A decision method for a two-vehicle simultaneous lane-changing scene and a related device

Abstract

The application discloses a decision-making method for a two-vehicle simultaneous lane-changing scene, and relates to the technical field of auxiliary driving. The method comprises the following steps: determining risk sub-scores of a preceding vehicle in multiple dimensions based on motion state data of the preceding vehicle; performing weighted summation on the risk sub-scores in the multiple dimensions to obtain a comprehensive risk score; determining a lane-changing decision-making cost based on the product of the relative state data of a host vehicle and the preceding vehicle and the comprehensive risk score; determining a decision-making cost for keeping the original lane based on the relative state data of the host vehicle and the preceding vehicle; and determining a target driving behavior of the host vehicle according to the size relationship between the lane-changing decision-making cost and the decision-making cost for keeping the original lane. The application evaluates the emergency U-turn intention of the preceding vehicle through multi-dimensional parameters, predicts the collision risk in advance in the case that the vision of the host vehicle is limited, avoids sudden accidents of the target lane caused by the preceding vehicle shielding, and improves the safety and reliability of the lane-changing decision-making.

Description

Technical Field

[0001] This application relates to the field of driver assistance technology, and in particular to a decision-making method and related equipment for a scenario where two vehicles change lanes simultaneously. Background Technology

[0002] In the field of autonomous driving or assisted driving, a common driving scenario is two vehicles simultaneously changing lanes from adjacent lanes to the same target lane. When both vehicles intend to change lanes at the same time, the vehicle in front may obstruct the driver's view of part of the target lane, preventing the driver from perceiving the traffic conditions within that lane. Existing technologies typically rely on the vehicle's sensors to perceive the surrounding environment in real time for trajectory planning or decision-making. However, in the aforementioned situation with limited visibility, if the vehicle in front suddenly turns back to its original lane due to an obstacle or other unforeseen event in the target lane, the driver, due to the obstruction, cannot obtain timely risk information about the target lane. If the driver continues driving according to the original lane-changing plan, it may only discover the emergency situation in the target lane after the vehicle in front has completely turned back, potentially leading to a collision. Therefore, improving the driver's ability to predict potential risks and make accurate decisions in scenarios where two vehicles are changing lanes simultaneously with limited visibility is a pressing technical problem that needs to be solved. Summary of the Invention

[0003] The summary section introduces a series of simplified concepts, which will be further explained in detail in the detailed description section. This summary section is not intended to limit the key and essential technical features of the claimed technical solutions, nor is it intended to determine the scope of protection of the claimed technical solutions.

[0004] Firstly, this application provides a decision-making method for a scenario where two vehicles change lanes simultaneously, including: Based on the motion state data of the preceding vehicle, a risk sub-score for the preceding vehicle is determined in multiple dimensions; The risk sub-scores under the multiple dimensions are weighted and summed to obtain the comprehensive risk score. The cost of lane change decision is determined by multiplying the relative state data of the vehicle and the vehicle in front by the comprehensive risk score. Based on the relative state data between the vehicle and the vehicle in front, the cost of maintaining the original lane is determined; The target driving behavior of the vehicle is determined based on the relationship between the lane change decision cost and the lane-keeping decision cost.

[0005] In some implementations, the multiple dimensions include a yaw rate dimension, a lateral position dimension, a longitudinal deceleration dimension, and a turn signal dimension.

[0006] In some implementations, the risk sub-score includes a yaw rate risk sub-score, a lateral position risk sub-score, a longitudinal deceleration risk sub-score, and a turn signal risk sub-score. Determining the risk sub-score of the preceding vehicle across multiple dimensions based on the preceding vehicle's motion state data includes: Based on the relationship between the yaw rate of the preceding vehicle and a preset yaw rate threshold, and the relationship between the rate of change of the yaw rate of the preceding vehicle and a preset rate of change threshold, the yaw rate risk sub-score is determined. Based on the relationship between the lateral speed of the preceding vehicle and a preset lateral speed threshold, and the relationship between the lateral distance of the preceding vehicle from the original lane centerline and a preset lateral position offset threshold, the lateral position risk sub-score is determined. Based on the relationship between the longitudinal deceleration of the preceding vehicle and a preset longitudinal deceleration threshold, and whether there is an obstacle in front of the preceding vehicle, the longitudinal deceleration risk sub-score is determined. A risk sub-score for the turn signal is determined based on whether the turn signal of the preceding vehicle changes state.

[0007] In some implementations, the preset yaw rate threshold includes a first yaw rate threshold and a second yaw rate threshold, and the preset rate of change threshold includes a first rate of change threshold and a second rate of change threshold. Determining the yaw rate risk sub-score based on the relationship between the yaw rate of the preceding vehicle and the preset yaw rate threshold, and the relationship between the rate of change of the yaw rate of the preceding vehicle and the preset rate of change threshold, includes: The absolute value of the yaw rate of the preceding vehicle is compared with the absolute values ​​of the first yaw rate threshold and the second yaw rate threshold. Based on the comparison result, a first normalized risk value is determined, wherein the absolute value of the first yaw rate threshold is less than the absolute value of the second yaw rate threshold. The absolute value of the rate of change of the yaw rate of the preceding vehicle is compared with the absolute values ​​of the first rate of change threshold and the second rate of change threshold. A second normalized risk value is determined based on the comparison result, wherein the absolute value of the first rate of change threshold is less than the absolute value of the second rate of change threshold. Calculate the arithmetic mean of the first normalized risk value and the second normalized risk value, and determine the arithmetic mean as the yaw rate risk sub-score.

[0008] In some implementations, determining the first normalized risk value based on the comparison result includes: When the absolute value of the yaw rate of the vehicle in front is less than or equal to the absolute value of the first yaw rate threshold, the first normalized risk value is determined to be 0. When the absolute value of the yaw rate of the vehicle in front is greater than or equal to the absolute value of the second yaw rate threshold, the first normalized risk value is determined to be 1. When the absolute value of the yaw rate of the preceding vehicle is greater than the absolute value of the first yaw rate threshold and less than the absolute value of the second yaw rate threshold, the yaw rate of the preceding vehicle is calculated by linear interpolation to obtain the first normalized risk value.

[0009] In some implementations, the preset lateral speed threshold includes a first lateral speed threshold and a second lateral speed threshold. The determination of the lateral position risk sub-score based on the relationship between the lateral speed of the preceding vehicle and the preset lateral speed threshold, and the relationship between the lateral distance of the preceding vehicle from the original lane centerline and the preset lateral position offset threshold, includes: The absolute value of the lateral speed of the preceding vehicle is compared with the absolute values ​​of the first lateral speed threshold and the second lateral speed threshold. A third normalized risk value is determined based on the comparison result, wherein the absolute value of the first lateral speed threshold is less than the absolute value of the second lateral speed threshold. The absolute value of the lateral distance between the preceding vehicle and the original lane centerline is compared with the magnitude of the preset lateral position offset threshold and half a preset lane width. A fourth normalized risk value is determined based on the comparison result, wherein the preset lateral position offset threshold is less than half a preset lane width. Calculate the arithmetic mean of the third normalized risk value and the fourth normalized risk value, and determine the arithmetic mean as the lateral position risk sub-score.

[0010] In some implementations, the preset longitudinal deceleration threshold includes a first longitudinal deceleration threshold and a second longitudinal deceleration threshold. The determination of the longitudinal deceleration risk sub-score based on the relationship between the longitudinal deceleration of the preceding vehicle and the preset longitudinal deceleration threshold, and whether there is an obstacle in front of the preceding vehicle, includes: The absolute value of the longitudinal deceleration of the preceding vehicle is compared with the absolute values ​​of the first longitudinal deceleration threshold and the second longitudinal deceleration threshold. A fifth normalized risk value is determined based on the comparison result, wherein the absolute value of the first longitudinal deceleration threshold is less than the absolute value of the second longitudinal deceleration threshold. Based on whether there is an obstacle in front of the vehicle, an obstacle influence factor is determined, wherein the value of the obstacle influence factor ranges from 0 to 1; Calculate the difference between the preset constant and the obstacle influence factor, and determine the difference as the difference factor; Calculate the product of the fifth normalized risk value and the difference factor, and determine the product as the longitudinal deceleration risk sub-score.

[0011] In some implementations, determining the turn signal risk sub-score based on whether the turn signal signal of the preceding vehicle has undergone a state change includes: Detect whether the turn signal of the vehicle in front changes from on to off, or from off to on, or from on one side to on the other. When a change in the state of the turn signal is detected, the risk sub-score of the turn signal is determined to be 1; When no change in the state of the turn signal is detected, the risk sub-score of the turn signal is determined to be 0.

[0012] In some implementations, the weighted summation of the risk sub-scores under the multiple dimensions to obtain a comprehensive risk score includes: The comprehensive risk score is determined by performing a weighted summation based on the yaw rate risk sub-score, the lateral position risk sub-score, the longitudinal deceleration risk sub-score, the turn signal risk sub-score, and their respective weight coefficients.

[0013] In some implementations, determining the lane-changing decision cost based on the product of the relative state data between the vehicle and the preceding vehicle and the comprehensive risk score includes: Based on the relative distance, relative speed, and collision time between the vehicle and the vehicle in front, a risk adjustment factor is calculated; Calculate the product of the comprehensive risk score and the risk adjustment factor, and determine the product as the lane change decision cost.

[0014] In some implementations, determining the cost of maintaining the original lane based on the relative state data between the vehicle and the vehicle in front includes: Based on the relative distance, relative speed, and collision time between the vehicle and the vehicle in front, a risk adjustment factor is calculated; Obtain preset calibration coefficients obtained through real vehicle calibration. These preset calibration coefficients are used to characterize the baseline risk level when the vehicle maintains its original lane. Calculate the product of the preset calibration coefficient and the risk adjustment factor, and determine the product as the cost of the decision to maintain the original lane.

[0015] In some implementations, calculating the risk adjustment factor based on the relative distance, relative speed, and collision time between the vehicle and the vehicle in front includes: Calculate the difference between the relative distance between the vehicle and the vehicle in front and the minimum tolerance distance, and determine the reciprocal of the difference as the first risk adjustment factor. When the relative distance is less than or equal to the minimum tolerance distance, the first risk adjustment factor is set to a preset maximum value. Calculate the value of an exponential function with a base of natural constant and an exponent of the ratio of the absolute value of the relative speed between the vehicle and the vehicle in front to a preset speed threshold, and determine the value of the exponential function as the second risk adjustment factor; Calculate the larger of the collision time between the vehicle and the vehicle in front and the minimum acceptable collision time, and divide 1 by the larger value to determine the third risk adjustment factor; The risk adjustment factor is determined by weighted summation of the first risk adjustment factor, the second risk adjustment factor, and the third risk adjustment factor.

[0016] Secondly, this application proposes a decision-making device for a scenario where two vehicles change lanes simultaneously, comprising: The risk scoring unit is used to determine the risk sub-scores of the preceding vehicle in multiple dimensions based on the motion state data of the preceding vehicle. The comprehensive scoring unit is used to perform a weighted summation of the risk sub-scores under the multiple dimensions to obtain a comprehensive risk score. The lane change cost unit is used to determine the lane change decision cost based on the product of the relative state data between the vehicle and the preceding vehicle and the comprehensive risk score. The original lane cost unit is used to determine the decision cost of maintaining the original lane based on the relative state data between the vehicle and the vehicle in front. A behavior decision unit is used to determine the target driving behavior of the vehicle based on the relationship between the lane change decision cost and the lane-keeping decision cost.

[0017] Thirdly, an electronic device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program stored in the memory to implement the steps of the decision-making method for a two-vehicle simultaneous lane-changing scenario as described in any of the first aspects.

[0018] Fourthly, this application also proposes a computer-readable storage medium storing a computer program thereon, wherein when the computer program is executed by a processor, it implements the steps of the decision-making method for a two-vehicle lane-changing scenario as described in any of the first aspects.

[0019] In summary, the decision-making method for a two-vehicle simultaneous lane-changing scenario provided in this application helps to understand whether the preceding vehicle intends to turn back by acquiring its motion state data. Based on the preceding vehicle's motion state data, corresponding risk sub-scores are determined from the preceding vehicle's behavior across multiple dimensions, quantifying the probability of the preceding vehicle returning to its original lane from different perspectives, thus avoiding misjudgments that may result from relying on a single parameter. A weighted summation of the risk sub-scores across multiple dimensions yields a comprehensive risk score, integrating multi-dimensional information into a quantitative indicator that comprehensively reflects the risk level of the preceding vehicle's turnback, making risk assessment more accurate. Furthermore, by combining the relative state data between the vehicle and the preceding vehicle, the lane-changing decision cost and the decision cost of maintaining the original lane are determined separately. This process correlates the preceding vehicle's turnback risk with the actual interaction state, such as the current relative distance and relative speed between the vehicle and the preceding vehicle, ensuring that the lane-changing decision cost and the decision cost of maintaining the original lane not only reflect the strength of the preceding vehicle's turnback intention but also the urgency of a conflict occurring under that intention. By comparing the decision costs of changing lanes and maintaining the original lane, the target driving behavior of the vehicle is determined. This comparison process directly selects the optimal option based on the aforementioned cost calculation results. When the decision cost of changing lanes is relatively low, it indicates that the risk of continuing to change lanes is relatively controllable. When the decision cost of maintaining the original lane is relatively low, it indicates that abandoning the lane change is safer. In the case where two vehicles change lanes at the same time and the vehicle in front may have to turn back urgently due to obstructed vision, the vehicle can anticipate potential dangers and make evasive decisions in advance based on the behavior of the vehicle in front, even without directly observing the risk of the target lane. This avoids collisions caused by perception delay due to obstruction by the vehicle in front. Attached Figure Description

[0020] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit this specification. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A schematic flowchart illustrating a decision-making method for a scenario where two vehicles change lanes simultaneously, provided in an embodiment of this application; Figure 2 A schematic diagram of a decision-making device for a scenario where two vehicles change lanes simultaneously, provided in an embodiment of this application; Figure 3 This is a schematic diagram of a decision-making electronic device for a scenario where two vehicles change lanes simultaneously, provided as an embodiment of this application. Detailed Implementation

[0021] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus. The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0022] This application applies to a typical hazardous situation in the field of vehicle-assisted driving or autonomous driving: a scenario where the vehicle and the vehicle in front are traveling in the same lane, both intending to change lanes into the same adjacent target lane, and the vehicle cannot directly observe the traffic conditions in the target lane due to the obstruction of the vehicle in front. In this application scenario, the vehicle and the vehicle in front are in the same lane, with the vehicle in front in front of the vehicle, and both have the same direction of travel and similar lane-changing intentions. Due to the obstruction of the vehicle in front's view, the vehicle's sensors (such as cameras and radar) have a limited range of perception of the target lane, making it impossible to detect in advance whether there are stationary obstacles, slow-moving vehicles, or other sudden risks in the target lane. At this time, if the vehicle in front discovers an obstacle in the target lane and abruptly abandons the lane change and returns to its original lane, the vehicle, due to its obstructed view, may still continue to travel towards the target lane according to the original lane-changing plan until the vehicle in front has completely turned back and the vehicle's view is restored, at which point the vehicle discovers the dangerous situation in the target lane, but by then it is too late to brake or avoid it, which can easily lead to a collision between the vehicle and the obstacle in the target lane or the vehicle in front. The technical solution of this application is designed for the specific application scenario of "two vehicles changing lanes to adjacent target lanes at the same time in the same lane and the vehicle's visibility is limited, and the vehicle in front may make an emergency turnback". It aims to predict the risk of turning back in advance by analyzing the dynamic behavior of the vehicle in front, thereby assisting the vehicle to make a safer decision.

[0023] To more clearly describe the technical solution of this application, the meanings of some specific terms involved in this application will be explained in detail below.

[0024] The "vehicle" in this application refers to a vehicle equipped with this decision-making method, i.e., the target vehicle that needs to make driving behavior decisions.

[0025] The vehicle preceding the vehicle in this application refers to a vehicle in the same lane, located in front of the vehicle, and with the intention to change lanes to the same adjacent target lane. The same lane refers to the original lane in which the vehicle and the vehicle preceding the vehicle were traveling together before the lane change operation was initiated.

[0026] The target lane in this application refers to another lane adjacent to the original lane into which the vehicle and the vehicle in front intend to change lanes.

[0027] Simultaneous lane change in this application refers to a situation where both the vehicle and the vehicle in front have the intention to move laterally toward the target lane, and their lane change actions overlap or are sequential in time.

[0028] The limited visibility in this application refers to the state in which the vehicle is unable to fully acquire information about traffic participants and obstacles in the target lane due to the physical obstruction of the detection range of the vehicle's sensors by the vehicle body of the preceding vehicle.

[0029] The term "turnaround" in this application refers to the behavior of a vehicle in front changing its lateral direction of movement and returning from the target lane to its original lane during the lane change process or the preparation stage for a lane change.

[0030] The motion state data in this application refers to a set of parameters characterizing the dynamic behavior and attitude of the vehicle in front, including but not limited to yaw rate, rate of change of yaw rate, lateral velocity, lateral position from the current lane centerline, longitudinal deceleration, and changes in turn signal status.

[0031] The yaw rate in this application refers to the angular velocity of the vehicle in front rotating about its vertical axis, which is used to reflect the degree of steering aggression of the vehicle in front. The positive and negative values ​​correspond to the left and right turns, respectively.

[0032] The rate of change of yaw rate in this application refers to the derivative of the yaw rate of the preceding vehicle with respect to time, which is used to reflect the changing trend of the preceding vehicle's steering intention.

[0033] The lateral velocity in this application refers to the velocity component of the preceding vehicle in the direction perpendicular to the lane, which is used to characterize how quickly the preceding vehicle deviates from or returns to the original lane centerline.

[0034] In this application, lateral position refers to the lateral offset distance of the preceding vehicle relative to its original lane centerline.

[0035] The longitudinal deceleration in this application refers to the deceleration acceleration of the preceding vehicle in the direction of travel, which is used to reflect the braking behavior of the preceding vehicle.

[0036] The turn signal in this application refers to the change in the state of the light signal emitted by the vehicle in front to indicate the intention to change lanes or turn back.

[0037] The obstacle information in this application refers to information related to static or dynamic obstacles located in front of the vehicle being driven by the vehicle ahead, detected by the vehicle's sensors.

[0038] The risk sub-score of this application refers to the normalized score obtained after quantitatively evaluating the performance of the preceding vehicle in a single dimension (such as yaw rate dimension, lateral position dimension, longitudinal deceleration dimension, and turn signal dimension), which is used to characterize the probability or urgency of the preceding vehicle turning back under that dimension, and the value ranges from 0 to 1.

[0039] The comprehensive risk score in this application refers to the fusion index obtained by weighted summation of multiple risk sub-scores, which is used to comprehensively characterize the overall probability of the preceding vehicle turning back.

[0040] The lane change decision cost in this application refers to the quantified value of the expected risk that the vehicle needs to bear to continue performing the lane change operation. This value is determined by the comprehensive risk score and the relative state between the vehicle and the vehicle in front.

[0041] The lane-keeping decision cost in this application refers to the expected risk cost calculated based on the relative state data of the vehicle and the vehicle in front when the vehicle chooses to abandon the lane change and keep driving in the current lane. This cost is used to measure the risk level of abandoning the lane change option.

[0042] The target driving behavior in this application refers to the action that the vehicle ultimately determines after comparing the decision cost of changing lanes with the decision cost of maintaining the original lane, including changing lanes or maintaining the original lane.

[0043] The relative state data in this application refers to the interaction parameters between the vehicle and the vehicle in front, including relative distance, relative speed, and collision time. Relative distance refers to the actual longitudinal distance between the front of the vehicle and the rear of the vehicle in front. Relative speed is the difference between the speed of the vehicle and the speed of the vehicle in front; the relative speed is positive when the speed of the vehicle in front is greater than the speed of the vehicle in front. Collision time refers to the time required for the vehicle and the vehicle in front to reach the current relative distance at the current relative speed.

[0044] The risk adjustment factor in this application refers to the amplification coefficient calculated based on relative state data. It is used to increase the decision cost when the interaction between the self-vehicle and the vehicle in front is urgent, thereby guiding the self-vehicle to choose a safer driving behavior.

[0045] Based on the above description, the technical solution described in this application quantifies the risk of the vehicle turning back due to a risk outside its line of sight when two vehicles change lanes to the same adjacent target lane in the same lane, and the vehicle in front turns back due to a risk outside its line of sight. It also calculates the cost of different decisions based on the multi-dimensional motion state data of the vehicle in front, and calculates the cost of different decisions based on the relative interaction state between the vehicle and the vehicle in front. Finally, it selects the driving behavior with the lower cost, thereby achieving early avoidance of potential collision accidents when the vehicle cannot directly observe the risk of the target lane.

[0046] Please see Figure 1This is a flowchart illustrating a decision-making method for a scenario where two vehicles change lanes simultaneously, provided in an embodiment of this application. Specifically, it may include: S110. Based on the motion state data of the preceding vehicle, determine the risk sub-score of the preceding vehicle in multiple dimensions; For example, based on the motion state data of the vehicle in front, the risk level of the vehicle in front is assessed from multiple dimensions such as yaw rate, lateral position, longitudinal deceleration and turn signal, and the assessment result of each dimension corresponds to a risk sub-score.

[0047] S120. The risk sub-scores under multiple dimensions are weighted and summed to obtain the comprehensive risk score; For example, based on the risk sub-scores of the preceding vehicle in multiple dimensions such as yaw rate, lateral position, longitudinal deceleration, and turn signal, each risk sub-score is multiplied by its corresponding preset weight coefficient and then summed. The weighted summation is used to calculate the comprehensive risk score, which is used to quantify the overall probability of the preceding vehicle making an emergency return to its original lane.

[0048] S130. Determine the lane change decision cost based on the product of the relative state data of the vehicle and the vehicle in front and the comprehensive risk score; For example, based on the comprehensive risk score and the relative state data between the vehicle and the vehicle in front, a factor for adjusting the risk level is first determined according to the relative state data. Then, the comprehensive risk score is multiplied by the factor to calculate the lane change decision cost, so that the higher the urgency of the interaction between the vehicle and the vehicle in front, the greater the lane change decision cost.

[0049] S140. Based on the relative state data between the vehicle and the vehicle in front, determine the cost of maintaining the original lane. For example, based on the relative state data between the vehicle and the vehicle in front, a risk adjustment factor reflecting the urgency of the current interaction is calculated. This factor is then multiplied by a pre-calibrated coefficient to obtain the decision cost that the vehicle needs to bear if it abandons the lane change and continues to drive in the original lane.

[0050] S150. Determine the target driving behavior of the vehicle based on the relationship between the cost of lane changing and the cost of maintaining the original lane.

[0051] For example, by comparing the decision cost of changing lanes and the decision cost of staying in the original lane, the target driving behavior to be executed by the vehicle is determined by judging the relationship between the two. If the decision cost of changing lanes is smaller, it means that the expected risk of continuing to change lanes is lower than giving up changing lanes, and the vehicle should perform the lane change operation. If the decision cost of changing lanes is greater than or equal to the decision cost of staying in the original lane, it means that giving up changing lanes is safer, and the vehicle should stay in the original lane, thereby achieving a driving behavior selection based on the principle of minimizing risk.

[0052] In summary, the decision-making method for a two-vehicle simultaneous lane-changing scenario provided in this application helps to understand whether the preceding vehicle intends to turn back by acquiring its motion state data. Based on the preceding vehicle's motion state data, corresponding risk sub-scores are determined from the preceding vehicle's behavior across multiple dimensions, quantifying the probability of the preceding vehicle returning to its original lane from different perspectives, thus avoiding misjudgments that may result from relying on a single parameter. A weighted summation of the risk sub-scores across multiple dimensions yields a comprehensive risk score, integrating multi-dimensional information into a quantitative indicator that comprehensively reflects the risk level of the preceding vehicle's turnback, making risk assessment more accurate. Furthermore, by combining the relative state data between the vehicle and the preceding vehicle, the lane-changing decision cost and the decision cost of maintaining the original lane are determined separately. This process correlates the preceding vehicle's turnback risk with the actual interaction state, such as the current relative distance and relative speed between the vehicle and the preceding vehicle, ensuring that the lane-changing decision cost and the decision cost of maintaining the original lane not only reflect the strength of the preceding vehicle's turnback intention but also the urgency of a conflict occurring under that intention. By comparing the decision costs of changing lanes and maintaining the original lane, the target driving behavior of the vehicle is determined. This comparison process directly selects the optimal option based on the aforementioned cost calculation results. When the decision cost of changing lanes is relatively low, it indicates that the risk of continuing to change lanes is relatively controllable. When the decision cost of maintaining the original lane is relatively low, it indicates that abandoning the lane change is safer. In the case where two vehicles change lanes at the same time and the vehicle in front may have to turn back urgently due to obstructed vision, the vehicle can anticipate potential dangers and make evasive decisions in advance based on the behavior of the vehicle in front, even without directly observing the risk of the target lane. This avoids collisions caused by perception delay due to obstruction by the vehicle in front.

[0053] In some instances, multiple dimensions include yaw rate, lateral position, longitudinal deceleration, and turn signal.

[0054] For example, the multiple dimensions selected in this application specifically include the yaw rate dimension, the lateral position dimension, the longitudinal deceleration dimension, and the turn signal dimension. These four dimensions jointly characterize the features of the preceding vehicle's emergency turnaround behavior from different physical levels. Among them, the yaw rate dimension is used to capture the angular velocity of the preceding vehicle rotating around the vertical axis and its rate of change. When the absolute value of the yaw rate increases and the direction is towards the original lane, it indicates that the preceding vehicle is performing a turnaround maneuver, and its rate of change further reflects the degree of abrupt change in steering intention. The lateral position dimension is used to capture the lateral offset state of the preceding vehicle relative to the center line of the original lane. When the preceding vehicle performs a turnaround maneuver, the direction of its lateral velocity will change from towards the target lane to towards the original lane, and its lateral position from the center line of the original lane will gradually decrease from the maximum deviation. By monitoring the change in the direction of the lateral velocity and the return trend of the lateral position, the progress of the preceding vehicle abandoning the lane change and returning to the original lane is quantified. The longitudinal deceleration dimension is used to capture the braking behavior of the vehicle ahead. When the vehicle ahead detects an obstacle in the target lane and makes an emergency turnaround, it usually does so with significant deceleration to reduce its speed and create safe conditions for the turnaround. By monitoring the absolute value of the longitudinal deceleration of the vehicle ahead, the intensity of the emergency avoidance measure can be quantified. Simultaneously, combined with information about the obstacle ahead, the cause of the deceleration can be identified, ruling out normal following or traffic congestion-related deceleration. The turn signal dimension is used to capture the explicit expression of the driving intention of the vehicle ahead. When the vehicle ahead changes its intention from lane changing to turning back, its turn signal usually changes state, for example, from turning on the turn signal towards the target lane to turning off the turn signal or turning on the turn signal towards the original lane. By detecting changes in the turn signal state, the change in intention expressed by the vehicle ahead through its light signals can be quantified.

[0055] In summary, this application uses a weighted approach to assess the risk of a preceding vehicle turning back, combining yaw rate and rate of change, lateral displacement and velocity, longitudinal deceleration with obstacle factors and turn signal status changes. This allows the vehicle to perceive potential hazards before its own visibility recovers, avoiding misjudgments based on a single parameter. The longitudinal deceleration dimension incorporates obstacle factors to differentiate braking causes, preventing normal obstacle avoidance from being misjudged as a turnback. These dimensions enhance the early warning capability and confidence of the risk score, enabling the vehicle to indirectly infer target lane risks based on the behavior of the preceding vehicle under limited visibility conditions, making evasive decisions in advance and effectively reducing the probability of collisions caused by perception lag.

[0056] In some instances, risk sub-scores include yaw rate risk sub-scores, lateral position risk sub-scores, longitudinal deceleration risk sub-scores, and turn signal risk sub-scores. Based on the motion state data of the vehicle ahead, risk sub-scores for the vehicle ahead are determined in multiple dimensions, including: Based on the relationship between the yaw rate of the preceding vehicle and a preset yaw rate threshold, and the relationship between the rate of change of the yaw rate of the preceding vehicle and a preset rate of change threshold, a yaw rate risk sub-score is determined. Based on the relationship between the lateral speed of the vehicle in front and a preset lateral speed threshold, and the relationship between the lateral distance of the vehicle in front from the original lane centerline and a preset lateral position offset threshold, a lateral position risk sub-score is determined. Based on the relationship between the longitudinal deceleration of the preceding vehicle and the preset longitudinal deceleration threshold, and whether there is an obstacle in front of the preceding vehicle, a longitudinal deceleration risk sub-score is determined. A risk sub-score for the turn signal is determined based on whether the turn signal of the vehicle in front changes state.

[0057] In some instances, the preset yaw rate threshold includes a first yaw rate threshold and a second yaw rate threshold, and the preset rate of change threshold includes a first rate of change threshold and a second rate of change threshold. Based on the relationship between the yaw rate of the preceding vehicle and the preset yaw rate threshold, and the relationship between the rate of change of the yaw rate of the preceding vehicle and the preset rate of change threshold, a yaw rate risk sub-score is determined, including: Compare the absolute value of the yaw rate of the vehicle in front with the absolute values ​​of the first and second yaw rate thresholds, and determine the first normalized risk value based on the comparison result, including: When the absolute value of the yaw rate of the current vehicle is less than or equal to the absolute value of the first yaw rate threshold, the first normalized risk value is set to 0. When the absolute value of the yaw rate of the current vehicle is greater than or equal to the absolute value of the second yaw rate threshold, the first normalized risk value is set to 1. When the absolute value of the yaw rate of the vehicle in front is greater than the absolute value of the first yaw rate threshold and less than the absolute value of the second yaw rate threshold, the yaw rate of the vehicle in front is linearly interpolated to obtain the first normalized risk value, wherein the absolute value of the first yaw rate threshold is less than the absolute value of the second yaw rate threshold. Compare the absolute value of the rate of change of the yaw rate of the vehicle in front with the absolute values ​​of the first rate of change threshold and the second rate of change threshold, and determine the second normalized risk value based on the comparison results, wherein the absolute value of the first rate of change threshold is less than the absolute value of the second rate of change threshold. Calculate the arithmetic mean of the first normalized risk value and the second normalized risk value, and determine the arithmetic mean as the yaw rate risk sub-score.

[0058] For example, the signs of the first and second yaw rate thresholds are determined based on the turning direction of the vehicle ahead. When the turning direction is left, the turning direction is positive, and both the first and second yaw rate thresholds are taken as positive numbers; when the turning direction is right, the turning direction is negative, and both the first and second yaw rate thresholds are taken as negative numbers. The absolute value of the first yaw rate threshold represents the yaw rate boundary value at which a risk begins, and the absolute value of the second yaw rate threshold represents the yaw rate boundary value at which a very high risk occurs, and the absolute value of the first yaw rate threshold is less than the absolute value of the second yaw rate threshold.

[0059] After obtaining the yaw rate of the vehicle ahead, it is determined whether the absolute value of the yaw rate is less than or equal to the absolute value of the first yaw rate threshold. If this condition is met, it indicates that the steering amplitude of the vehicle ahead has not reached the level of risk of turning back, and the first normalized risk value is set to 0. If the absolute value of the yaw rate of the vehicle ahead is greater than or equal to the absolute value of the second yaw rate threshold, it indicates that the steering amplitude of the vehicle ahead has reached a very high level of risk, and the first normalized risk value is set to 1.

[0060] If the absolute value of the yaw rate of the vehicle in front is greater than the absolute value of the first yaw rate threshold but less than the absolute value of the second yaw rate threshold, a linear interpolation is performed on the yaw rate of the vehicle in front. The interpolation formula is: the first normalized risk value equals the difference between the absolute value of the yaw rate of the vehicle in front and the absolute value of the first yaw rate threshold, divided by the difference between the absolute value of the second yaw rate threshold and the absolute value of the first yaw rate threshold, and the calculation result is restricted to a closed interval between 0 and 1. Since the absolute values ​​of the first and second yaw rate thresholds are both positive, this interpolation calculation avoids the ambiguity of the negative number domain and reflects the degree of transition from no risk to high risk in the yaw rate.

[0061] For the rate of change of the yaw rate of the preceding vehicle, the same rules as for the yaw rate itself are applied. The signs of the first and second rate of change thresholds are determined based on the direction of the preceding vehicle's turnaround: both thresholds are positive when the turnaround direction is positive, and negative when the turnaround direction is negative. The absolute value of the first rate of change threshold represents the boundary value of the yaw rate of change at which a risk begins, and the absolute value of the second rate of change threshold represents the boundary value of the rate of change with a very high risk. Furthermore, the absolute value of the first rate of change threshold is less than the absolute value of the second rate of change threshold.

[0062] After obtaining the rate of change of the yaw rate of the vehicle in front, it is determined whether the absolute value of this rate of change is less than or equal to the absolute value of the first rate of change threshold. If so, the second normalized risk value is set to 0. If the absolute value of the rate of change is greater than or equal to the absolute value of the second rate of change threshold, the second normalized risk value is set to 1. If the absolute value of the rate of change is between the two, it is calculated using the same linear interpolation formula: the second normalized risk value is equal to the difference between the absolute value of the rate of change of the yaw rate of the vehicle in front and the absolute value of the first rate of change threshold, divided by the difference between the absolute value of the second rate of change threshold and the absolute value of the first rate of change threshold, and the result is restricted to a closed interval between 0 and 1. The second normalized risk value obtained in this way quantifies the urgency of the sudden change in the steering intention of the vehicle in front.

[0063] After obtaining the first and second normalized risk values, they are arithmetically averaged. The yaw rate risk sub-score is calculated as the sum of the first and second normalized risk values ​​divided by 2. This arithmetic mean integrates the current amplitude and rate of change of the yaw rate, reflecting both the current steering posture of the vehicle ahead and the changing trend of its steering intention, as well as the steering characteristics of the vehicle ahead during an emergency turnaround. This avoids the influence of single-dimensional noise or instantaneous fluctuations on risk assessment.

[0064] In summary, the embodiments of this application, relying solely on yaw rate to determine the turning intention of the preceding vehicle, are easily affected by steering fluctuations during normal lane changes. For example, the preceding vehicle may experience a slight change in yaw rate due to lane line correction during a lane change, but this is not a true turning back. By introducing the yaw rate change rate, a distinction can be made between normal steering corrections and sudden changes in intention. When the yaw rate change rate is small, even if the absolute value of the yaw rate is large, the second normalized risk value is low, and the average yaw rate risk sub-score is lowered, avoiding misjudging normal lane centering adjustments as emergency turning back. Conversely, when the preceding vehicle truly turns back due to a sudden obstacle ahead, its steering direction will quickly switch from pointing towards the target lane to pointing towards the original lane. This process is inevitably accompanied by a sharp increase in the yaw rate change rate, and the second normalized risk value approaches 1. Even if the absolute value of the yaw rate has not yet reached the second yaw rate threshold, the average score can reflect the increased risk in advance. Therefore, when the yaw rate has not yet fully reached the high-risk threshold, the early warning signal of the rate of change can be used to identify the intention to turn back in advance, giving the vehicle more time to make a decision and respond. At the same time, it avoids false alarms caused by fluctuations in a single parameter, thus improving the accuracy and timeliness of risk prediction.

[0065] In some instances, the preset lateral speed threshold includes a first lateral speed threshold and a second lateral speed threshold. A lateral position risk sub-score is determined based on the relationship between the lateral speed of the preceding vehicle and the preset lateral speed threshold, and the relationship between the lateral distance of the preceding vehicle from the original lane centerline and the preset lateral position offset threshold. This sub-score includes: The absolute value of the lateral speed of the vehicle in front is compared with the absolute values ​​of the first lateral speed threshold and the second lateral speed threshold. A third normalized risk value is determined based on the comparison results, wherein the absolute value of the first lateral speed threshold is less than the absolute value of the second lateral speed threshold. Compare the absolute value of the lateral distance between the preceding vehicle and the original lane centerline with the preset lateral position offset threshold and half a preset lane width, and determine the fourth normalized risk value based on the comparison results. The preset lateral position offset threshold is less than half a preset lane width. Calculate the arithmetic mean of the third and fourth normalized risk values, and determine the arithmetic mean as the lateral position risk sub-score.

[0066] For example, the signs of the first and second lateral speed thresholds are determined based on the turning direction of the vehicle ahead. Both thresholds are positive when the turning direction is left and negative when the turning direction is right. The absolute value of the first lateral speed threshold represents the lateral speed boundary value at which a risk begins, and the absolute value of the second lateral speed threshold represents the lateral speed boundary value at which a very high risk occurs. Furthermore, the absolute value of the first lateral speed threshold is less than the absolute value of the second lateral speed threshold.

[0067] After obtaining the lateral speed of the vehicle ahead, it is determined whether the absolute value of this lateral speed is less than or equal to the absolute value of the first lateral speed threshold. If this condition is met, it indicates that the lateral speed of the vehicle ahead has not yet reached the level of risk of turning back, and the third normalized risk value is set to 0. If the absolute value of the lateral speed of the vehicle ahead is greater than or equal to the absolute value of the second lateral speed threshold, it indicates that the lateral speed of the vehicle ahead has reached or exceeded a very high risk level, and the third normalized risk value is set to 1.

[0068] If the absolute value of the lateral speed of the vehicle in front is greater than the absolute value of the first lateral speed threshold but less than the absolute value of the second lateral speed threshold, a third normalized risk value is calculated using a linear interpolation formula. This formula is: the third normalized risk value equals the difference between the absolute value of the lateral speed of the vehicle in front and the absolute value of the first lateral speed threshold, divided by the difference between the absolute values ​​of the second and first lateral speed thresholds. The result is then limited to a closed interval between 0 and 1. Through segmented processing, the third normalized risk value quantifies the risk level corresponding to the lateral movement speed of the vehicle in front towards its original lane direction. The larger the absolute value of the lateral speed, the closer this value is to 1, indicating a faster turning maneuver by the vehicle in front and a higher risk.

[0069] The lateral position offset threshold is used to define the lateral position offset at which a turning-back risk begins to arise. It is set to 0.3 meters, indicating that a turning-back risk begins when the current vehicle deviates more than this value from the original lane centerline. The preset lane width refers to the standard width of the original lane. Half of this width is taken as the boundary for extremely high-risk lateral position offset. This is because when the lateral position of the preceding vehicle from the original lane centerline is close to half the lane width, the preceding vehicle is already at its maximum deviation position during lane changing. If a turning-back maneuver is performed at this point, the risk of collision is high. The lateral position of the preceding vehicle from the lane centerline is defined as the vertical distance between the current center of gravity of the preceding vehicle and the original lane centerline, with the original lane centerline as the reference zero point and the direction towards the target lane as the positive direction.

[0070] When calculating the fourth normalized risk value, the absolute value of the lateral distance between the preceding vehicle and the original lane centerline is obtained. This absolute value represents the degree to which the preceding vehicle deviates from the original lane centerline. It is then determined whether this absolute value is less than or equal to a preset lateral position offset threshold. If it is, it indicates that the preceding vehicle has not significantly deviated from the original lane centerline, and the risk of turning back is low; in this case, the fourth normalized risk value is set to 0. If the absolute value is greater than or equal to half a preset lane width, it indicates that the preceding vehicle has approached or exceeded the lane boundary. In this case, if a turnback occurs, the risk is high, and the fourth normalized risk value is set to 1.

[0071] If the absolute value falls between a preset lateral position offset threshold and half a preset lane width, a fourth normalized risk value is calculated using a linear interpolation formula. This formula is: the fourth normalized risk value equals the absolute value of the preceding vehicle's lateral distance minus the preset lateral position offset threshold, divided by the difference between half a preset lane width and the preset lateral position offset threshold, with the result limited to a closed interval between 0 and 1. Through this calculation, the fourth normalized risk value quantifies the risk corresponding to the degree to which the preceding vehicle deviates from the original lane centerline. The greater the deviation distance, the closer the value is to 1, indicating a higher degree of lane change by the preceding vehicle, and thus a higher risk of returning to the original lane.

[0072] After obtaining the third and fourth normalized risk values, the arithmetic mean of the two is calculated. The lateral position risk sub-score is equal to the sum of the third and fourth normalized risk values ​​divided by 2. This arithmetic mean combines the lateral speed and lateral position information of the preceding vehicle when it turns around.

[0073] In summary, this application's embodiments describe the reversing behavior of the vehicle ahead by integrating two dimensions: lateral speed and lateral position. Relying solely on lateral position can easily misjudge the maximum deviation of a normal lane change as a reversal, while relying solely on lateral speed is susceptible to sensor noise or road surface bumps, leading to false alarms. After weighted averaging of the two dimensions, the score increases only when the lateral speed is moving towards the original lane and the lateral position still deviates significantly, effectively suppressing false alarms.

[0074] In some instances, the preset longitudinal deceleration threshold includes a first longitudinal deceleration threshold and a second longitudinal deceleration threshold. Based on the relationship between the longitudinal deceleration of the preceding vehicle and the preset longitudinal deceleration threshold, and the presence of an obstacle in front of the preceding vehicle, a longitudinal deceleration risk sub-score is determined, including: The absolute value of the longitudinal deceleration of the preceding vehicle is compared with the absolute values ​​of the first longitudinal deceleration threshold and the second longitudinal deceleration threshold. Based on the comparison results, a fifth normalized risk value is determined, wherein the absolute value of the first longitudinal deceleration threshold is less than the absolute value of the second longitudinal deceleration threshold. Based on whether there are obstacles in front of the vehicle, determine the obstacle influence factor, where the value of the obstacle influence factor ranges from 0 to 1; Calculate the difference between the preset constant and the obstacle influence factor, and determine the difference as the difference factor; Calculate the product of the fifth normalized risk value and the difference factor, and determine the product as the longitudinal deceleration risk sub-score.

[0075] For example, the first longitudinal deceleration threshold represents the boundary value of the longitudinal deceleration of the preceding vehicle when the risk begins, and the second longitudinal deceleration threshold represents the boundary value of the longitudinal deceleration of the preceding vehicle when the risk is very high. Since the longitudinal deceleration is negative when the preceding vehicle decelerates, both thresholds are negative, and the absolute value of the first longitudinal deceleration threshold is less than the absolute value of the second longitudinal deceleration threshold.

[0076] After obtaining the longitudinal deceleration of the vehicle ahead, the absolute value of this longitudinal deceleration is compared with the absolute values ​​of the first and second longitudinal deceleration thresholds. When the absolute value of the longitudinal deceleration is less than or equal to the absolute value of the first longitudinal deceleration threshold, it indicates that the longitudinal deceleration of the vehicle ahead is relatively mild and has not yet reached the level of potential backtracking risk; the fifth normalized risk value is set to 0. When the absolute value of the longitudinal deceleration is greater than or equal to the absolute value of the second longitudinal deceleration threshold, it indicates that the longitudinal deceleration of the vehicle ahead is severe and has reached a very high level of risk; the fifth normalized risk value is set to 1.

[0077] When the absolute value of the longitudinal deceleration lies between the absolute values ​​of the first and second longitudinal deceleration thresholds, a fifth normalized risk value is calculated using a linear interpolation formula. This formula is: the fifth normalized risk value equals the difference between the absolute value of the preceding vehicle's longitudinal deceleration and the absolute value of the first longitudinal deceleration threshold, divided by the difference between the absolute value of the second and first longitudinal deceleration thresholds. The result is then restricted to a closed interval between 0 and 1. Through this segmented processing, the fifth normalized risk value quantifies the risk level corresponding to the preceding vehicle's longitudinal deceleration behavior. The larger the absolute value of the longitudinal deceleration, the closer this value is to 1, indicating that the preceding vehicle is braking more aggressively and the likelihood of taking emergency evasive action is higher.

[0078] The obstacle impact factor is determined based on whether there are obstacles in front of the vehicle. Obstacle information in front of the vehicle is obtained through the vehicle's own sensors or vehicle-to-infrastructure communication, and is used to determine whether the vehicle's deceleration is caused by static or dynamic obstacles in front of it.

[0079] When there is no obstacle in front of the vehicle, it indicates that the vehicle's deceleration is not due to avoiding a target ahead; its deceleration may be due to normal following or traffic congestion. Therefore, the obstacle influence factor is set to 0. When there is an obstacle in front of the vehicle, and this obstacle is the main cause of the vehicle's deceleration, it indicates that the vehicle's deceleration is directly caused by a risk ahead, and its turning-back intention is strongly correlated with the obstacle. Therefore, the obstacle influence factor is set to 1.

[0080] When there is an obstacle in front of the vehicle but it cannot be confirmed whether the obstacle is the main cause of the vehicle's deceleration, the obstacle influence factor is set to an intermediate value between 0 and 1. This intermediate value is calibrated based on the degree of matching between the vehicle's deceleration and the distance to the obstacle. The obstacle influence factor ranges from 0 to 1, with a higher value indicating a higher probability that the vehicle's deceleration was caused by the obstacle in front.

[0081] The preset constant is set to 1. The difference between the preset constant and the obstacle influence factor is calculated, i.e., the difference factor equals 1 minus the obstacle influence factor. The product of the fifth normalized risk value and the difference factor is calculated, and this product is determined as the longitudinal deceleration risk sub-score.

[0082] When the obstacle impact factor is 0, the difference factor is 1, and the longitudinal deceleration risk sub-score is equal to the fifth normalized risk value multiplied by 1. At this time, the deceleration of the vehicle in front is considered to be caused by non-obstacles, and its deceleration behavior itself constitutes the basis for the risk of turning back.

[0083] When the obstacle impact factor is 1, the difference factor is 0, and the longitudinal deceleration risk sub-score is equal to the fifth normalized risk value multiplied by 0, that is, the result is 0. At this time, the deceleration of the vehicle in front is confirmed to be caused by the obstacle. Its deceleration behavior reflects the existence of risk ahead rather than the vehicle in front's own intention to turn back. This deceleration behavior is not included in the turnback risk assessment to avoid misjudging the deceleration of the vehicle in front due to avoidance of the obstacle ahead as a signal of the vehicle in front turning back.

[0084] When the obstacle influence factor takes a value between 0 and 1, the difference factor also falls between 0 and 1. The longitudinal deceleration risk sub-score equals the fifth normalized risk value multiplied by this difference factor. This means the fifth normalized risk value is attenuated according to the confidence level of the obstacle influence factor; the larger the obstacle influence factor, the greater the attenuation, indicating a higher proportion of the deceleration behavior explained by the obstacle and a lower proportion reflecting the preceding vehicle's own turning intention. Through this product calculation, the longitudinal deceleration risk sub-score is adjusted within the range of 0 to 1, retaining only the risk component of the preceding vehicle's deceleration behavior that cannot be explained by the obstacle ahead.

[0085] In summary, this application's embodiments attenuate the longitudinal deceleration risk score by introducing an obstacle influence factor. When there is an obstacle in front of the vehicle and its deceleration behavior can be explained, the obstacle influence factor approaches 1, and the longitudinal deceleration risk score is attenuated to near 0, preventing the vehicle from mistakenly abandoning a lane change due to the normal deceleration of the vehicle in front. When there is no obstacle in front of the vehicle but it decelerates violently, the obstacle influence factor is 0, the longitudinal deceleration risk score is retained, allowing the vehicle to perceive the abnormality in advance and take evasive measures.

[0086] In some instances, a turn signal risk sub-score is determined based on whether the turn signal of the vehicle ahead changes state, including: Detect whether the turn signal of the vehicle in front changes from on to off, from off to on, or from on one side to on the other. When a change in the state of the turn signal is detected, the risk sub-score for the turn signal is set to 1. If no change in the state of the turn signal is detected, the risk sub-score for the turn signal is set to 0.

[0087] For example, the system collects the turn signal from the vehicle in front. This signal is detected by a visual sensor to identify the flashing state of the turn signal at the rear of the vehicle in front, or by obtaining the turn signal control command from the vehicle in front via vehicle-to-everything (V2X) communication. The detection process includes continuously monitoring the state value of the turn signal from the vehicle in front, which includes at least three states: off, left turn on, and right turn on. Specifically, three types of state changes are detected: the first is from on to off, for example, if the vehicle in front was originally flashing its left or right turn signal and then the turn signal went out, it indicates that the vehicle in front may have abandoned its original lane change intention; the second is from off to on, for example, if the vehicle in front was originally not using its turn signal and then turned on one of its turn signals, it indicates that the vehicle in front is beginning to express a new lane change or turning back intention; the third is from one side on to the other, for example, if the vehicle in front originally used its left turn signal and then turned off its left turn signal and turned on its right turn signal, or vice versa, it indicates that the vehicle in front has changed its expected driving direction, from intending to change lanes to the left to change lanes to the right or turn back. When any of the above-mentioned state changes is detected, it is determined that the turn signal of the vehicle in front has changed, indicating that the vehicle in front has expressed a change in its driving intention through the light signal. If no of the above-mentioned state changes is detected within the continuous monitoring time window, that is, the turn signal state remains constant, it is determined that no state change has occurred.

[0088] After detecting changes in the turn signal status, the specific value of the turn signal risk sub-score is determined based on the detection results. When a change in the turn signal status of the vehicle ahead is detected, the turn signal risk sub-score is set to 1, because any change in the turn signal status indicates that the vehicle ahead is altering its original driving plan; this signal is the most direct and explicit expression of the driver's intention. When no change in the turn signal status of the vehicle ahead is detected, the turn signal risk sub-score is set to 0, indicating that this dimension does not provide any risk indication signal and does not increase the assessment of turnback risk due to the turn signal status. This binary scoring method treats the turn signal as a switching input. In the weighted summation of the comprehensive risk score, it only participates in the calculation when a status change is detected; otherwise, its contribution is 0, ensuring that the comprehensive risk score responds to the intention change information conveyed by the vehicle ahead through its lights.

[0089] In some instances, a weighted summation of risk sub-scores across multiple dimensions is performed to obtain a comprehensive risk score, including: The comprehensive risk score is determined by weighted summation based on the risk sub-scores of yaw rate, lateral position, longitudinal deceleration, and turn signal, as well as their respective weight coefficients.

[0090] For example, risk sub-scores for each of the four dimensions are obtained. Each risk sub-score is pre-normalized and limited to a closed interval of 0 to 1, ensuring comparability of indicators with different physical dimensions. A preset weighting coefficient is assigned to each risk sub-score, the magnitude of which reflects the importance of the corresponding dimension in the assessment of the preceding vehicle's turnaround risk. The specific operation of the weighted summation calculation is as follows: the yaw rate risk sub-score is multiplied by its weighting coefficient to obtain the first product; the lateral position risk sub-score is multiplied by its weighting coefficient to obtain the second product; the longitudinal deceleration risk sub-score is multiplied by its weighting coefficient to obtain the third product; and the turn signal risk sub-score is multiplied by its weighting coefficient to obtain the fourth product. Then, the first, second, third, and fourth products are added together, and the sum is the comprehensive risk score. This comprehensive risk score is also constrained to the range of 0 to 1, used to quantify the overall probability of the preceding vehicle engaging in emergency turnaround behavior.

[0091] The weighting coefficients are assigned according to the varying strengths of each dimension in representing the turning behavior, and the sum of all weighting coefficients equals 1 for normalization. The lateral position risk sub-rating has the highest weighting coefficient, set at 0.4, because lateral position and its changes are the most direct spatial representation of the preceding vehicle's turning, including changes in lateral velocity direction and distance regression from the centerline, most intuitively reflecting whether the preceding vehicle is abandoning a lane change and returning to its original lane. The yaw rate risk sub-rating has the second highest weighting coefficient, set at 0.3, because yaw rate and its rate of change reflect the preceding vehicle's steering intention and the degree of sudden steering change. The longitudinal deceleration risk sub-rating has a weighting coefficient of 0.2, because deceleration is usually an accompanying phenomenon of turning, and normal deceleration caused by obstacles ahead has been eliminated through the obstacle influence factor; this dimension is only used as an auxiliary judgment. The turn signal risk sub-rating has the lowest weighting coefficient, set at 0.1, because although the intention of turn signal is clear, its reliability is relatively low, with situations such as the driver not using the signal, not performing the action after using the signal, or the signal being obstructed; therefore, it is given the lowest weight to reduce the impact of false detections. The sum of the four weighting coefficients 0.4, 0.3, 0.2, and 0.1 equals 1, which satisfies the normalization constraint and ensures that the comprehensive risk score always remains within a reasonable range of 0 to 1.

[0092] In some instances, risk adjustment factors are calculated based on the relative distance, relative speed, and time of collision between the vehicle and the vehicle in front, including: Calculate the difference between the relative distance between the vehicle and the vehicle in front and the minimum tolerance distance. The reciprocal of the difference is determined as the first risk adjustment factor. When the relative distance is less than or equal to the minimum tolerance distance, the first risk adjustment factor is set to the preset maximum value. Calculate the value of an exponential function with a base of natural constant and an exponent of the ratio of the absolute value of the relative speed between the vehicle and the vehicle in front to a preset speed threshold. Use the value of the exponential function as the second risk adjustment factor. Calculate the larger of the collision time between the vehicle and the vehicle in front and the minimum acceptable collision time, and divide 1 by the larger value to determine the third risk adjustment factor; The first risk adjustment factor, the second risk adjustment factor, and the third risk adjustment factor are weighted and summed to determine the risk adjustment factor.

[0093] For example, the actual longitudinal distance between the front of the vehicle and the rear of the vehicle in front is obtained as the relative distance. Simultaneously, a pre-calibrated minimum tolerance distance is obtained. This minimum tolerance distance represents the lower threshold for maintaining safe interaction between the vehicle and the vehicle in front, typically set based on vehicle braking performance and reaction time. The difference between the relative distance and the minimum tolerance distance is calculated. If this difference is greater than zero, it indicates that the longitudinal distance between the vehicle and the vehicle in front is still within a safe margin. In this case, the first risk adjustment factor is set as the reciprocal of the difference between the relative distance and the minimum tolerance distance, using a very small positive number as the lower limit of the denominator to prevent division by zero errors. That is, the first risk adjustment factor equals 1 divided by the larger of the difference between the relative distance and the minimum tolerance distance and 0.1 meters. This calculation ensures that the closer the relative distance is to the minimum tolerance distance, the smaller the difference and the larger its reciprocal, amplifying the urgency of the risk at close range. If the relative distance is less than or equal to the minimum tolerance distance, it indicates that the vehicle and the vehicle in front have entered a close proximity range, and any lane-changing attempt will easily lead to a collision risk. In the above manner, the first risk adjustment factor describes the degree of proximity between the vehicle and the vehicle in front in the longitudinal space. The closer the relative distance is to the minimum tolerance distance, the larger the value of this factor, which in turn increases the lane change decision cost in the cost function, so that the vehicle will give priority to abandoning the lane change.

[0094] When determining the second risk adjustment factor, the difference between the vehicle's speed and the speed of the vehicle in front is used as the relative speed. A positive relative speed indicates the vehicle is approaching the vehicle ahead, while a negative relative speed indicates the distance between the two vehicles is increasing. For turnaround risk scenarios, the situation where the vehicle's speed is greater than the vehicle ahead's is the most dangerous because the vehicle ahead often decelerates sharply during an emergency turnaround. If the vehicle is still approaching at a high speed, a collision is inevitable. Therefore, a preset speed threshold is set as a positive speed boundary value to distinguish between normal and dangerous speed differences. The ratio of the absolute value of the relative speed to the preset speed threshold is calculated and then expressed as a natural constant. e Using the base as the base, and this ratio as the exponent, an exponential operation is performed to obtain the second risk adjustment factor, that is, the second risk adjustment factor equals... eThe exponential function calculates the risk factor by raising the ratio of the absolute relative speed to a preset speed threshold. When the absolute relative speed is less than or equal to the preset speed threshold, the exponential value is small, and the second risk adjustment factor is close to 1, indicating that the speed difference has not significantly amplified the risk. When the absolute relative speed is greater than the preset speed threshold, the exponential value increases, and the second risk adjustment factor grows rapidly, reflecting the trend of increased collision energy and shortened avoidance time caused by the vehicle's excessive speed relative to the vehicle in front. This exponential function causes the risk adjustment factor to rise exponentially once the speed difference exceeds the threshold, prioritizing maintaining the original lane rather than risky lane changes under high relative speed conditions.

[0095] When determining the third risk adjustment factor based on the collision time between the vehicle and the vehicle in front and the minimum acceptable collision time, the collision time is calculated based on the relative distance and relative speed. If the relative speed is greater than zero, the collision time is equal to the relative distance divided by the relative speed. This value indicates how many seconds it will take for the vehicle to rear-end the vehicle in front if the current relative speed remains constant. If the relative speed is less than or equal to zero, it indicates that the vehicle is not approaching the vehicle in front or is moving away from it. In this case, the collision time is set to infinity to indicate that there is no risk of collision.

[0096] A pre-calibrated minimum acceptable collision time is obtained, typically 1.5 seconds as the safety threshold. This indicates that if the collision time is below this threshold, the vehicle and the vehicle in front are in a dangerously close proximity. The third risk adjustment factor is calculated as follows: the larger of the collision time and the minimum acceptable collision time is used as the denominator, and 1 is used as the numerator. The third risk adjustment factor is equal to 1 divided by the larger of the collision time and the minimum acceptable collision time. When the collision time is greater than the minimum acceptable collision time, the denominator is the collision time, and the third risk adjustment factor is less than or equal to 2 / 3. The larger the collision time, the smaller the factor, indicating sufficient time margin and low risk. When the collision time is less than or equal to the minimum acceptable collision time, the denominator is the minimum acceptable collision time, and the third risk adjustment factor is constant at approximately 0.67. However, in reality, because the collision time is already very small, even with a fixed factor value, the overall risk adjustment factor will still increase due to the amplification effect of the first and second factors. This design ensures that when the collision time is below the safety threshold, the third risk adjustment factor contributes at least a moderately high benchmark amplification factor, avoiding numerical instability caused by an excessively small denominator due to the collision time approaching zero.

[0097] After obtaining the first, second, and third risk adjustment factors, they are weighted and summed to determine the final risk adjustment factor. Pre-defined weight coefficients are assigned to each factor, reflecting the differences in the contribution of different relative state dimensions to the urgency of risk. The first factor, based on relative distance, is typically given the highest weight because spatial distance is the most direct determinant of collision occurrence; the second factor, based on relative velocity, is given the second highest weight because velocity difference determines collision energy and response time; and the third factor, based on collision time, is given a lower weight as an auxiliary verification indicator. The first factor is multiplied by its weight, the second factor by its weight, and the third factor by its weight, and then the three products are summed; the resulting sum is the risk adjustment factor.

[0098] In summary, this embodiment calculates a risk adjustment factor by integrating three dimensions: relative distance, relative speed, and collision time, to perceive the urgency of the interaction between the vehicle and the vehicle in front. When the vehicle is too close to the vehicle in front, the relative speed is too high, or the collision time is insufficient, the risk adjustment factor amplifies the cost of lane-changing decisions, forcing the vehicle to prioritize abandoning the lane change. This effectively avoids collisions caused by forced lane changes in dangerous following situations, improving decision-making safety and scenario adaptability under conditions of limited visibility.

[0099] In some instances, the cost of a lane-changing decision is determined based on the product of the relative state data of the vehicle and the vehicle in front, and a comprehensive risk score, including: The risk adjustment factor is calculated based on the relative distance, relative speed, and collision time between the vehicle and the vehicle in front. Calculate the product of the comprehensive risk score and the risk adjustment factor, and determine the cost of the lane change decision.

[0100] For example, the comprehensive risk score calculated above is obtained. This comprehensive risk score is a normalized value obtained by weighting and summing the yaw rate risk sub-score, the lateral position risk sub-score, the longitudinal deceleration risk sub-score, and the turn signal risk sub-score. Its value ranges from 0 to 1, and it reflects the overall probability of the preceding vehicle making an emergency turnaround. The closer the comprehensive risk score is to 1, the stronger the preceding vehicle's intention to turn around and the higher the probability of turning around.

[0101] Simultaneously, a risk adjustment factor is obtained based on the relative state data between the vehicle and the vehicle in front. This risk adjustment factor is an amplification coefficient obtained by weighted summation of the first risk adjustment factor, the second risk adjustment factor, and the third risk adjustment factor. Its value range varies according to the relative distance, relative speed, and collision time between the vehicle and the vehicle in front.

[0102] After obtaining the comprehensive risk score and risk adjustment factor, the comprehensive risk score and risk adjustment factor are multiplied together. The resulting product is the lane-change decision cost. The physical meaning of this multiplication is that the lane-change decision cost depends not only on the strength of the preceding vehicle's intention to turn back, but also on the urgency of the conflict determined by the current interaction between the vehicle and the preceding vehicle. When the preceding vehicle's intention to turn back is weak but the distance between the vehicle and the preceding vehicle is close, the amplification of the risk adjustment factor keeps the lane-change decision cost at a high level, forcing the vehicle to abandon the lane change. When the preceding vehicle's intention to turn back is strong but the distance between the vehicle and the preceding vehicle is sufficient and the speed difference is small, the risk adjustment factor is at a low level, and the lane-change decision cost may still be lower than the decision cost of maintaining the original lane, allowing the vehicle to choose to continue changing lanes. By multiplying these two dimensions of risk quantification, a comprehensive cost index is formed. The larger the value of this lane-change decision cost, the higher the expected risk faced by the vehicle in performing the lane change operation; the smaller the value, the safer the lane-change operation.

[0103] In some instances, the cost of maintaining the original lane is determined based on the relative state data of the vehicle and the vehicle in front, including: The risk adjustment factor is calculated based on the relative distance, relative speed, and collision time between the vehicle and the vehicle in front. Obtain the preset calibration coefficients obtained through real vehicle calibration. The preset calibration coefficients are used to characterize the baseline risk level when the vehicle maintains its original lane. Calculate the product of the preset calibration coefficient and the risk adjustment factor, and determine the product as the decision cost of maintaining the original lane.

[0104] For example, a risk adjustment factor is calculated based on the relative distance, relative speed, and collision time between the vehicle and the vehicle in front. This risk adjustment factor is an amplification coefficient used to quantify the urgency of the conflict between the vehicle and the vehicle in front in the current interaction state. Its value increases monotonically as the relative distance decreases, the relative speed increases, or the collision time shortens. Simultaneously, a preset calibration coefficient, obtained through real-vehicle calibration, is acquired. This preset calibration coefficient is a constant value representing the basic risk level borne by the vehicle in maintaining its original lane under normal following conditions. This basic risk covers the risk of a rear-end collision between the vehicle and the vehicle in front, as well as the opportunity loss risk of abandoning the lane change. The preset calibration coefficient and the risk adjustment factor are multiplied together; the resulting product is the decision cost of maintaining the original lane. This product scales the basic risk level according to the urgency of the interaction. When the distance between the vehicle and the vehicle in front is too close, the relative speed is too high, or the collision time is insufficient, the risk adjustment factor increases, and the decision cost of maintaining the original lane increases accordingly, reflecting that abandoning the lane change still faces a high risk of longitudinal conflict.

[0105] In summary, this application's embodiments do not rely on the turning-back risk score or turning-back intention prediction of the preceding vehicle. Instead, they calculate the cost solely based on the relative state data between the vehicle and the preceding vehicle, providing the vehicle with a benchmark independent of the lane-changing decision cost. When the vehicle cannot directly observe the target lane risk due to obstruction by the preceding vehicle, the cost of maintaining the original lane only reflects the longitudinal interaction risk between the vehicle and the preceding vehicle, avoiding decision-making bias caused by misjudgment of the preceding vehicle's turning-back. By using a preset calibration coefficient as a fixed benchmark point and scaling it in conjunction with a risk adjustment factor, the rationality of the cost of maintaining the original lane is ensured under different operating conditions, while also ensuring that the cost of maintaining the original lane increases synchronously when the interaction between the vehicle and the preceding vehicle is urgent.

[0106] In some instances, the target driving behavior of the vehicle is determined based on the relationship between the cost of lane-changing decisions and the cost of maintaining the original lane, including: When the cost of making a lane change decision is less than the cost of maintaining the original lane decision, the target driving behavior is determined to be a lane change. When the cost of changing lanes is greater than or equal to the cost of staying in the original lane, the target driving behavior is determined to be staying in the original lane.

[0107] For example, when the cost of the lane-changing decision is less than the cost of the decision to stay in the original lane, it indicates that the expected risk of continuing to change lanes is lower than the risk of staying in the original lane after abandoning the lane change. In this case, the target driving behavior of the vehicle is determined to be to change lanes. When the cost of the lane-changing decision is greater than or equal to the cost of the decision to stay in the original lane, it indicates that the risk of continuing to change lanes is not lower than or even exceeds the risk of staying in the original lane. In this case, the target driving behavior is determined to be to stay in the original lane. By comparing the magnitudes of the two decision costs, the vehicle selects the target driving behavior based on the principle of risk minimization, avoiding the limitations of simply relying on fixed thresholds or a single risk indicator for decision-making.

[0108] Please see Figure 2 This is a schematic diagram of a decision-making device for a scenario where two vehicles change lanes simultaneously, provided in an embodiment of this application. The device includes: Risk scoring unit 21 is used to determine the risk sub-score of the preceding vehicle in multiple dimensions based on the motion state data of the preceding vehicle; The comprehensive scoring unit 22 is used to perform a weighted summation of the risk sub-scores under multiple dimensions to obtain a comprehensive risk score. Lane change cost unit 23 is used to determine the lane change decision cost based on the product of the relative state data of the vehicle and the preceding vehicle and the comprehensive risk score; The original lane cost unit 24 is used to determine the decision cost of maintaining the original lane based on the relative state data between the vehicle and the vehicle in front. The behavior decision unit 25 is used to determine the target driving behavior of the vehicle based on the relationship between the cost of lane changing and the cost of maintaining the original lane.

[0109] Please see Figure 3 This application also provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor. When the processor 320 executes the computer program 311, it implements the steps of a decision-making method for a scenario where two vehicles change lanes simultaneously.

[0110] Since the electronic device described in this embodiment is the device used by the decision-making device for the scenario of two vehicles changing lanes simultaneously in the embodiment of this application, those skilled in the art can understand the specific implementation method and various variations of the electronic device in this embodiment based on the method described in the embodiment of this application. Therefore, how the electronic device implements the method in the embodiment of this application will not be described in detail here. Any device used by those skilled in the art to implement the method in the embodiment of this application is within the scope of protection of this application.

[0111] In practice, when the computer program 311 is executed by the processor, it can implement any of the embodiments corresponding to the first aspect.

[0112] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0113] Those skilled in the art will understand that embodiments of this application can provide methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-readable storage media containing computer-readable program code.

[0114] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0115] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0116] These computer program instructions can also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0117] This application also provides a computer program product, which includes computer software instructions that, when executed on a processing device, cause the processing device to perform... Figure 1 The flowchart of a decision-making method for a scenario where two vehicles change lanes simultaneously, as described in the corresponding embodiment.

[0118] A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, computer instructions may be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium may be any usable medium that a computer can store or a data storage device such as a server or data center that integrates one or more usable media. The usable medium may be a magnetic medium, an optical medium, or a semiconductor medium, etc.

[0119] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0120] In the several embodiments provided in this application, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Furthermore, the mutual couplings or direct couplings or communication connections shown or discussed may be indirect couplings or communication connections through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0121] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in the form of hardware and / or software functional units.

[0122] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory, magnetic disks, or optical disks.

[0123] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

[0124] Although preferred embodiments have been described in this specification, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications that fall outside the scope of this specification.

[0125] Obviously, those skilled in the art can make various modifications to this specification without departing from its spirit and scope. Therefore, this specification also intends to include any modifications that fall within the scope of the claims and their equivalents.

Claims

1. A decision-making method for a scenario where two vehicles change lanes simultaneously, characterized in that, include: Based on the motion state data of the preceding vehicle, a risk sub-score for the preceding vehicle is determined in multiple dimensions; The risk sub-scores under the multiple dimensions are weighted and summed to obtain the comprehensive risk score. The cost of lane change decision is determined by multiplying the relative state data of the vehicle and the vehicle in front by the comprehensive risk score. Based on the relative state data between the vehicle and the vehicle in front, the cost of maintaining the original lane is determined; The target driving behavior of the vehicle is determined based on the relationship between the lane change decision cost and the lane-keeping decision cost.

2. The method according to claim 1, characterized in that, The multiple dimensions include yaw rate dimension, lateral position dimension, longitudinal deceleration dimension, and turn signal dimension.

3. The method according to claim 2, characterized in that, The risk sub-scores include yaw rate risk sub-scores, lateral position risk sub-scores, longitudinal deceleration risk sub-scores, and turn signal risk sub-scores. The determination of the risk sub-scores of the preceding vehicle across multiple dimensions based on its motion state data includes: Based on the relationship between the yaw rate of the preceding vehicle and a preset yaw rate threshold, and the relationship between the rate of change of the yaw rate of the preceding vehicle and a preset rate of change threshold, the yaw rate risk sub-score is determined. Based on the relationship between the lateral speed of the preceding vehicle and a preset lateral speed threshold, and the relationship between the lateral distance of the preceding vehicle from the original lane centerline and a preset lateral position offset threshold, the lateral position risk sub-score is determined. Based on the relationship between the longitudinal deceleration of the preceding vehicle and a preset longitudinal deceleration threshold, and whether there is an obstacle in front of the preceding vehicle, the longitudinal deceleration risk sub-score is determined. A risk sub-score for the turn signal is determined based on whether the turn signal of the preceding vehicle changes state.

4. The method according to claim 3, characterized in that, The preset yaw rate threshold includes a first yaw rate threshold and a second yaw rate threshold; the preset rate of change threshold includes a first rate of change threshold and a second rate of change threshold; the determination of the yaw rate risk sub-score based on the relationship between the yaw rate of the preceding vehicle and the preset yaw rate threshold, and the relationship between the rate of change of the yaw rate of the preceding vehicle and the preset rate of change threshold, includes: The absolute value of the yaw rate of the preceding vehicle is compared with the absolute values ​​of the first yaw rate threshold and the second yaw rate threshold. Based on the comparison result, a first normalized risk value is determined, wherein the absolute value of the first yaw rate threshold is less than the absolute value of the second yaw rate threshold. The absolute value of the rate of change of the yaw rate of the preceding vehicle is compared with the absolute values ​​of the first rate of change threshold and the second rate of change threshold. A second normalized risk value is determined based on the comparison result, wherein the absolute value of the first rate of change threshold is less than the absolute value of the second rate of change threshold. Calculate the arithmetic mean of the first normalized risk value and the second normalized risk value, and determine the arithmetic mean as the yaw rate risk sub-score.

5. The method according to claim 4, characterized in that, The step of determining the first normalized risk value based on the comparison results includes: When the absolute value of the yaw rate of the vehicle in front is less than or equal to the absolute value of the first yaw rate threshold, the first normalized risk value is determined to be 0. When the absolute value of the yaw rate of the vehicle in front is greater than or equal to the absolute value of the second yaw rate threshold, the first normalized risk value is determined to be 1. When the absolute value of the yaw rate of the preceding vehicle is greater than the absolute value of the first yaw rate threshold and less than the absolute value of the second yaw rate threshold, the yaw rate of the preceding vehicle is calculated by linear interpolation to obtain the first normalized risk value.

6. The method according to claim 3, characterized in that, The preset lateral speed threshold includes a first lateral speed threshold and a second lateral speed threshold. The determination of the lateral position risk sub-score based on the relationship between the lateral speed of the preceding vehicle and the preset lateral speed threshold, and the relationship between the lateral distance of the preceding vehicle from the original lane centerline and the preset lateral position offset threshold, includes: The absolute value of the lateral speed of the preceding vehicle is compared with the absolute values ​​of the first lateral speed threshold and the second lateral speed threshold. A third normalized risk value is determined based on the comparison result, wherein the absolute value of the first lateral speed threshold is less than the absolute value of the second lateral speed threshold. The absolute value of the lateral distance between the preceding vehicle and the original lane centerline is compared with the magnitude of the preset lateral position offset threshold and half a preset lane width. A fourth normalized risk value is determined based on the comparison result, wherein the preset lateral position offset threshold is less than half a preset lane width. Calculate the arithmetic mean of the third normalized risk value and the fourth normalized risk value, and determine the arithmetic mean as the lateral position risk sub-score.

7. The method according to claim 3, characterized in that, The preset longitudinal deceleration threshold includes a first longitudinal deceleration threshold and a second longitudinal deceleration threshold. The determination of the longitudinal deceleration risk sub-score based on the relationship between the longitudinal deceleration of the preceding vehicle and the preset longitudinal deceleration threshold, and whether there is an obstacle in front of the preceding vehicle, includes: The absolute value of the longitudinal deceleration of the preceding vehicle is compared with the absolute values ​​of the first longitudinal deceleration threshold and the second longitudinal deceleration threshold. A fifth normalized risk value is determined based on the comparison result, wherein the absolute value of the first longitudinal deceleration threshold is less than the absolute value of the second longitudinal deceleration threshold. Based on whether there is an obstacle in front of the vehicle, an obstacle influence factor is determined, wherein the value of the obstacle influence factor ranges from 0 to 1; Calculate the difference between the preset constant and the obstacle influence factor, and determine the difference as the difference factor; Calculate the product of the fifth normalized risk value and the difference factor, and determine the product as the longitudinal deceleration risk sub-score.

8. The method according to claim 3, characterized in that, The determination of the turn signal risk sub-score based on whether the turn signal of the preceding vehicle has undergone a state change includes: Detect whether the turn signal of the vehicle in front changes from on to off, or from off to on, or from on one side to on the other. When a change in the state of the turn signal is detected, the risk sub-score of the turn signal is determined to be 1; When no change in the state of the turn signal is detected, the risk sub-score of the turn signal is determined to be 0.

9. The method according to claim 3, characterized in that, The weighted summation of the risk sub-scores under the multiple dimensions to obtain the comprehensive risk score includes: The comprehensive risk score is determined by performing a weighted summation based on the yaw rate risk sub-score, the lateral position risk sub-score, the longitudinal deceleration risk sub-score, the turn signal risk sub-score, and their respective weight coefficients.

10. The method according to claim 1, characterized in that, The determination of lane-changing decision cost based on the product of the relative state data between the vehicle and the vehicle in front and the comprehensive risk score includes: Based on the relative distance, relative speed, and collision time between the vehicle and the vehicle in front, a risk adjustment factor is calculated; Calculate the product of the comprehensive risk score and the risk adjustment factor, and determine the product as the lane change decision cost.

11. The method according to claim 1, characterized in that, The step of determining the cost of maintaining the original lane decision based on the relative state data between the vehicle and the vehicle in front includes: Based on the relative distance, relative speed, and collision time between the vehicle and the vehicle in front, a risk adjustment factor is calculated; Obtain preset calibration coefficients obtained through real vehicle calibration. These preset calibration coefficients are used to characterize the baseline risk level when the vehicle maintains its original lane. Calculate the product of the preset calibration coefficient and the risk adjustment factor, and determine the product as the cost of the decision to maintain the original lane.

12. The method according to claim 10 or 11, characterized in that, The risk adjustment factor is calculated based on the relative distance, relative speed, and collision time between the vehicle and the vehicle in front, including: Calculate the difference between the relative distance between the vehicle and the vehicle in front and the minimum tolerance distance, and determine the reciprocal of the difference as the first risk adjustment factor. When the relative distance is less than or equal to the minimum tolerance distance, the first risk adjustment factor is set to a preset maximum value. Calculate the value of an exponential function with a base of natural constant and an exponent of the ratio of the absolute value of the relative speed between the vehicle and the vehicle in front to a preset speed threshold, and determine the value of the exponential function as the second risk adjustment factor; Calculate the larger of the collision time between the vehicle and the vehicle in front and the minimum acceptable collision time, and divide 1 by the larger value to determine the third risk adjustment factor; The risk adjustment factor is determined by weighted summation of the first risk adjustment factor, the second risk adjustment factor, and the third risk adjustment factor.

13. A decision-making device for a scenario where two vehicles change lanes simultaneously, characterized in that, include: The risk scoring unit is used to determine the risk sub-scores of the preceding vehicle in multiple dimensions based on the motion state data of the preceding vehicle. The comprehensive scoring unit is used to perform a weighted summation of the risk sub-scores under the multiple dimensions to obtain a comprehensive risk score. The lane change cost unit is used to determine the lane change decision cost based on the product of the relative state data between the vehicle and the preceding vehicle and the comprehensive risk score. The original lane cost unit is used to determine the decision cost of maintaining the original lane based on the relative state data between the vehicle and the vehicle in front. A behavior decision unit is used to determine the target driving behavior of the vehicle based on the relationship between the lane change decision cost and the lane-keeping decision cost.

14. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program stored in the memory to implement the steps of a decision-making method for a two-vehicle lane-changing scenario as described in any one of claims 1 to 12.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the decision-making method for a scenario where two vehicles change lanes simultaneously as described in any one of claims 1 to 12.

Images