Real-time escape decision-making method and system for autonomous vehicles based on active safety

Abstract

This invention provides a real-time escape decision-making method and system for autonomous vehicles based on active safety. It constructs a real-time collision risk assessment model based on the collision time and safety escape margin of the two vehicles in front and behind. The model determines in real-time whether there is a collision risk between the current autonomous vehicle (AEV) and the rearmost vehicle (RCV) in the current lane. Based on the collision risk assessment model, it performs either longitudinal escape or lane-changing escape. The longitudinal acceleration escape strategy is prioritized; if this is not met, an emergency lane-changing escape mode is activated, and the decision-making system is continuously optimized in real-time at regular time intervals.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving control technology, and specifically to a real-time escape decision-making method and system for autonomous vehicles based on active safety. Background Technology

[0002] With the rapid iteration of intelligent driving technology and the acceleration of industrialization, autonomous vehicles (AEVs) are gradually moving from closed testing grounds to open road environments.

[0003] While AEVs possess technological advantages such as multimodal environmental perception, collaborative decision-making algorithms, and drive-by-wire execution architecture, their dynamic risk avoidance capabilities in normal driving scenarios still face severe challenges when dealing with complex urban road conditions. In particular, the challenges posed by dangerous rear-end traffic participants in traffic flow have not been fully resolved. On one hand, the behavioral uncertainty of heterogeneous traffic participants leads to a 17.3% misjudgment rate in traditional collision prediction models. Furthermore, dangerous behaviors from rear-end traffic participants are not taken into account. When rear-end traffic participants pose a significant danger, the damage to the AEV and the entire traffic environment can be incalculable.

[0004] Therefore, there is an urgent need to build an intelligent risk avoidance system that integrates real-time risk situation assessment and multi-objective optimization control. By establishing a real-time collision risk judgment model and an AEV basic escape model, the active safety defense capability of AEV in normal driving scenarios can be upgraded. This has become a key technical path to break through the bottleneck of intelligent driving commercialization. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a real-time escape decision-making method for autonomous vehicles based on active safety, comprising the following steps:

[0006] Step S1: Construct a real-time collision risk assessment model based on the collision time and safety escape margin of the two vehicles in front and behind;

[0007] Step S2: The collision risk real-time determination model determines in real time whether there is a collision risk between the current autonomous vehicle AEV and the RCV in the current lane. If there is a collision risk, proceed to step S3.

[0008] Step S3: Based on the real-time collision risk determination model, calculate the safe intrusion acceleration range of the AEV relative to the FV in the current lane and the safe escape acceleration range of the AEV relative to the RCV. If the two ranges intersect, generate a linear longitudinal escape acceleration function based on the maximum speed limit and acceleration range, and drive the AEV to accelerate and escape according to the linear longitudinal escape acceleration function; otherwise, proceed to step S4.

[0009] Step S4: Based on the real-time collision risk determination model, calculate the collision time and safety escape margin of the vehicles in front and behind the target lane to determine the feasibility of lane changing. If the safety conditions are met, generate a linear lateral escape acceleration function based on the lane width and the initial lateral velocity, and drive the AEV to implement lane changing.

[0010] Preferably, the expression for the real-time collision risk determination model is:

[0011]

[0012] In the formula, TTC is the time of collision between the rear vehicle and the front vehicle; T safe The maximum time required for an autonomous vehicle to react under normal conditions; SEM represents the safe escape margin between vehicles behind and in front; D rv-fv It indicates the real-time distance between the vehicle behind and the vehicle in front during longitudinal travel.

[0013] Preferably, the calculation expression for the time to collision (TTC) between a rear vehicle and a front vehicle is:

[0014]

[0015] In the formula, and These represent the distances traveled by the vehicle behind and the vehicle in front, respectively, from their current travels until a collision occurs; L fv v is the actual length of the vehicle in front; rv and v fv These are the real-time speeds of the vehicle behind and the vehicle in front, respectively; a in This accelerates the real-time intrusion of vehicles behind.

[0016] Preferably, the expression for calculating the safe escape margin between the rear vehicle and the front vehicle is:

[0017]

[0018] In the formula, and These represent the distances traveled by the vehicle behind and the vehicle in front within the maximum reaction time of the autonomous vehicle; T safe D is the maximum time required for an autonomous vehicle to react under normal conditions. safe This refers to the maximum safe following distance between two vehicles.

[0019] Preferably, in step S3, the linear longitudinal escape acceleration function is a piecewise linear function, and its expression is:

[0020]

[0021] In the formula, a AEV(y) (t) represents the linear longitudinal escape acceleration of the current autonomous vehicle (AEV) over time; T represents the acceleration from the start of the longitudinal escape acceleration behavior until the vehicle speed reaches the maximum speed limit v. max Time spent; j y a is the rate of change of acceleration when the AEV adopts longitudinal escape acceleration behavior; (AEV)min This represents the minimum longitudinal escape acceleration contained within the acceleration range; a (AEV)max This represents the maximum longitudinal escape acceleration contained within the acceleration range; v AEV This indicates the real-time speed of the AEV.

[0022] Preferably, in step S4, the expression for the linear lateral escape acceleration function is:

[0023] a AEV(x) (t)=a AEV(x) (0)+j x t, t∈[0,TTC1];

[0024]

[0025] In the formula, a AEV(x) (t) represents the linear lateral escape acceleration of the current autonomous vehicle (AEV) over time; j x a is the rate of change of acceleration when the AEV adopts lateral escape acceleration behavior; AEV(x) (0) represents the initial lateral escape acceleration of the AEV; D represents the current lane width; TTC1 represents the collision time.

[0026] The present invention also provides a real-time escape decision system for autonomous vehicles based on active safety, including a collision risk assessment module, a longitudinal acceleration escape module, and an emergency lane change escape module;

[0027] The collision risk determination module is used to calculate the collision time (TTC) and safety escape margin (SEM) between the autonomous vehicle (AEV) and the rear-end collision vehicle (RCV), and outputs a collision risk trigger signal when TTC ≤ maximum system reaction time and actual front-end distance ≤ SEM.

[0028] The longitudinal acceleration escape module is used to calculate the safe intrusion acceleration range of the AEV relative to the FV of the vehicle in front in the current lane and the safe escape acceleration range of the AEV relative to the RCV after receiving the collision risk trigger signal. If the two ranges intersect, a linear longitudinal escape acceleration function based on the maximum speed limit and acceleration range is generated, and the AEV is driven to accelerate according to the function. Otherwise, the collision risk trigger signal is transmitted to the emergency lane change escape module.

[0029] The emergency lane change escape module calculates lane change feasibility based on TTC and SEM between the AEV and the vehicles in front and behind in the target lane. If the safety conditions are met, it generates a linear lateral escape acceleration function based on the lane width and initial lateral velocity, and drives the AEV to perform the lane change.

[0030] Preferably, in the collision risk assessment module, TTC is calculated according to the following formula:

[0031]

[0032] In the formula, and These represent the distances traveled by the vehicle behind and the vehicle in front, respectively, from their current travels until a collision occurs; L fv v is the actual length of the vehicle in front; rv and v fv These are the real-time speeds of the vehicle behind and the vehicle in front, respectively; a in This accelerates the real-time intrusion of vehicles behind.

[0033] Preferably, in the collision risk assessment module, the SEM is calculated according to the following formula:

[0034]

[0035] In the formula, and These represent the distances traveled by the vehicle behind and the vehicle in front within the maximum reaction time of the autonomous vehicle; T safe D is the maximum time required for an autonomous vehicle to react under normal conditions. safe This refers to the maximum safe following distance between two vehicles.

[0036] Preferably, in the longitudinal acceleration escape module, the linear longitudinal escape acceleration function is a piecewise linear function, and its expression is:

[0037]

[0038] In the formula, a AEV(y) (t) represents the linear longitudinal escape acceleration of the current autonomous vehicle (AEV) over time; T represents the acceleration from the start of the longitudinal escape acceleration behavior until the vehicle speed reaches the maximum speed limit v. max Time spent; j y a is the rate of change of acceleration when the AEV adopts longitudinal escape acceleration behavior; (AEV)min This represents the minimum longitudinal escape acceleration contained within the acceleration range; a (AEV)max This represents the maximum longitudinal escape acceleration contained within the acceleration range; v AEV This indicates the real-time speed of the AEV.

[0039] The beneficial effects of this invention include at least the following: This invention proposes a real-time escape decision-making system for autonomous vehicles based on active safety. By constructing a closed-loop control architecture of "perception-decision-execution," it enables the autonomous vehicle (AEV) to dynamically respond to rear-end vehicles (RCVs). The system uses time parameter T as the time reference unit to establish a real-time collision risk assessment model, determining whether the RCV poses a risk of colliding with the AEV. When a potential collision risk is detected, the system employs a two-layer progressive escape strategy optimization mechanism: First, in the longitudinal control dimension, by establishing acceleration constraint equations, the optimal longitudinal acceleration solution that satisfies the safety conditions is solved; then, from the lateral control dimension, based on a comfortable lane-changing mode, the optimal lateral acceleration solution for the AEV is calculated. The system prioritizes the execution of the longitudinal acceleration escape strategy; when this is not met, an emergency lane-changing escape mode is activated, and the decision-making system is continuously optimized in real-time at certain time intervals. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention;

[0041] Figure 2 This is a schematic diagram of vehicle classification markings according to an embodiment of the present invention;

[0042] Figure 3 This is a schematic diagram of a collision risk determination scenario for a real-time collision risk determination model according to an embodiment of the present invention.

[0043] Figure 4 This is a schematic diagram of a collision risk assessment scenario for autonomous vehicles according to an embodiment of the present invention;

[0044] Figure 5 This is a schematic diagram illustrating the intrusion acceleration solution scenario according to an embodiment of the present invention;

[0045] Figure 6 This is a schematic diagram of the escape acceleration solution scenario according to an embodiment of the present invention;

[0046] Figure 7 This is a schematic diagram of the longitudinal escape acceleration function of the AEV according to an embodiment of the present invention;

[0047] Figure 8 This is a schematic diagram of an AEV emergency lane change scenario according to an embodiment of the present invention;

[0048] Figure 9 This is a graph of the lateral escape acceleration function of the AEV according to an embodiment of the present invention; Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0050] like Figure 1 As shown, this embodiment of the invention provides a real-time escape decision-making method for autonomous vehicles based on active safety, including the following steps:

[0051] Step S1: Construct a real-time collision risk assessment model based on the collision time and safety escape margin of the two vehicles in front and behind.

[0052] This invention only studies the normal driving process of vehicles, and classifies vehicles under this heterogeneous traffic flow as follows: Figure 2 As shown.

[0053] AEV stands for Automatic Escape Vehicle, which is the autonomous vehicle researched and invented.

[0054] RCV stands for Rear Collision Vehicle, which is the dangerous vehicle referred to in the invention.

[0055] FV refers to the vehicle ahead of the vehicle on the road in which this vehicle is traveling, as described in the invention.

[0056] ALFV, which refers to the vehicle in front of the target lane when the present invention performs a lane-changing operation;

[0057] ALRV refers to the vehicle behind the target lane when the present invention performs a lane-changing operation.

[0058] AEVs can obtain real-time vehicle trajectory information, such as distance to other vehicles in the road network, vehicle speed, and vehicle acceleration, through roadside equipment deployed in the road network. Unless otherwise specified, distance is in meters (m); v refers to vehicle speed in m / s; time is in seconds (s); and acceleration is in m / s². 2 .

[0059] During the normal operation of an autonomous vehicle, if a dangerous vehicle rapidly approaches from behind, posing a rear-end collision risk, the autonomous vehicle must actively avoid the danger by accelerating or changing lanes to ensure its own safety. Such actions are collectively referred to as escape behavior. However, relying solely on qualitative risk perception is insufficient to accurately trigger escape decisions. Therefore, a real-time collision risk assessment model needs to be constructed from two dimensions: temporal and spatial risk. This model quantitatively analyzes the dynamic interaction between the rear and front vehicles in a collision scenario involving an autonomous vehicle, calculates the risk in real time, and determines whether the current scenario meets the critical conditions for triggering an escape decision. Collision risk assessment scenarios include... Figure 3 As shown.

[0060] Based on the collision risk assessment scenario and the quantification of this collision risk from both temporal and spatial perspectives, the temporal aspect considers the collision time between the rear vehicle (rv) and the front vehicle (fv), i.e., when the collision occurs when the two vehicles are currently traveling in their respective states; the spatial aspect considers the safe escape margin between rv and fv, i.e., the actual head-to-head distance between the two vehicles before fv reacts and while traveling in their respective states, still maintaining the minimum safe distance. A collision risk assessment model is established based on this, and the assessment formula is shown in Equation (1).

[0061]

[0062] In the formula: TTC is the collision time between vehicles rv and fv, in seconds; T safe The maximum time required for an autonomous vehicle to react under normal conditions, measured in seconds; SEM is the safety escape margin between rv and fv, measured in meters; D rv-fv This represents the real-time actual distance between rv and fv during longitudinal travel, in meters.

[0063] From a time perspective, since autonomous vehicles need a certain amount of time to react to decision-making behavior, in order to prevent the fv from being collided with the rv in scenarios with collision risk, it is necessary to calculate from a time perspective when the front of the rv and the rear of the fv collide on the longitudinal path of the vehicle and the following vehicle. That is, according to equation (1), TTC needs to be calculated in order to determine the collision risk and establish a basic collision time calculation model.

[0064] Among them, the headway of rv and fv traveling within the TTC at the time of determination is given by equations (2) and (3), and the relationship between the difference in the distance traveled by rv and fv and the actual headway of the two vehicles at the time of determination is given by equation (4).

[0065]

[0066] In equations (2) to (4): L represents the distance traveled by rv and fv in their current vehicle states until the two vehicles collide, in meters (m). fv fv is the actual vehicle length, in meters (m); v rv v fv The real-time actual vehicle speeds of rv and fv are respectively, in m / s; a in For RV's real-time intrusion acceleration, m / s 2 The TTC solution value here is the minimum positive solution value.

[0067] According to equation (1), the safety escape margin (SEM) needs to be calculated from a spatial perspective, thus requiring the establishment of a safety escape margin calculation model. The safety escape margin considers that both rv and fv will travel a certain distance during the process of fv reacting to the decision-making behavior. To prevent collisions between the two vehicles during this process, it is necessary to calculate the initial headway between rv and fv that allows them to safely travel within the maximum reaction time of the autonomous vehicle under the current driving state, provided that a minimum safe following distance is reserved. This is how the safety escape margin calculation model is established.

[0068] The safety escape margin model should first calculate the distance traveled by rv and fv within the maximum reaction time of the autonomous vehicle. The distance calculation formulas are (5) and (6).

[0069]

[0070] In equations (5) and (6): t represents the distance traveled by rv and fv within the maximum reaction time of the autonomous vehicle, in meters (m); safe The maximum time required for the AEV to make a system response under normal conditions, expressed in seconds.

[0071] Based on the physical motion model, a formula for calculating the safety escape margin is established, which is formula (7).

[0072]

[0073] In the formula: D safe The maximum safe following distance between two vehicles is the distance between the front of the following vehicle and the rear of the preceding vehicle, expressed in meters (m).

[0074] Step S2: The collision risk real-time judgment model determines in real time whether there is a collision risk between the current autonomous vehicle (AEV) and the RCV (Rear Vehicle) in the current lane. If there is a collision risk, proceed to step S3.

[0075] Specifically, during the normal operation of an autonomous vehicle (AEV), if a dangerous vehicle (RCV) appears behind, a real-time collision risk assessment model is required to determine the risk. The collision risk assessment scenario between the RCV and the AEV is described in [link to relevant documentation]. Figure 4 .

[0076] The collision time between the RCV and AEV is calculated using equations (8) to (10).

[0077]

[0078] In the formula: and These are the distances traveled by the RCV and AEV in their current vehicle states until a collision occurs, respectively, in meters (m). AEV The actual length of the AEV is in meters (m); v RCV and v AEV These are the real-time actual vehicle speeds of the RCV and AEV, respectively, in m / s; a in Real-time intrusion acceleration of RCV, in m / s² 2 Here, the solution value of TTC1 is taken as its minimum positive solution value.

[0079] The safety escape margins of RCV and AEV are calculated using equations (11) to (13).

[0080]

[0081] In the formula: and T represents the distance traveled by the RCV and AEV within the maximum reaction time of the AEV, respectively, in meters (m); safe D is the maximum time required for the AEV to make a system response under normal conditions, expressed in seconds. safe The maximum safe following distance between two vehicles is the distance between the front of the following vehicle and the rear of the preceding vehicle, expressed in meters (m).

[0082] Based on the collision risk assessment model, the current collision risk between the RCV and AEV is assessed, and the assessment formula is shown in formula (14).

[0083]

[0084] In the formula: TTC1 is the collision time between the RCV and the AEV, in seconds; T safe SEM1 is the maximum time required for the AEV to react under normal conditions, in seconds; SEM1 is the safety escape margin between the RCV and the AEV, in meters; D1 is the real-time actual distance between the RCV and the AEV during longitudinal driving, in meters. When a risk is determined to exist, proceed to step S3.

[0085] Step S3: Based on the real-time collision risk judgment model, calculate the safe intrusion acceleration range of the AEV relative to the FV in the current lane and the safe escape acceleration range of the AEV relative to the RCV. If the two ranges intersect, generate a linear longitudinal escape acceleration function based on the maximum speed limit and acceleration range, and drive the AEV to accelerate and escape according to the linear longitudinal escape acceleration function; otherwise, proceed to step S4.

[0086] Specifically, when a collision risk is determined between the RCV and the AEV, the AEV needs to take relevant escape actions to ensure its own safety. Because the AEV has good responsiveness and a relatively fast reaction time during driving, and can perform various different escape actions, this step will classify and establish a basic escape model for the AEV. Based on the type of escape action, this model is divided into a longitudinal acceleration escape model and an emergency lane-change escape model.

[0087] This model is divided into two parts: a longitudinal acceleration feasibility assessment model and an AEV longitudinal escape acceleration calculation model; if the AEV speed has reached the maximum speed limit v that can be achieved on the current road. max AEV skips the longitudinal acceleration escape mode and enters the emergency lane change escape mode directly, that is, it skips the longitudinal acceleration escape model and enters the emergency lane change escape model.

[0088] Since the AEV needs to consider its relationship with both the FV and RCV during longitudinal acceleration escape behavior in order to make the correct escape action, it is necessary to discuss the intrusion acceleration range 'a' of the AEV relative to the FV under no-collision-risk conditions. in(AEV) And to determine the escape acceleration range 'a' of the AEV relative to the RCV under no-collision risk conditions. out(AEV) Only a correlation can determine whether an AEV can perform longitudinal acceleration escape behavior.

[0089] That is, if a in(AEV) and a out(AEV) If the calculated acceleration intervals intersect, then the AEV can use this intersection as the acceleration interval for longitudinal acceleration escape; if a in(AEV) and a out(AEV) If the calculated acceleration intervals do not overlap, it is determined that the AEV cannot perform longitudinal acceleration escape behavior, and the AEV then enters the emergency lane change escape process. The expression for determining the feasibility of longitudinal acceleration is Equation (15).

[0090]

[0091] In the formula: a in(AEV) With a out(AEV) The intrusion acceleration range of AEV relative to FV and the safe escape acceleration range of AEV relative to RCV are given in m / s².2 ;a (AEV)min a (AEV)max For a in(AEV) ∩a out(AEV) The subsequent acceleration range includes the minimum and maximum longitudinal escape acceleration of AEV, in m / s². 2 .

[0092] The method for solving the intrusion acceleration range based on the longitudinal relationship between AEV and FV is as follows. In this step, before executing the longitudinal acceleration escape mode, AEV needs to determine whether it can accelerate relative to FV. The scenario is as follows: Figure 5 As shown.

[0093] That is, assuming that AEV escapes with a specific intrusion acceleration relative to FV, in order to take into account the principle that AEV should not change the driving state of other vehicles when making relevant driving behavior changes, AEV and the preceding vehicle FV also need to meet the no-collision risk condition of the collision risk determination model with FV as the target vehicle during this escape process, and the condition is shown in Equation (16).

[0094] TTC2≥T safe Or D2≥SEM2 (16)

[0095] In the formula: TTC2 is the collision time between AEV and FV under this assumed state, in seconds; D2 is the real-time actual distance between AEV and FV during longitudinal travel, in meters; SEM2 is the safe escape margin between AEV and FV under this assumed state, in meters.

[0096] To satisfy equation (16), AEV will calculate its own acceptable acceleration range on the premise that the above conditions are met. The calculation of the acceleration range needs to start from the collision time between AEV and FV and the safe escape margin of AEV relative to FV.

[0097] First, we establish the relationship function between the intrusion acceleration of the AEV and the time of collision, starting from the collision time between the AEV and the FV. The initial calculation formula is referenced from formulas (2) to (4), and the results are formulas (17) to (19).

[0098]

[0099] In the formula: and TTC2 represents the distance traveled by AEV and FV in their current vehicle states until the collision occurs, in meters (m); TTC2 represents the time of collision between AEV and FV, in seconds (s); D2 represents the real-time actual distance traveled by AEV and FV during longitudinal travel, in meters (m); L FV These are the actual vehicle lengths of FV, in meters; vAEV v FV These are the real-time actual vehicle speeds of AEV and FV, respectively, in m / s; a 1in The real-time intrusion acceleration of the AEV under collision time conditions is expressed in m / s². 2 .

[0100] Because in equations (17) and (18) above, a 1in Both TTC2 and TTC2 are unknowns. An equation is established based on equation (19), where TTC2 is x and a... 1in Let y be the equation (20).

[0101]

[0102] Where: D2-L FV This expression is always greater than 0, and its unit is m; because v FV and v AEV The magnitude of the relationship depends on the actual driving scenario. Therefore, it is necessary to define different situations to determine the real-time intrusion acceleration of the AEV under the vehicle collision time condition. After differentiation and solving the feature points, the real-time intrusion acceleration range of the AEV under different states is obtained, as shown in equation (21).

[0103]

[0104] Then, starting with the safety escape margin of AEV relative to FV, establish a relationship between D2≥SEM2 and a 2in The inequality discriminant is used to establish the reference equations (5) to (7) for SEM2 calculation in the discriminant. The solution result of the inequality discriminant is equation (25).

[0105]

[0106] In the formula: and The distance traveled by the AEV and FV within the maximum reaction time of the AEV, in meters; a 2in Real-time intrusion acceleration under AEV safety escape margin conditions, in m / s² 2 .

[0107] When the AEV intrusion acceleration range 'a' that satisfies the collision time condition and the safety escape margin condition is determined... 1in and a 2in In order for the AEV to maintain a safe driving state with the FV during the acceleration escape process, i.e., satisfying equation (16), the actual intrusion acceleration value adopted by the AEV should simultaneously satisfy the acceleration range under the two conditions, thus having equations (26) and (27).

[0108] ain(AEV) ∈a 1in ∪a 2in (26)

[0109] a in(AEV) ∈[0,a in(AEV)max (27)

[0110] In equations (26) and (27): a in(AEV) To accelerate AEV intrusion, a in(AEV)max The maximum intrusion acceleration achievable under AEV intrusion conditions, in m / s². 2 .

[0111] At this point, according to equation (27), when considering the relationship between AEV and FV, a can be calculated. in(AEV) The corresponding maximum longitudinal escape acceleration will be determined next, and the relationship between RCV and AEV will be discussed next.

[0112] The escape acceleration range for the longitudinal relationship between RCV and AEV is as follows: When the AEV enters the longitudinal acceleration escape mode, the acceleration used must satisfy the requirement that the AEV achieves successful longitudinal acceleration escape before accelerating to the maximum vehicle speed on the current road. The scenario is as follows: Figure 6 As shown.

[0113] To achieve this goal, RCV and AEV should satisfy the following conditions, which are given by equation (28).

[0114] TTC3≥T safe Or D1≥SEM3 (28)

[0115] To satisfy equation (28), it is also necessary to calculate the longitudinal escape acceleration range of the AEV from the perspective of the collision time condition between the RCV and the AEV and the safety escape margin condition. First, the collision time between the RCV and the AEV, TTC3≥T, is used as an example. safe We first establish the escape acceleration versus time function of AEV during collision time. The initial calculation formulas are referenced from equations (2) to (4), and the results are equations (29) to (31).

[0116]

[0117] In the formula: and These represent the distances traveled by the RCV and AEV in their current vehicle states until the collision, respectively, in meters (m); TTC3 represents the collision time between the RCV and AEV, in seconds (s); a 1out The real-time escape acceleration of the AEV under collision time conditions is expressed in m / s². 2 .

[0118] Because in equations (29) and (30), a 1out Both TTC3 and TTC3 are unknowns. An equation is established based on equation (31), where TTC3 is x and a... 1out Let y be the equation (32).

[0119]

[0120] In the formula: (D1-L AEV This formula is always greater than 0, and the unit is m; from the previous condition, we know that AEV has already passed the collision risk assessment model, and it already satisfies the condition that TTC3 < T when AEV acceleration is 0. safe That is, when y = 0, the corresponding x < T safe Because of v AEV v RCV The magnitude relationship depends on the actual driving scenario, so the next step is to define different situations to determine the real-time escape acceleration of AEV under the vehicle collision time condition. After differentiation and solving the feature points, the real-time escape acceleration range of AEV under different states is obtained, as shown in equation (33).

[0121]

[0122] Then, starting with the safety escape margin of RCV relative to AEV, establish a relationship between D1≥SEM3 and a. 2out The inequality discriminant is used to establish the SEM2 calculation formulas (34) to (36) in the discriminant, and the reference formulas (5) to (7) are established. The solution result of the inequality discriminant is formula (37).

[0123]

[0124] In the formula: and This refers to the distance traveled by the RCV and AEV within the AEV's maximum reaction time, in meters (m). 2out For the real-time escape acceleration under the AEV safety escape margin condition, a max The maximum acceleration that an AEV can achieve while driving on normal roads is expressed in m / s². 2 .

[0125] When the AEV escape acceleration range 'a' that satisfies the collision time condition and the safe escape margin condition is determined... 1out and a 2out In order for the AEV to successfully achieve its escape purpose during the acceleration escape process, it must satisfy equation (28). At this time, the actual escape acceleration value adopted by the AEV should simultaneously satisfy the acceleration range under the two conditions, which gives equations (38) and (39).

[0126] a out(AEV)∈a 1out ∪a 2out (38)

[0127] a out(AEV) ∈[a out(AEV)min ,a out(AEV)max (39)

[0128] In equations (38) and (39): a out(AEV) For AEV escape acceleration, a out(AEV)min Let a be the minimum escape acceleration required under AEV escape conditions. out(AEV)max The maximum escape acceleration that can be achieved under AEV escape conditions, in m / s². 2 ;

[0129] Finally, the longitudinal escape acceleration of the AEV is calculated. Considering that using only a single acceleration during the acceleration phase would greatly reduce driver comfort, and as shown in equation (15), when [a] is calculated... (AEV)min ,a (AEV)max It will be found that AEVs can use various acceleration methods for longitudinal acceleration, and because the vehicle has a maximum speed limit (v) on the relevant roads. max Therefore, a linear longitudinal escape acceleration a can be established based on this. AEV(y) (t), which is a linear function of one variable that changes with time as the AEV begins to accelerate longitudinally. For this function, a in equation (15) (AEV)min Let a be its intercept value and minimum value, and let a in equation (15) be... (AEV)max Let it be the maximum value of the function, and its function expression is equation (40).

[0130]

[0131] In the formula: a AEV(y) (t) represents the linear longitudinal escape acceleration of the AEV over time, in m / s². 2 T represents the speed of the AEV from the moment it begins its longitudinal escape acceleration behavior until it reaches v. max The time spent, in seconds (s); j y The rate of change of acceleration when the AEV adopts longitudinal escape acceleration behavior, in m / s². 3 ; t is the initial time when the longitudinal escape acceleration begins, in seconds; the function graph is as follows. Figure 7 As shown.

[0132] Clearly, this AEV exhibits a linear longitudinal escape acceleration a that varies with time. AEV(y) The function expression (t) has two unknowns j. yAnd T, as mentioned above, because the AEV has a maximum speed v. max Limited, and when the AEV accelerates to At that time, a AEV(y) For a (AEV)max At this point, we can establish an equation to solve for j and T, and the solution is equation (41).

[0133]

[0134] In the formula: v max This represents the maximum speed that an AEV can achieve on the current road, expressed in m / s.

[0135] The equation (41) above is used to solve for j and T, and its expression is given by equation (42).

[0136]

[0137] Based on the above results, the accurate linear longitudinal escape acceleration a of AEV over time can be obtained. AEV(y) The function expression (t) allows the AEV to use this acceleration for longitudinal escape behavior. While maintaining this accelerated longitudinal driving state, the AEV updates relevant lane environment information in real time at certain time intervals and continuously calculates the new linear longitudinal escape acceleration a of the AEV over time. AEV(y) (t). If the AEV's speed reaches the maximum speed limit v max AEV then enters emergency lane change escape mode.

[0138] Step S4: Based on the real-time collision risk assessment model, calculate the collision time and safety escape margin of the vehicles in front and behind the target lane to determine the feasibility of lane changing. If the safety conditions are met, generate a linear lateral escape acceleration function based on the lane width and the initial lateral velocity, and drive the AEV to implement lane changing.

[0139] Specifically, when the AEV is determined or calculated to be unable to perform longitudinal acceleration escape behavior by the longitudinal acceleration escape model, the AEV enters the emergency lane change escape model. In this model, due to the short lane change time, the AEV, ALRV, and ALFV are all considered to be traveling at a constant speed in the longitudinal direction from the start of the lane change preparation to the final successful lane change. This model is also divided into an emergency lane change escape feasibility determination model and an AEV lateral escape acceleration calculation model.

[0140] Similarly, when the AEV enters the emergency lane change escape model, it is also necessary to determine the feasibility of the emergency lane change escape. This involves determining whether the TTC and SEM relationships between the AEV, ALRV, and ALFV are safe when the AEV successfully changes lanes to an adjacent lane in its current state. The determination scenario is as follows: Figure 8As shown.

[0141] In order to ensure safe driving of AEV from the start of lane change preparation to the successful lane change, the longitudinal spatiotemporal relationship between AEV and ALFV and the longitudinal spatiotemporal relationship between ALRV and AEV should meet the safe driving state without collision risk, so there is a judgment formula (43).

[0142]

[0143] In the formula: TTC4 is the collision time between AEV and ALFV, and TTC5 is the collision time between ALRV and AEV, in seconds; SEM4 is the safe escape margin between AEV and ALFV, and SEM5 is the safe escape margin between ALRV and AEV, in meters; D3 and D4 are the longitudinal distances between AEV and ALFV and ALRV in the current lane after AEV successfully changes lanes, in meters.

[0144] Since the AEV, ALRV, and ALFV are all considered to be traveling at a constant speed longitudinally from the start of the lane change preparation to the successful lane change, that is, after the AEV changes to the adjacent lane, the AEV, ALRV, and ALFV are still traveling at the speed before the judgment. Therefore, when making the judgment calculation, the speed at the current detection time can be used directly. In order to ensure that the AEV can safely complete the lane change, the time from the start of the lane change to the successful lane change should be equal to the collision time between the RCV and the AEV. First, the longitudinal position relative of the AEV after the successful lane change in the current state is calculated. The calculation results are shown in equations (44) and (45).

[0145] D3 = D ALFV -(v AEV -v ALFV TTC1 (44)

[0146] D4 = D ALRV -(v ALRV -v AEV TTC1 (45)

[0147] In equations (44) and (45): D3 and D4 are the longitudinal distances between the AEV and the ALFV and ALRV in the current lane after the AEV successfully changes lanes, in meters; v ALFV and v ALRV The speeds of the ALFV and ALRV that are changing lanes for the AEV in the target lane, in m / s.

[0148] For the longitudinal spatiotemporal relationship between AEV and ALFV, refer to equations (2) to (7) and there are spatiotemporal relationship equations (46) to (51).

[0149]

[0150] In the formula: and These are the distances traveled by the AEV and ALFV in their current vehicle states while traveling in the target lane of the AEV lane change until a collision occurs, respectively, in meters. and TTC4 represents the distance traveled by the AEV and ALFV within the AEV's maximum reaction time, in meters (m); TTC4 represents the collision time between the AEV and ALFV, in seconds (s); L ALFV The actual lengths of the ALFV are shown in meters (m); SEM4 represents the safety escape margin between the AEV and the ALFV, also in meters.

[0151] For the longitudinal spatiotemporal relationship between ALRV and AEV, refer to equations (2) to (7) for spatiotemporal relationship equations (52) to (57).

[0152]

[0153] In the formula: and These are the distances traveled by the ALRV and AEV in their current vehicle states while the AEV is in the target lane for lane changing until a collision occurs, respectively, in meters. and TTC5 is the distance traveled by the ALRV and AEV within the maximum reaction time of the AEV, in meters; TTC5 is the collision time between the ALRV and AEV, in seconds; SEM5 is the safe escape margin between the ALRV and AEV, in meters.

[0154] After the emergency lane change escape feasibility assessment model determines that the AEV can perform an emergency lane change escape, it is necessary to calculate the AEV's lateral escape acceleration to consider driver comfort during the lane change process. First, the lane change time of the AEV is determined. The emergency lane change escape feasibility assessment model states that the AEV should successfully change lanes when the collision time between the RCV and the AEV is reached. Next, the lane change distance of the AEV is determined; it should be equal to the current lane width. It is also obvious that the lateral velocity of the AEV should eventually return to zero. Based on this, a linear lateral escape acceleration α for the AEV is established. AEV(x) (t), which is a linear function in one variable that changes with time as the AEV begins to take lateral acceleration behavior, and its function expression is Equation (58).

[0155] a AEV(x) (t)=a AEV(x) (0)+j x t,t∈[0,TTC1] (58)

[0156] In the formula: a AEV(x)(t) represents the linear lateral escape acceleration of the AEV over time, in m / s². 2 j x The rate of change of acceleration when the AEV adopts lateral escape acceleration behavior, in m / s². 3 ;t is the initial time taken as the start of the lateral escape acceleration behavior, in seconds; a AEV(x) (0) represents the initial lateral escape acceleration of the AEV, in m / s². 2 The graph of the function is as follows: Figure 9 As shown, the positive acceleration direction is the lane-changing direction of the AEV.

[0157] Since the lane-changing distance of the AEV should be equal to the current lane width, and when t is TTC1, the lateral escape acceleration of the AEV is the negative of the initial lateral acceleration of the AEV, an equation can be established for j. x and a AEV(x) (0) Solve the equation, and the solution is equation (59).

[0158]

[0159] In the formula: d is the current lane width, in meters.

[0160] The equation above can be solved to obtain j. x and a AEV(x) (0), its expression is equation (60).

[0161]

[0162] Based on the above results, the accurate linear lateral escape acceleration a of AEV over time can be obtained. AEV(x) The function expression (t) allows the AEV to use lateral acceleration for emergency lane changing and escape until it successfully changes to the target lane. After a successful lane change, the AEV continues to update the vehicle information in the current lane in real time.

[0163] The present invention also provides a real-time escape decision system for autonomous vehicles based on active safety, including a collision risk assessment module, a longitudinal acceleration escape module, and an emergency lane change escape module;

[0164] The collision risk assessment module is used to calculate the collision time (TTC) and safety escape margin (SEM) between the autonomous vehicle (AEV) and the rear-end collision vehicle (RCV). When TTC ≤ the maximum system reaction time and the actual frontal distance ≤ SEM, it outputs a collision risk trigger signal.

[0165] The longitudinal acceleration escape module is used to calculate the safe intrusion acceleration range of the AEV relative to the FV of the vehicle in front in the current lane and the safe escape acceleration range of the AEV relative to the RCV after receiving the collision risk trigger signal. If the two ranges intersect, a linear longitudinal escape acceleration function based on the maximum speed limit and acceleration range is generated and the AEV is driven to accelerate according to the function. Otherwise, the collision risk trigger signal is transmitted to the emergency lane change escape module.

[0166] The emergency lane change escape module calculates lane change feasibility based on TTC and SEM between the AEV and the vehicles in front and behind in the target lane. If the safety conditions are met, it generates a linear lateral escape acceleration function based on lane width and initial lateral velocity, and drives the AEV to perform the lane change.

[0167] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described; only preferred embodiments of the present invention are illustrated. The descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. As long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0168] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this invention should be determined by the appended claims.

Claims

1. A real-time escape decision-making method for autonomous vehicles based on active safety, characterized in that: Includes the following steps: Step S1: Construct a real-time collision risk assessment model based on the collision time and safety escape margin of the two vehicles in front and behind; Step S2: The collision risk real-time determination model determines in real time whether there is a collision risk between the current autonomous vehicle AEV and the RCV in the current lane. If there is a collision risk, proceed to step S3. Step S3: Based on the real-time collision risk determination model, calculate the safe intrusion acceleration range of the AEV relative to the FV in the current lane and the safe escape acceleration range of the AEV relative to the RCV. If the two ranges intersect, generate a linear longitudinal escape acceleration function based on the maximum speed limit and acceleration range, and drive the AEV to accelerate and escape according to the linear longitudinal escape acceleration function; otherwise, proceed to step S4. Step S4: Based on the real-time collision risk judgment model, calculate the collision time and safety escape margin of the vehicles in front and behind the target lane to determine the feasibility of lane changing. If the safety conditions are met, generate a linear lateral escape acceleration function based on the lane width and the initial lateral velocity, and drive the AEV to implement lane changing. The expression for the real-time collision risk assessment model is: ； In the formula, The time of collision between the vehicle behind and the vehicle in front; This indicates the maximum time required for an autonomous vehicle to make a system response under normal conditions; To provide a safe escape margin between vehicles behind and vehicles in front; It indicates the real-time distance between the vehicle behind and the vehicle in front during longitudinal travel; Time of collision between the vehicle behind and the vehicle in front The calculation expression is: ； ； ； In the formula, and These represent the distances traveled by the vehicle behind and the vehicle in front, respectively, while maintaining their current vehicle states until a collision occurs. This refers to the actual length of the vehicle in front. and These are the real-time speeds of the vehicles behind and in front, respectively. Accelerates the real-time intrusion of vehicles behind; The formula for calculating the safe escape margin between the vehicle behind and the vehicle in front is: ； ； ； In the formula, and These represent the distances traveled by the vehicle behind and the vehicle in front within the maximum reaction time of the autonomous vehicle; The maximum time required for an autonomous vehicle to make a system response under normal conditions; This refers to the maximum safe following distance between two vehicles.

2. The real-time escape decision-making method for autonomous vehicles based on active safety according to claim 1, characterized in that: In step S3, the linear longitudinal escape acceleration function is a piecewise linear function, and its expression is: ； ； ； In the formula, The linear longitudinal escape acceleration of the current autonomous vehicle (AEV) varies with time. For AEVs, from the moment they begin longitudinal escape acceleration to when they reach the maximum speed limit The time spent; The rate of change of acceleration when the AEV adopts longitudinal escape acceleration behavior; This represents the minimum longitudinal escape acceleration contained within the acceleration range; This indicates the maximum longitudinal escape acceleration contained within the acceleration range; This indicates the real-time speed of the AEV.

3. The real-time escape decision-making method for autonomous vehicles based on active safety according to claim 1, characterized in that: In step S4, the expression for the linear lateral escape acceleration function is: ； ； ； In the formula, The linear lateral escape acceleration of the current autonomous vehicle (AEV) varies with time. The rate of change of acceleration when the AEV adopts lateral escape acceleration behavior; Let be the initial lateral escape acceleration of the AEV; The current lane width; Indicates the time of collision.

4. A real-time escape decision-making system for autonomous vehicles based on active safety, characterized in that: This includes a collision risk assessment module, a longitudinal acceleration escape module, and an emergency lane change escape module; The collision risk assessment module is used to calculate the collision time between the autonomous vehicle (AEV) and the rear-end collision vehicle (RCV). and safety escape margin , and when ≤System maximum response time and actual vehicle headway≤ At that time, output a collision risk trigger signal; The longitudinal acceleration escape module is used to calculate the safe intrusion acceleration range of the AEV relative to the FV of the vehicle in front in the current lane and the safe escape acceleration range of the AEV relative to the RCV after receiving the collision risk trigger signal. If the two ranges intersect, a linear longitudinal escape acceleration function based on the maximum speed limit and acceleration range is generated, and the AEV is driven to accelerate according to the function. Otherwise, the collision risk trigger signal is transmitted to the emergency lane change escape module. The emergency lane change escape module, based on the distance between the AEV and the vehicles in front and behind in the target lane... and The feasibility of lane changing is calculated. If the safety conditions are met, a linear lateral escape acceleration function based on lane width and initial lateral velocity is generated, and the AEV is driven to perform a lane change.

5. The real-time escape decision system for autonomous vehicles based on active safety according to claim 4, characterized in that: In the collision risk assessment module Calculate using the following formula: ； ； ； In the formula, and These represent the distances traveled by the vehicle behind and the vehicle in front, respectively, while maintaining their current vehicle states until a collision occurs. This refers to the actual length of the vehicle in front. and These are the real-time speeds of the vehicles behind and in front, respectively. This accelerates the real-time intrusion of vehicles behind.

6. The real-time escape decision system for autonomous vehicles based on active safety according to claim 4, characterized in that: In the collision risk assessment module Calculate using the following formula: ； ； ； In the formula, and These represent the distances traveled by the vehicle behind and the vehicle in front within the maximum reaction time of the autonomous vehicle; The maximum time required for an autonomous vehicle to make a system response under normal conditions; This refers to the maximum safe following distance between two vehicles.

7. A real-time escape decision system for autonomous vehicles based on active safety according to claim 4, characterized in that: In the longitudinal acceleration escape module, the linear longitudinal escape acceleration function is a piecewise linear function, and its expression is: ； ； ； In the formula, The linear longitudinal escape acceleration of the current autonomous vehicle (AEV) varies with time. For AEVs, from the moment they begin longitudinal escape acceleration to when they reach the maximum speed limit The time spent; The rate of change of acceleration when the AEV adopts longitudinal escape acceleration behavior; This represents the minimum longitudinal escape acceleration contained within the acceleration range; This indicates the maximum longitudinal escape acceleration contained within the acceleration range; This indicates the real-time speed of the AEV.

Images