A method for early warning of collision risk of autonomous road traffic system vehicle

Abstract

The application discloses a kind of autonomous road traffic system vehicle conflict risk early warning method, belong to road traffic safety technical field.Method includes: initialization, input main car planning track and target car predicted track, set initial time window;Track point extraction, track point pair is extracted from track according to preset time difference and constructs to be detected set;Collision detection, the collision relationship of track point pair is judged based on two-way collision detection area model;Loop control, according to the detection result, update time window or output conflict index;The result is fed back to trajectory planning system and continuously iterated execution.The application solves the technical limitations of traditional methods in complex motion trajectory processing and two-dimensional collision relationship identification through collision detection area model and progressive time window detection mechanism, significantly improves the conflict identification accuracy and system robustness, provides reliable technical support for real-time decision-making of autonomous vehicle.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of road traffic safety, and particularly relates to a self-driving road traffic system vehicle conflict risk early warning method. BACKGROUND

[0002] With the continuous development of automatic driving, vehicle-road cooperation and high-precision perception and positioning technologies, the environmental perception, behavior prediction and autonomous decision-making capabilities of vehicles are continuously enhanced, and urban traffic systems are evolving towards intelligent and autonomous directions. Under the background of autonomous traffic systems, the safety of autonomous vehicles becomes increasingly important. Relying on the autonomous perception and behavior prediction capabilities of vehicles, autonomous identification and early warning of potential conflict risks through intelligent algorithms have become a key direction to improve the active safety and traffic efficiency of autonomous traffic systems.

[0003] In the autonomous traffic system, thanks to the widespread deployment of high-precision sensors and significant advances in communication technologies, the motion trajectory information and entity contour data of road participants become more easily accessible and more accurate; at the same time, autonomous vehicles can autonomously predict the behavior of surrounding traffic participants, thereby obtaining their future trajectories. The above data provides feasibility and accuracy guarantees for autonomous identification of trajectory-based conflict risks.

[0004] Traffic conflict identification is an important prerequisite for active safety early warning, which usually relies on the analysis of the trajectory interaction relationship between traffic participants to identify potential collision risks in advance. The widely used conflict identification indicators currently include Time to Collision (TTC) and Post Encroachment Time (PET), etc. The former reflects the time required for two vehicles to collide at the current speed, and the latter measures the time interval between one party leaving the conflict area and the other party entering. These indicators have been widely used in vehicle conflict identification, risk assessment and traffic modeling, etc.

[0005] However, the existing traditional conflict identification methods mainly face traditional traffic systems, and when faced with multi-agent interaction, complex trajectory behavior in autonomous traffic systems, and autonomous decision-making scenarios with high timeliness and high precision requirements, there are still some deficiencies: on the one hand, traditional models are mostly based on idealized assumptions (such as uniform linear motion), which are difficult to reflect the dynamic characteristics of actual traffic participants in non-linear behaviors such as variable speed, turning, and avoiding; on the other hand, existing methods mostly focus on one-dimensional longitudinal conflict situations, ignoring the influence of lateral motion characteristics and entity contours on collision risks. In real road scenarios, two-dimensional conflicts are widespread, and the direction, shape and relative position of traffic participants have an important influence on the occurrence of collisions. Although existing methods attempt to implement geometric-level collision detection through rectangular boundary modeling, this approach has the problems of complex classification and high computational cost. SUMMARY

[0006] The present application aims to solve the problems of the prior art, and provides a vehicle collision risk warning method for an autonomous road traffic system, which introduces a collision detection zone model to accurately determine the collision relationship between traffic participants in two-dimensional space and avoid complex classification processing. At the same time, a non-aligned progressive time window detection mechanism is proposed, which dynamically adjusts the time offset relationship of the trajectory points of two vehicles, breaks through the traditional time alignment restriction, and accurately calculates the warning collision time index (similar to TTC) and the dangerous time window index (similar to PET). In addition, by constructing a warning-decision closed-loop system, the prediction and planning data of the connected vehicle are received in real time, and the risk identification result is fed back to the trajectory planning system to form a closed loop. The method significantly improves the adaptability to dynamic trajectories and two-dimensional collision characteristics, and provides strong support for the automatic driving vehicle to realize efficient and reliable autonomous decision-making in the automatic driving process.

[0007] To achieve the above-mentioned purpose, the present application provides the following scheme: a vehicle collision risk warning method for an autonomous road traffic system, comprising the following steps:

[0008] S1, detecting whether there is a target vehicle meeting the warning condition in the automatic driving process of a host vehicle; when there is a target vehicle meeting the warning condition, obtaining a planning trajectory of the host vehicle and a prediction trajectory of the target vehicle, and initializing a time window parameter k;

[0009] S2, constructing a to-be-detected set under the time window parameter k; the to-be-detected set contains a plurality of trajectory point pairs; each trajectory point pair is composed of a trajectory point in the planning trajectory and a trajectory point in the prediction trajectory meeting a preset condition;

[0010] S3, constructing a collision detection zone model, and sequentially performing collision judgment on all trajectory point pairs in the to-be-detected set based on the collision detection zone model;

[0011] S4, when there is a trajectory point pair with collision, outputting a warning information; when there is no trajectory point pair with collision, outputting a result without warning; updating the time window parameter k and returning to S2;

[0012] S5, feeding back the result of S4 to the trajectory planning system of the host vehicle, and performing a new decision cycle judgment based on the updated planning trajectory of the host vehicle and the prediction trajectory of the target vehicle.

[0013] Further preferably, in S1, the planning trajectory and the prediction trajectory each contain N trajectory points, the sampling time of the trajectory points with the same serial number is consistent, and each trajectory point contains a rectangular contour center point coordinate, an orientation angle, a rectangular contour length and a width of the vehicle in a global coordinate system.

[0014] Further preferably, the collision detection zone model comprises a target vehicle collision detection zone model and a host vehicle collision detection zone model;

[0015] The target vehicle collision detection zone model is used to determine whether the center point of the host vehicle is located in the collision detection zone of the target vehicle;

[0016] The host vehicle collision detection zone model is used to determine whether the center point of the target vehicle is located in the collision detection zone of the host vehicle;

[0017] The method for determining whether a collision occurs in S3 comprises:

[0018] The center point of the host vehicle is located in the collision detection zone of the target vehicle and the center point of the target vehicle is located in the collision detection zone of the host vehicle.

[0019] Further preferably, the calculation process of the target vehicle collision detection zone model comprises:

[0020] S31, establishing a local coordinate system with the center point of the target vehicle as the origin Obtaining the relative coordinates and the relative orientation angle of the host vehicle in the local coordinate system through coordinate transformation;

[0021] S32, obtaining the collision detection zone of the target vehicle based on the relative orientation angle;

[0022] S33, obtaining a first collision index based on the relative coordinates, and determining whether the center point of the host vehicle is in the collision detection zone of the target vehicle based on the first collision index.

[0023] Further preferably, the calculation method of the relative coordinates comprises:

[0024] x i|j =sinθ j (x i -x j )-cosθ j (y i -y j );

[0025] y i|j =cosθ j (x i -x j )+sinθ j (y i -y j );

[0026] In the formula, (x i|j ,y i|j ) is the relative coordinates of the host vehicle; x i , y​i These are the x and y coordinates of the center point of the rectangular outline of the main vehicle; x j y j These are the x and y coordinates of the center point of the rectangular outline of the target vehicle, respectively; θ j The heading angle of the target vehicle;

[0027] The method for calculating the relative orientation angle includes:

[0028] θ i|j =θ i -θ j +90°;

[0029] In the formula, θ i The heading angle of the main vehicle.

[0030] More preferably, the method for calculating the collision detection zone of the target vehicle includes:

[0031] D j =d j +l i |cosθ i|j |+d i |sinθ i|j |;

[0032] L j =l j +l i |sinθ i|j |+d i |cosθ i|j |;

[0033] In the formula, D j L j These are the width and length of the collision detection zone for the target vehicle, respectively; i d i These represent the length and width of the rectangular outline of the main vehicle, respectively; j d j These represent the length and width of the rectangular outline of the target vehicle, respectively; i |cosθ i|j |+d i |sinθ i|j | represents the rectangular outline of the main vehicle in the local coordinate system. The projected length on the x-axis; l i |sinθ i|j |+d i |cosθ i|j | Represents the rectangular outline of the target vehicle in the local coordinate system The projected length on the y-axis.

[0034] More preferably, the method for calculating the first collision index includes:

[0035]

[0036] When , it indicates that the center point of the host vehicle is in the collision detection area of the target vehicle; when , it indicates that the center point of the host vehicle is not in the collision detection area of the target vehicle.

[0037] Further preferably, in S4, the pre-warning information includes a danger time window and a pre-warning collision time.

[0038] The danger time window includes:

[0039] T1=k*Delta t;

[0040] The pre-warning collision time includes:

[0041] T2=m'*Delta t;

[0042] In the formula, m' represents the minimum serial number of the host vehicle trajectory point in all collision trajectory point pairs in the to-be-detected set; and Delta t represents a fixed time interval.

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

[0044] The present application realizes accurate determination of two-dimensional space collision relationship by innovatively introducing a collision detection area model, avoiding the complexity of traditional collision type classification; at the same time, a progressive time window detection mechanism is adopted, effectively solving the complex trajectory processing problem. This method significantly improves the robustness and applicability of the system in complex traffic scenarios, providing reliable safety guarantee for automatic driving decision. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the present application, the following briefly introduces the drawings needed to be used in the embodiments. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0046] Figure 1 The figure is a flowchart of the vehicle collision risk pre-warning method of the autonomous road traffic system of the embodiment of the present application.

[0047] Figure 2 The figure is a two-vehicle trajectory diagram in the embodiment of the present application.

[0048] Figure 3 The figure is a two-vehicle position and collision detection area diagram in the embodiment of the present application.

[0049] Figure 4A schematic diagram for modeling a collision detection zone in an embodiment of the present application. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present application will be apparently and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0051] In order to make the above objectives, characteristics and advantages of the present application more apparent, obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0052] Embodiment one:

[0053] As shown in the figure, the present embodiment provides a self-driving road traffic system vehicle collision risk early warning method, which comprises the following steps: Figure 1 S1, detecting whether there is a target vehicle j meeting the early warning condition in the automatic driving process of the host vehicle i; when there is a target vehicle j meeting the early warning condition, obtaining the planning trajectory of the host vehicle i and the predicted trajectory of the target vehicle j, and initializing the time window parameter k=0 and setting the maximum time window parameter K (K≤N, N is the number of trajectory points contained in the trajectory).

[0054] In the present embodiment, the early warning condition refers to that the distance between the two vehicles is less than a preset safety threshold, the planning trajectory of the host vehicle i is the expected driving path of the vehicle generated by the automatic driving algorithm, and the predicted trajectory of the target vehicle j is the future driving path of the vehicle calculated based on the vehicle-mounted or roadside sensor data and through the existing trajectory prediction model (such as the trajectory prediction model of the automatic driving vehicle).

[0055] The planning trajectory and the predicted trajectory are both composed of a series of trajectory points sampled at a fixed time interval Δt, which are used to represent or predict the motion position and state of the corresponding vehicle at a future time. The planning trajectory and the predicted trajectory both contain N trajectory points, and the sampling time of the trajectory points with the same serial number is strictly consistent; each trajectory point contains the rectangular profile center point coordinates, the orientation angle, the rectangular profile length and the width of the vehicle in the global coordinate system. The global coordinate system is a common reference coordinate system for unifying the positions of the host vehicle i and the target vehicle j. The outer profile of the vehicle is represented by a rectangle.

[0056] S2, constructing a to-be-detected set under the time window parameter k; the to-be-detected set contains a plurality of trajectory point pairs; each trajectory point pair is composed of a host vehicle trajectory point in a planning trajectory and a target vehicle trajectory point in a predicted trajectory meeting a preset condition.

[0057] ​

[0058] In this embodiment, the preset condition is that the sequence number difference between the target vehicle trajectory point and the main vehicle trajectory point is k. For a positive integer m not greater than N, the main vehicle trajectory point p is selected from the planned trajectory. i,m If the predicted trajectory contains trajectory point p j,m+k and / or p j,m-k Then the trajectory point pair (p) i,m ,p j,m+k ) and / or (p i,m ,p j,m-k Add to the set to be detected P k .

[0059] Assume that both the main vehicle and the target vehicle have 5 points on their trajectories, and k = 2.

[0060] ① Take the trajectory point 1 of the main vehicle and select the corresponding trajectory point 3 of the target vehicle to form a trajectory point pair (1, 3).

[0061] ② Select trajectory point 2 of the main vehicle and corresponding trajectory point 4 of the target vehicle to form a trajectory point pair (2, 4).

[0062] ③ Select the main vehicle's trajectory point 3, and correspondingly select the target vehicle's trajectory points 1 and 5 to form trajectory point pairs (3, 1) and (3, 5).

[0063] ④ Select trajectory point 4 of the main vehicle and select trajectory point 2 of the target vehicle to form a trajectory point pair (4, 2).

[0064] ⑤ Select trajectory point 5 of the main vehicle and corresponding trajectory point 3 of the target vehicle to form a trajectory point pair (5, 3).

[0065] ⑥ Finally, when k=2, there are trajectory point pairs (1,3)(2,4)(3,5)(3,1)(4,2)(5,3) in the detection set. The sequence number difference of the trajectory points in all trajectory point pairs is 2, and all trajectory point pairs that meet the conditions are put into the detection set. In the next k, the detection set needs to be recalculated, and the method is the same as above.

[0066] S3. Construct a collision detection zone model and perform collision judgment on trajectory point pairs based on the collision detection zone model.

[0067] In this embodiment, the collision detection zone model includes a target vehicle collision detection zone model and a host vehicle collision detection zone model; the target vehicle collision detection zone model is used to determine whether the center point of the host vehicle i is located in the collision detection zone of the target vehicle j; the host vehicle collision detection zone model is used to determine whether the center point of the target vehicle j is located in the collision detection zone of the host vehicle i. Each collision detection zone is dedicated to collision detection of a specific two vehicles, and each collision detection zone is fixedly associated with one of the vehicles participating in the collision detection (hereinafter referred to as the associated vehicle); the collision detection zone is a rectangle, which is concentric and coaxial with the rectangular outline of the associated vehicle; the collision detection zone boundary determination method is that when any side of the rectangular outline of the associated vehicle is in contact with the corner point of the rectangular outline of the other vehicle, the center point of the other vehicle is located on the boundary of the collision detection zone; the collision detection zone length calculation method is that the length of the outline of the associated vehicle plus the projection amount of the outline of the other vehicle in the length direction of the associated vehicle; the collision detection zone width calculation method is that the width of the outline of the associated vehicle plus the projection amount of the outline of the other vehicle in the width direction of the associated vehicle. When calculating the range of the collision detection zone, the calculation is performed in the local coordinate system of the associated vehicle; the local coordinate system takes the center point of the outline of the associated vehicle as the origin, and the coordinate axes are parallel to the length and width of the rectangular outline thereof.

[0068] Further implementation is that the calculation process of the target vehicle collision detection zone model includes:

[0069] S31, establishing a local coordinate system with the center point of the target vehicle j as the origin The y-axis positive direction of which is consistent with the orientation angle θ j of the target vehicle j. The relative coordinates (x i|j , y i|j ) and the relative orientation angle θ i|j of the host vehicle i in the local coordinate system are obtained through coordinate transformation.

[0070] The calculation method of the relative coordinates includes:

[0071] x i|j = sinθ j (x i -x j )-cosθ j (y i -y j ); (1)

[0072] y i|j = cosθ j (x i -x j )+sinθ j (y i -y j ); (2)

[0073] In the formula, x iy i These are the x and y coordinates of the center point of the rectangular outline of the main vehicle i, respectively; x j y j These are the x and y coordinates of the center point of the rectangular outline of the target vehicle, respectively.

[0074] Methods for calculating relative orientation angles include:

[0075] θ i|j =θ i -θ j +90°; (3)

[0076] In the formula, θ i The heading angle of the main vehicle i.

[0077] S32. Obtain the collision detection zone of the target vehicle j based on the relative orientation angle.

[0078] The calculation method for the collision detection zone of target vehicle j includes:

[0079] D j =d j +l i |cosθ i|j |+d i |sinθ i|j |; (4)

[0080] L j =l j +l i |sinθ i|j |+d i |cosθ i|j |; (5)

[0081] In the formula, D j L j These are the width and length of the collision detection zone for the target vehicle j, respectively; i d i Let l be the length and width of the rectangular outline of the main vehicle i; j d j Let l be the length and width of the rectangular outline of the target vehicle j, respectively; i |cosθ i|j |+d i |sinθ i|j | represents the rectangular outline of the main vehicle i in the local coordinate system. The projected length on the x-axis; l i |sinθ i|j |+d i |cosθ i|j | represents the rectangular outline of the target vehicle j in the local coordinate system. The projected length on the y-axis.

[0082] S33, obtaining a first collision index of the center point of the host vehicle i and the collision detection area of the target vehicle j based on the relative coordinates, and determining whether the center point of the host vehicle i is in the collision detection area of the target vehicle j based on the first collision index.

[0083] The calculation method of the first collision index comprises:

[0084]

[0085] When , it indicates that the center point of the host vehicle i is in the collision detection area of the target vehicle j; when , it indicates that the center point of the host vehicle i is not in the collision detection area of the target vehicle j.

[0086] Similarly, when determining whether the center point of the target vehicle j is in the collision detection area of the host vehicle i, the steps are basically the same as S31-S33, and the second collision index of the center point of the target vehicle j and the collision detection area of the host vehicle i can be calculated according to the formulas (1)-(6).

[0087] Only when the center points of the two vehicles are in the collision detection area of the other vehicle, that is, when , it is determined that the trajectory point pair exists collision.

[0088] After generating the to-be-detected set, the collision judgment is applied to each group of trajectory point pairs in it once, and as long as any one of the trajectory point pairs collides, the pre-warning information output of S4 will be triggered. When the pre-warning information is output, the earliest one of the trajectory point pairs is considered for the host vehicle among all the colliding trajectory point pairs.

[0089] S4, when the trajectory point pair exists collision, output the pre-warning information and end the loop; when the trajectory point pair does not exist collision, output the result that no pre-warning is needed; update the time window parameter k and return to S2.

[0090] If P k detects at least one colliding trajectory point pair, output the pre-warning information; the pre-warning information comprises a dangerous time window and a pre-warning collision time.

[0091] The dangerous time window is the current time window in which the collision is detected, comprising:

[0092] T1=k×Δt. (7)

[0093] The pre-warning collision time is the time difference of the earliest time of the host vehicle trajectory point in all the colliding trajectory point pairs relative to the current time, comprising:

[0094] T2=m′×Δt; (8)

[0095] In the formula, m' represents the minimum serial number of the host vehicle trajectory point in all the collision trajectory point pairs in the to-be-detected set; and Δt represents a fixed time interval.

[0096] If P k is not detected in the trajectory point pair, and the current k < K, then k: = k + 1 and S2-S4 are re-executed, otherwise a no-warning result is output and the loop is ended.

[0097] S5, the result of S4 is fed back to the trajectory planning system of the host vehicle, a new decision cycle is determined based on the updated planning trajectory of the host vehicle and the predicted trajectory of the target vehicle, and whether to trigger S1 is re-evaluated.

[0098] Embodiment Two:

[0099] The present embodiment provides an application example for further illustrating the method proposed in the present application.

[0100] S1, data initialization: in the present embodiment, the host vehicle i will experience lane narrowing in front, at this time the target vehicle j located behind plans to overtake the host vehicle i at the lane narrowing place. When the distance between the two vehicles is less than a preset threshold, the host vehicle i starts a warning program, inputs the planning trajectory of itself and the predicted trajectory of the target vehicle j, and the trajectories of the two vehicles are as shown in FIG. 1, while initializing the time window parameter k = 0. Figure 2

[0101] In the formula, the planning trajectory and the predicted trajectory each contain N = 50 continuous trajectory points, the trajectory point sampling interval Δt = 0.1 seconds; each trajectory point contains the rectangular profile center point coordinates, the heading angle, the rectangular profile length and the width in the global coordinate system; the number and the time sequence of the trajectory points of the two vehicles are strictly aligned, that is, for any integer m ∈ [1, 50], the mth trajectory point of the two trajectories has the same sampling time.

[0102] S2, establishing a to-be-detected set: for all positive integers m satisfying m ≤ 50, the trajectory point p i,m is selected from the trajectory of the host vehicle i, if there are trajectory points p j,m+k and / or p j,m-k in the trajectory of the target vehicle j, the corresponding trajectory point pair is added to the to-be-detected set P k , and finally P k = {(p i,1 , p j,1+k ), (p i,1+k , p j,1 ), (p i,2 , p j,2+k ), (p i,2+k , p j,2 ), …, (p i,50-k , p j,50 ), (p i,50 ​,p j,50-k )}, where the special case is P0={(p i,1 ,p j,1 ),(p i,2 ,p j,2 ),...,(p i,50 ,p j,50 )}.

[0103] It should be noted that in practical applications, P can be adjusted according to the specific scenario. k To simplify, for example in a typical car-following scenario, when the lead vehicle i is always behind the target vehicle j, kinematic analysis can determine all (p i,m ,p j,m+k If there is no possibility of collision, then the corresponding trajectory point pair does not need to be added to P. k .

[0104] S3, Collision Detection: For P k For each pair of trajectory points, it is calculated whether the center point of the main vehicle i is located within the collision detection zone of the target vehicle j, and whether the center point of the target vehicle j is located within the collision detection zone of the main vehicle i; the pair of trajectory points is determined to have collided if and only if the center points of both vehicles are located within the collision detection zone of the other. For a detailed implementation of S3, please refer to Example 1.

[0105] S4. Loop Control and Early Warning Output: In this embodiment, when the time window parameter k∈{0,1,...,5}, the corresponding P k No collisions were detected for any of the trajectory point pairs, so k:=k+1 and jump to S2; when k=6, a collision was detected (p i,42 ,p j,36 ), (p i,43 ,p j,37 ), (p i,44 ,p j,38 A series of trajectory point pairs collide, so the output danger time window is 6×0.1=0.6 seconds, the warning collision time is 42×0.1=4.2 seconds, and the loop ends.

[0106] S5. Feedback Execution: The dangerous time window and warning collision time output in step S4 are fed back to the master vehicle trajectory planning system in real time. Based on this warning information, master vehicle i dynamically optimizes its trajectory planning scheme and simultaneously updates the trajectory prediction results for surrounding vehicles. In the next decision cycle, when the distance between the two vehicles is continuously detected to be lower than the preset safety threshold, the system automatically triggers a new round of warning procedures.

[0107] Example 3:

[0108] This embodiment uses the collision detection of two vehicles at a certain trajectory point on their respective vehicles as an example to illustrate the specific execution process of the collision detection procedure S3. All length data in this embodiment are in meters. Figure 3 The positions of the main vehicle i and the target vehicle j on their respective trajectory points are as follows: Center point O of the rectangular outline of the main vehicle i i =(x i ,y i = (39.5, 65.2), Orientation angle θ i =23°, rectangular outline length l i =4.1 and width d i =1.8; Center point O of the rectangular outline of the target vehicle j j =(x j ,y j = (38.7, 61.4), Orientation angle θ j =53°, rectangular outline length l j =4.4 and width d j =1.9.

[0109] (1) Target vehicle collision detection zone model: Determine the center point O of the main vehicle i. i Is it located within the collision detection zone of the target vehicle j (denoted as R)? j The specific steps are as follows:

[0110] a) Take O j Establish a local coordinate system for the origin. Its positive y-axis direction is parallel to θ. j Consistent, obviously The y-axis is parallel to the length of the rectangular outline of the target vehicle j, and the x-axis is parallel to the width of the rectangular outline; the coordinate transformation is used to calculate the position of the main vehicle i. The relative coordinates O in i|j =(x i|j ,y i|j ) and relative orientation angle θ i|j In this embodiment, O can be calculated. i|j =(x i|j ,y i|j )=(-1.65,3.52), θ i|j =60°.

[0111] b) Calculate the collision detection zone R j The dimensions can be determined using the following principles: For example... Figure 4 As shown, translate the outline of the main vehicle i so that the corner of its rectangular outline just touches any side of the rectangular outline of the target vehicle j. At this time, O i It must lie on a straight line segment parallel to the contact edge, and the parallel extension of this line segment is R. ja boundary line. Since the collision detection region has central symmetry, only the position of one boundary line in the length direction and one boundary line in the width direction is needed to complete the region definition. In this embodiment, the calculation gives D j = 5.51, L j = 10.45.

[0112] c) Calculate the first collision index of the center point of the host vehicle i and the collision detection region of the target vehicle j In this embodiment, the calculation gives , which means that O i is located outside R j .

[0113] (2) Host vehicle collision detection region model: Determine whether the center point O j of the target vehicle j is located inside the collision detection region (denoted as R i ) of the host vehicle i. The specific steps are basically the same as above. In this embodiment, the calculation gives O j|i = (x j|i , y j|i ) = (3.19, -2.22), θ i|j = 120°, D j = 5.74, L j = 10.48, and finally the second collision index , which means that O j is located inside R i .

[0114] (3) Collision determination condition: Since is not satisfied, it is determined that there is no collision between the host vehicle i and the target vehicle j.

[0115] It should be noted that in actual applications, if is calculated, the subsequent steps can be skipped, and a non-collision result can be directly output according to the collision determination condition.

[0116] The above-described embodiments only describe the preferred modes of the present application and do not limit the scope of the present application. Without departing from the design spirit of the present application, various modifications and improvements to the technical solutions of the present application made by those of ordinary skill in the art should fall within the protection scope of the present application as defined by the claims.

Claims

1. A method for early warning of vehicle conflict risk in an autonomous road traffic system, characterized in that, Includes the following steps: S1. Detect whether there is a target vehicle that meets the warning conditions during the automatic driving of the master vehicle; when there is a target vehicle that meets the warning conditions, obtain the planned trajectory of the master vehicle and the predicted trajectory of the target vehicle, and initialize the time window parameters. ; S2, Build on time window parameters The set to be detected is defined as follows: the set to be detected includes several pairs of trajectory points; each pair of trajectory points consists of a main vehicle trajectory point in a planned trajectory and a target vehicle trajectory point in a predicted trajectory that meets preset conditions. S3. Construct a collision detection zone model, and perform collision judgment on all trajectory point pairs in the set to be detected based on the collision detection zone model. S4. When there are colliding trajectory point pairs, output a warning message; when there are no colliding trajectory point pairs, output a result indicating no warning is needed; regarding the time window parameters... Perform an update and return to S2; S5. Feed the results of S4 back to the trajectory planning system of the master vehicle, and make a new decision cycle judgment based on the updated planned trajectory of the master vehicle and the predicted trajectory of the target vehicle. The collision detection zone model includes a target vehicle collision detection zone model and a main vehicle collision detection zone model; The target vehicle collision detection zone model is used to determine whether the center point of the main vehicle is located within the target vehicle's collision detection zone; The main vehicle collision detection zone model is used to determine whether the center point of the target vehicle is located within the main vehicle's collision detection zone; The methods for determining whether a collision has occurred in S3 include: The center point of the main vehicle is located within the collision detection zone of the target vehicle, and the center point of the target vehicle is located within the collision detection zone of the main vehicle. The calculation process of the target vehicle collision detection zone model includes: S31. Establish a local coordinate system with the center point of the target vehicle as the origin. The coordinate transformation is used to obtain the main vehicle in the local coordinate system. The relative coordinates and relative orientation angles in the data; S32. Obtain the collision detection zone of the target vehicle based on the relative orientation angle; S33. Obtain a first collision index based on the relative coordinates, and determine whether the center point of the main vehicle is within the collision detection area of ​​the target vehicle based on the first collision index.

2. The method for early warning of vehicle conflict risk in an autonomous road traffic system according to claim 1, characterized in that, In S1, both the planned trajectory and the predicted trajectory include There are 10 trajectory points, and trajectory points with the same sequence number are sampled at the same time. Each trajectory point includes the coordinates of the center point of the rectangular contour of the vehicle in the global coordinate system, the orientation angle, the length and width of the rectangular contour.

3. The method for early warning of vehicle conflict risk in an autonomous road traffic system according to claim 1, characterized in that, The method for calculating the relative coordinates includes: ； ； In the formula, The relative coordinates of the main vehicle; , These are the x and y coordinates of the center point of the rectangular outline of the main vehicle, respectively. , These are the x and y coordinates of the center point of the rectangular outline of the target vehicle, respectively. The heading angle of the target vehicle; The method for calculating the relative orientation angle includes: ； In the formula, The heading angle of the main vehicle.

4. The method for early warning of vehicle conflict risk in an autonomous road traffic system according to claim 3, characterized in that, The methods for calculating the collision detection zone of the target vehicle include: ； ； In the formula, , These are the width and length of the collision detection zone for the target vehicle, respectively. , These are the length and width of the rectangular outline of the main vehicle, respectively. , These are the length and width of the rectangular outline of the target vehicle, respectively. The rectangular outline of the main vehicle in the local coordinate system of Projected length on the axis; The rectangular outline of the target vehicle in the local coordinate system of The projected length on the axis.

5. The method for early warning of vehicle conflict risk in an autonomous road traffic system according to claim 4, characterized in that, The calculation method for the first collision index includes: ； when When, it indicates that the center point of the main vehicle is within the collision detection zone of the target vehicle; when When the center point of the main vehicle is not within the collision detection zone of the target vehicle, it indicates that the center point of the main vehicle is not within the collision detection zone of the target vehicle.

6. The method for early warning of vehicle conflict risk in an autonomous road traffic system according to claim 1, characterized in that, In S4, the warning information includes a danger time window and a warning collision time; The dangerous time window includes: ； The collision warning time includes: ； In the formula, This represents the minimum index of the main vehicle trajectory point among all collision trajectory point pairs in the set to be detected; Indicates a fixed time interval.

Images