enlarge or reduce the width of the risk field as the vehicle heading changes;

[0079] Therefore, the dynamic prediction field of the obstacle vehicle labeled i can be preliminarily expressed as:

[0080]

[0081] The schematic diagram of the dynamic prediction field is as follows Figure 4 As shown in the figure, the risk field completely extends along the predicted trajectory. In the direction of the trajectory, the risk gradually decreases as the distance from the vehicle increases. In the direction perpendicular to the trajectory, the field strength reaches the maximum value at each trajectory point and decays to both sides. At the same time, as the trajectory extends, the maximum field strength gradually decreases. At this point, the shape of the risk field is no longer limited to the traditional ellipse or ring, but is completely extended based on the motion state of the obstacle vehicle.

[0082] Although the above dynamic prediction field can well describe the risk characteristics of the obstacle vehicle during its driving process and take into account the speed, trajectory and other characteristics of the obstacle vehicle, the ultimate purpose of the risk field is to serve the subsequent decision-making of the ego vehicle. Therefore, the relative motion between the ego vehicle and other vehicles still needs to be taken into account in the dynamic prediction field.

[0083] A relative trajectory approach is used to incorporate relative motion into the risk field without changing the field function. Figure 5 As shown, assuming the vehicle speed is V obs The obstacle vehicle's trajectory is a curve during the entire lane change process. 1 At this moment, the predicted trajectory is output. However, from the perspective of the ego vehicle, due to the relative motion relationship between the two vehicles, the risks generated do not completely follow the predicted trajectory, but the so-called "relative trajectory". For example, if the speed of the obstacle vehicle in front is much higher than that of the ego vehicle or the vehicle behind is much slower than that of the ego vehicle, even if the obstacle vehicle is very close to the ego vehicle, it will not have much impact on the driver's driving; on the contrary, only when there is a slow car in front and a fast car behind, a large number of potential risks will be generated.

[0084] To this end, this paper outputs the predicted trajectory based on the relative speed ΔV = V ego -V obs Adjusted and optimized, and constructed the relative trajectory sequence p relv :

[0085]

[0086] like Figure 5 As shown, there is an obstacle car at T 1 The predicted trajectory within 5 seconds is equivalent to the relative trajectory of the ego vehicle when it is stationary. When the ego vehicle is slower than the obstacle vehicle, the obstacle vehicle is moving forward relative to the ego vehicle, and the relative trajectory is shown by the dotted line. When the speeds of the ego vehicle and the obstacle vehicle are equal, the relative motion of the obstacle vehicle only exists in the lateral direction, and the relative trajectory is perpendicular to the horizontal direction. When the ego vehicle is faster than the obstacle vehicle, the obstacle vehicle is moving backward relative to the ego vehicle, and the relative trajectory is transferred to the rear of the obstacle vehicle.

[0087] To this end, the predicted trajectory sequence p in the dynamic prediction field function calculation is replaced by the relative trajectory sequence p relv This is the final dynamic prediction field model;

[0088] When a vehicle is driving on the road, it will definitely be affected by the lane lines. Lane lines not only play the role of dividing the road, so that vehicles on the road can drive in an orderly manner; on the other hand, the double yellow lines in the lane lines can serve as a warning to the opposite lanes. It is forbidden to cross the double yellow lines unless in special circumstances. Therefore, from the perspective of potential field design, the road field should have the following characteristics: the road boundary has the highest degree of danger, followed by the double yellow line, and the ordinary lane line has a lower risk. Vehicles should stay in the center of the lane as much as possible, so the center of each lane has the lowest risk. The road risk field is only related to the road structure and should extend along the lane, and has nothing to do with traffic participants. After analysis, the steps for constructing the road field are as follows: Analyze the road conditions, determine the number of lanes, lane width, lane line type, and road curvature; determine the field strength value of the road field based on the number of lanes, lane width, lane line type, and road curvature.

[0089]

[0090] Constant η LANE is the lane line type coefficient to distinguish the danger level of different lane lines. The white dashed line has the smallest value, the double yellow line has a larger value, and the road boundary has the largest value. In the present invention, the road boundary, double yellow line and dashed line are 8, 3, and 1 respectively; is the minimum Euclidean distance from the vehicle's current position to the lane centerline; d LANE is the lane width; σ LANE The road adjustment coefficient ensures that the field strength varies within a reasonable range and changes its speed with lateral distance, which is pre-configured;

[0091] Road field diagram Figure 6 As shown in the figure, the risk is kept at the minimum value in the center of each lane, which is intuitively displayed by the dark blue area. As the distance from the center of the lane increases, the risk gradually increases and reaches the maximum value at the boundary. For ordinary road boundary dotted lines, vehicles are allowed to cross, so even on the boundary line, the risk is only transitioned with a slightly higher field strength value, represented by the light blue area. This enables the road field to guide the vehicle to move along the center of the current lane, while meeting the needs of vehicle motion planning across the lane, so that the vehicle can complete driving tasks such as lane changing. However, for the lowest road boundary (near Y = 0), the vehicle should avoid entering or approaching these areas during driving, so the potential field color is the darkest, the field strength is the largest, and the rise is the largest; similarly, in the double yellow line area near Y = 12, the vehicle should also stay as far away from this area as possible to avoid collision risks, so the risk is also relatively large and has a large rise.

[0092] Different drivers have different behavioral characteristics and driving abilities, which leads to different risk judgments. The risk judgments determined by the driver's own characteristics are summarized into the following three categories: driving ability, driving style, mental state, and driver factor σ dr It is a dimensionless coefficient between 0 and 1 determined based on the above three aspects of the driver's performance. Its value reflects the driver's sensitivity and acceptance of environmental risks during driving. The larger the value, the more attention the driver pays to environmental risks at this moment. The steps for configuring driving factors are as follows: evaluate the driver's ability and determine the driver's ability factor; evaluate the driver's driving style and determine the driver's style factor; evaluate the driver's mental state and determine the driver's mental state factor; determine the driving factor based on the driver's ability factor, driver style factor and driver state factor. The driver factors that affect the distribution of risk fields are defined as:

[0093] σ dr =η 1 σ skill +η 2 σ style +η 3 σ phy&psy ;

[0094] Among them, σ skill is the driver capability factor, σ style is the driver style factor, σ phy&psy is the driver's mental state factor, η 1 , η 2 , η 3 is the weight coefficient corresponding to each factor.

[0095] When constructing a risk field for a complex traffic scenario, the size of the risk field needs to be fully considered, taking into account the motion relationship between the surrounding vehicles and the vehicle itself and the driver's field of view. For the design of the risk field range, it is necessary to strike a balance between prediction effect and computational efficiency. The larger the size of the field, the more content needs to be calculated at the same time, and the worse the real-time performance of the algorithm; conversely, the smaller the size, the less comprehensive the features described in the field, and subsequent decisions are prone to fall into local optimal solutions. Therefore, if Figure 7 As shown, the scope of the field is limited from two aspects: the feasible area of ​​the vehicle in the short term and the observation area of ​​the driver.

[0096] The feasible area is mainly determined by the speed limit of the road and the acceleration performance of the vehicle, and is defined as the acceleration of the vehicle from the current speed v e Accelerate to maximum speed v max and decelerate to the minimum speed v min The range included. Within this range, every position in the risk field can be reached through appropriate acceleration and deceleration behavior of the ego vehicle, ensuring the feasibility of subsequent strategy calculation and control execution. To ensure that the established risk field can move with the ego vehicle, this value is calculated using relative distance.

[0097] From the current position x e Starting from the beginning, the area reached by the vehicle through acceleration is:

[0098]

[0099] where a e is the vehicle acceleration, take 2m / s 2 ;v max Determined by the larger value of the road's maximum speed limit and the vehicle's desired speed.

[0100] Similarly, the rear area can also get:

[0101]

[0102] where d e is the vehicle deceleration, take -3m / s 2 ;v min It is determined by the smaller value of the road's minimum speed limit and the vehicle's desired speed.

[0103] Drivers usually reserve a certain amount of forward and rearward vision distances during daily driving to ensure that they can analyze the environmental status in advance and respond. Therefore, in addition to the feasible area calculated based on vehicle motion, an observation area is defined to reflect the driver's observation range of the environment. The observation area is determined by the headway time (TH). Considering that the rear vehicle generally adjusts its strategy autonomously according to the movement of the vehicle, the danger level is relatively low, while the sudden behavior of the front vehicle will pose a major threat. Therefore, the rear observation area is shortened, and the front and rear headways are TH and TH, respectively. f =3,TH r = -1. The observation area is calculated as follows:

[0104] x r =v e ·TH r ;

[0105] x f =v e ·TH f ;

[0106] Finally, the longitudinal range of the risk field takes the larger value of the feasible area and the observation area to meet the requirements of driver behavior and vehicle control at the same time, that is:

[0107] X r =max(Δx r , x r );

[0108] X f =max(Δx f , x f );

[0109] In the lateral direction, the extent of the risk field is related to the number of lanes and the lane width and is defined as:

[0110] Y = nW;

[0111] Where n is the number of lanes and W is the lane width.

[0112] The scope of the risk field is as follows Figure 7 As shown, the state of the vehicle at time t can be expressed as:

[0113]

[0114] in, Indicates the horizontal and vertical coordinates of the vehicle. Indicates vehicle speed. represents acceleration, represents the heading angle, Indicates the steering wheel angle.

[0115] At time t-1, the risk field area is shown in the dotted box. For vehicles and static obstacles outside the window, it is difficult to cause imminent collision risks, so they are temporarily ignored during calculations. As the ego vehicle travels to time t, the risk field also moves to the solid line box area. At this time, the obstacle vehicles and static obstacles enter the ego vehicle's field of view, and the risk field needs to take multiple traffic participants into consideration at the same time.

[0116] In order to achieve accurate assessment of the driver's various abilities, in one embodiment, the driving ability factor assessment steps are as follows:

[0117] Obtain the basic data of the driver; basic data includes: years of obtaining the driver's license, age, physical examination data, etc.;

[0118] Obtain the driver's historical driving data; the historical driving data includes: the vehicle model data driven, the total driving time or total mileage of various vehicle models, etc.;

[0119] A pre-configured first feature parameter extraction library is used to extract first feature parameters from driving data and historical driving data to obtain multiple first feature parameters; wherein the first feature parameters include: a parameter indicating the number of years of obtaining a driving license, a parameter indicating age, a parameter indicating a physical examination data evaluation score, a parameter indicating the total driving time or total mileage of vehicles of various models, etc.;

[0120] Using each first characteristic parameter as an index item, searching a pre-configured driving ability factor evaluation library to determine a corresponding driving ability factor; the first characteristic parameter in the driving ability factor evaluation library is correspondingly associated with the driving ability factor;

[0121] The driving style factor evaluation steps are as follows:

[0122] Obtain the driver's historical driving data; for example, use the driving data of the past year as the basis for analysis;

[0123] Based on the preset event library, the historical driving data is screened to obtain multiple event data; the event data includes: whether the yellow light rush occurs when passing the traffic light; whether the red light running occurs when passing the traffic light; whether the driver gives way to another vehicle at the intersection; whether the driver frequently changes lanes during driving (i.e., the number of lane changes is greater than the preset threshold); whether the driver frequently overtakes other vehicles during driving (i.e., the number of overtaking is greater than the preset threshold), etc.;

[0124] The data of multiple events are grouped and counted, and the statistical data are quantified. Based on the quantified parameter values, an analysis vector is constructed; the number of red light running, the ratio of yellow light running to the number of red light running, the number of courtesy at intersections, the ratio of frequent lane changes to the total number of driving, the ratio of frequent overtaking to the total number of driving, etc.

[0125] Retrieving a driving style factor corresponding to the analysis vector from a pre-configured driving style factor evaluation library using the analysis vector; the analysis vectors in the driving style factor evaluation library are associated with the driving style factors in a one-to-one correspondence;

[0126] The analysis steps of the driver's mental state factor are as follows:

[0127] Obtain driving data of the current driver; driving data includes: driving time;

[0128] Acquire a facial image of the driver during driving; input the facial image into a pre-trained converged neural network to obtain fatigue assessment data;

[0129] Based on driving time and fatigue assessment data, the corresponding driver mental state factor is retrieved from the pre-configured mental state factor analysis library.

[0130] In order to achieve the accuracy of the construction of the dynamic field; in one embodiment, after the trajectory of the dynamic obstacle is predicted, the trajectory can be corrected by constructing a risk field with the dynamic obstacle as the target. The specific correction steps are as follows:

[0131] Associating the risk fields within the preset range around each trajectory point on the predicted trajectory;

[0132] When the field strength of each point in the risk field corresponding to the trajectory point is less than or equal to the preset threshold, the position of the trajectory point is not adjusted;

[0133] When the field strength of each point in the risk field corresponding to the trajectory point is greater than the preset threshold, the nearest point is extracted as the reference point; the trajectory point is adjusted away from the reference point, and the adjustment distance is the difference between the distance between the reference point and the trajectory point and the threshold.

[0134] The present invention also provides a system for constructing an interactive prediction risk field based on dynamic environmental characteristics, such as Figure 8 As shown, it includes: an obstacle field analysis module 1, a road field analysis module 2, a driving factor analysis module 3 and an interactive fusion module 4;

[0135] Among them, the obstacle field analysis module 1 constructs the obstacle field based on the attributes and predicted behaviors of the surrounding vehicles; the road field analysis module 2 constructs the road field based on the road conditions; the driving factor analysis module 3 configures the driving factor based on the driver's own characteristics; the interactive fusion module 4 integrates the obstacle field, road field and driving factors to construct an interactive prediction risk field.

[0136] Among them, the obstacle field analysis module includes: a grouping unit, a static analysis unit and a dynamic analysis unit;

[0137] Among them, the grouping unit performs static and dynamic identification of surrounding vehicles and divides them into static group and dynamic group; the static analysis unit generates the risk field strength of the vehicles in the static group based on the center of mass and posture of the vehicle; the dynamic analysis unit predicts the trajectory of the vehicles in the dynamic group, and combines it with the relative speed of the vehicle itself to calculate the relative motion trajectory, samples the point of the relative motion trajectory, determines the field strength point, and generates the risk field strength of each field strength point according to the trajectory length and vehicle width corresponding to the field strength point.

[0138] Among them, the road field analysis module includes: a data analysis unit and a road field determination unit; wherein the data analysis unit analyzes the road conditions and determines the number of lanes, lane width, lane line type, and road curvature; the road field determination unit determines the field strength value of the road field based on the number of lanes, lane width, lane line type, and road curvature.

[0139] The driving factor analysis module includes: an ability assessment unit, a style assessment unit, a mental state assessment unit and a driving factor determination unit;

[0140] Among them, the ability assessment unit assesses the driver's ability and determines the driver's ability factor; the style assessment unit assesses the driver's driving style and determines the driver's style factor; the mental state assessment unit assesses the driver's mental state and determines the driver's mental state factor; the driving factor determination unit determines the driving factor based on the driver's ability factor, the driver's style factor and the driver's state factor.

[0141] The interactive fusion module performs the following operations:

[0142] The obstacle field, road field and driving factor are normalized to obtain the interactive prediction risk field. The normalization formula is as follows:

[0143]

[0144] Where m represents the number of obstacles in the current environment, n is the number of lane boundaries or lines; σ dr represents the driving factor; E road(j) (x, y) represents the road field strength at the coordinate (x, y); E k(i) (x, y) represents the dynamic field strength of the obstacle field at coordinates (x, y); E s(i) (x, y) represents the static field strength of the obstacle field at coordinates (x, y).

[0145] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for constructing an interactive prediction risk field based on dynamic environmental characteristics, characterized in that: include: Construct an obstacle field based on the attributes and predicted behaviors of surrounding vehicles; Construct the road field according to the road conditions; Configure driving factors based on the driver's own characteristics; The obstacle field, road field and driving factors are integrated and interacted to construct an interactive prediction risk field.

2. The interactive prediction risk field construction method based on dynamic environment characteristics according to claim 1 is characterized in that: The steps to construct the obstacle course are as follows: Perform static and dynamic identification of surrounding vehicles and divide them into static and dynamic groups; Generate the risk field strength of the static group of vehicles based on the center of mass and the posture of the vehicle; The trajectory of the vehicles in the dynamic group is predicted and combined with the relative speed between the vehicles to calculate the relative motion trajectory; Perform point sampling on the relative motion trajectory to determine the field strength point; The risk field strength of each field strength point is generated based on the trajectory length and vehicle width corresponding to the field strength point.

3. The interactive prediction risk field construction method based on dynamic environment characteristics according to claim 1 is characterized in that: The steps to build a road field are as follows: Analyze road conditions to determine the number of lanes, lane width, lane line type, and road curvature; The field strength value of the road field is determined based on the number of lanes, lane width, lane line type, and road curvature.

4. The interactive prediction risk field construction method based on dynamic environment characteristics according to claim 1 is characterized in that: The steps for driving factor configuration are as follows: Assess the driver's ability and determine the driver's ability factor; Evaluate the driver's driving style and determine the driver's style factor; Assess the driver's mental state and determine the driver's mental state factors; The driving factor is determined based on the driver ability factor, the driver style factor and the driver status factor.

5. The interactive prediction risk field construction method based on dynamic environment characteristics according to claim 1 is characterized in that: The steps of fusion interaction of obstacle field, road field and driving factor are as follows: The obstacle field, road field and driving factor are normalized to obtain the interactive prediction risk field. The normalization formula is as follows: Where m represents the number of obstacles in the current environment, n is the number of lane boundaries or lines; σ dr represents the driving factor; E road(j) (x, y) represents the road field strength at the coordinate (x, y); E k(i) (x, y) represents the dynamic field strength of the obstacle field at coordinates (x, y); E s(i) (x, y) represents the static field strength of the obstacle field at coordinates (x, y).

6. A system for constructing interactive prediction risk fields based on dynamic environmental characteristics, characterized in that: include: Obstacle field analysis module, road field analysis module, driving factor analysis module and interactive fusion module; Among them, the obstacle field analysis module constructs the obstacle field based on the attributes and predicted behaviors of surrounding vehicles; the road field analysis module constructs the road field based on road conditions; and the driving factor analysis module configures the driving factor based on the driver's own characteristics. The interactive fusion module integrates the obstacle field, road field and driving factors to construct an interactive prediction risk field.

7. The interactive prediction risk field construction system based on dynamic environmental characteristics according to claim 6, characterized in that: The obstacle field analysis module includes: a grouping unit, a static analysis unit and a dynamic analysis unit; Among them, the grouping unit performs static and dynamic identification of surrounding vehicles and divides them into static group and dynamic group; the static analysis unit generates the risk field strength of the vehicles in the static group based on the center of mass and posture of the vehicle; the dynamic analysis unit predicts the trajectory of the vehicles in the dynamic group, and combines it with the relative speed of the vehicle to calculate the relative motion trajectory; the relative motion trajectory is sampled to determine the field strength point, and the risk field strength of each field strength point is generated according to the trajectory length and vehicle width corresponding to the field strength point.

8. The interactive prediction risk field construction system based on dynamic environmental characteristics according to claim 6, characterized in that: The road field analysis module includes: a data analysis unit and a road field determination unit; wherein the data analysis unit analyzes the road conditions and determines the number of lanes, lane width, lane line type, and road curvature; the road field determination unit determines the field strength value of the road field based on the number of lanes, lane width, lane line type, and road curvature.

9. The interactive prediction risk field construction system based on dynamic environmental characteristics according to claim 6, characterized in that: The driving factor analysis module includes: an ability assessment unit, a style assessment unit, a mental state assessment unit and a driving factor determination unit; Among them, the ability assessment unit assesses the driver's ability and determines the driver's ability factor; the style assessment unit assesses the driver's driving style and determines the driver's style factor; the mental state assessment unit assesses the driver's mental state and determines the driver's mental state factor; the driving factor determination unit determines the driving factor based on the driver's ability factor, the driver's style factor and the driver's state factor.

10. The interactive prediction risk field construction system based on dynamic environment characteristics according to claim 6, characterized in that: The interactive fusion module performs the following operations: The obstacle field, road field and driving factor are normalized to obtain the interactive prediction risk field. The normalization formula is as follows: Where m represents the number of obstacles in the current environment, n is the number of lane boundaries or lines; σ dr represents the driving factor; E road(j) (x, y) represents the road field strength at the coordinate (x, y); E k(i) (x, y) represents the dynamic field strength of the obstacle field at coordinates (x, y); E s(i) (x, y) represents the static field strength of the obstacle field at coordinates (x, y).

Images