Driving scenario determination method and system for a vehicle

Abstract

Embodiments of the present application provide a driving scene determination method and system for a vehicle. The method comprises: obtaining a target time period after an interaction condition between the vehicle and at least one obstacle is met, and dividing the target time period into a plurality of time windows, wherein the target time period is used to represent a time range in which the vehicle and the obstacle interact; obtaining state information of the vehicle and the obstacle in the plurality of time windows respectively, wherein the state information is used to represent a motion state of the vehicle and the obstacle in the time window; determining at least one interaction indicator of the vehicle and the obstacle in the time window based on the state information, wherein the interaction indicator is used to represent a dynamic evolution relationship of the interaction behavior in the time-space dimension; determining an interaction type to which the interaction behavior belongs based on the interaction indicator; and determining a driving scene in which the vehicle is located based on the interaction type. The present application solves the technical problem of low accuracy of determining the driving scene of the vehicle.

Description

Technical Field

[0001] This application relates to the field of intelligent driving, and more specifically, to a method and system for determining the driving scenario of a vehicle. Background Technology

[0002] Currently, related technologies often rely on threshold judgment methods based on a single frame or a single time window to identify vehicle interaction behavior and driving scenarios, which leads to problems such as missed detections, false detections, and blurred driving scenario boundaries in multi-vehicle collaboration and progressive evolution scenarios.

[0003] Meanwhile, related technologies often rely on manually set fixed rules to determine the vehicle's interactive behavior. Since manually set fixed rules lack the ability to model the temporal evolution trend of interactive behavior, this leads to the technical problem of low accuracy in determining the vehicle's driving scenario.

[0004] There is currently no good solution to the above problems. Summary of the Invention

[0005] This application provides a method and system for determining a vehicle's driving scenario, to at least solve the technical problem of the accuracy of determining the vehicle's driving scenario.

[0006] According to one aspect of the embodiments of this application, a method for determining a vehicle's driving scenario is provided. The method includes: acquiring a target time period after the interaction conditions between the vehicle and at least one obstacle are met; dividing the target time period into multiple time windows, wherein the target time period is used to represent the time range in which the interaction behavior between the vehicle and the obstacle occurs; acquiring state information of the vehicle and the obstacle within the multiple time windows, wherein the state information is used to represent the motion state of the vehicle and the obstacle within the time windows; determining at least one interaction index between the vehicle and the obstacle within the time windows based on the state information, wherein the interaction index is used to characterize the dynamic evolution relationship of the interaction behavior in the spatiotemporal dimension; determining the interaction type to which the interaction behavior belongs based on the interaction index; and determining the driving scenario in which the vehicle is located based on the interaction type.

[0007] Furthermore, based on interaction metrics, the interaction type of the interaction behavior is determined, including: combining multiple interaction metrics in chronological order to obtain a feature sequence, wherein the feature sequence is used to reflect the trend of interaction behavior over time; comparing the same type of feature sequences within multiple time windows to obtain comparison results; and determining the interaction type based on the comparison results.

[0008] Furthermore, the interaction metrics include a lateral distance metric, which represents the evolution of the distance between the vehicle and the obstacle in the lateral direction. Based on the comparison results, the interaction type is determined, including: in response to the number of time windows being three, the feature sequences within the first time window, the feature sequences within the second time window, and the feature sequences within the third time window are determined respectively; in response to the comparison result that the lateral distance metric within the first time window is greater than the lateral distance metric within the second time window, and the lateral distance metric within the second time window is less than the lateral distance metric within the third time window, the interaction type is determined to be the first interaction type, wherein the first interaction type is used to indicate that the vehicle and the obstacle are close to each other in the first time window, run parallel in the second time window, and move away from each other in the third time window.

[0009] Furthermore, based on the interaction type, the driving scenario in which the vehicle is located is determined, including: in response to the interaction type being the first interaction type and the vehicle being in a deceleration state, the driving scenario is determined to be a defensive deceleration scenario.

[0010] Furthermore, the interaction metrics include a longitudinal collision time metric, a directional metric, and a speed metric. The longitudinal collision time metric represents the time required for the vehicle and obstacle to first collide in the longitudinal direction. The speed metric represents the closing velocity or discrete velocity between the vehicle and obstacle. The directional metric represents the degree of consistency or divergence in the vehicle's driving direction. Based on the comparison results, the interaction type is determined, including: if the longitudinal collision time metric in the second time window is less than the longitudinal collision time metric in the first time window, and the longitudinal collision time metric in the second time window is less than the longitudinal collision time metric in the third time window, and the speed metric is negative, the interaction type is determined to be the second interaction type, which indicates that there is a potential cutting-in behavior between the vehicle and the obstacle; if the directional metric gradually increases in the order of the first time window, the second time window, and the third time window, the interaction type is determined to be the third interaction type, which indicates that there is a bypassing behavior or lane-changing behavior between the vehicle and the obstacle.

[0011] Furthermore, the interaction metrics include a lateral collision time metric, which represents the time required for the vehicle and obstacle to first collide in the lateral direction. Based on the interaction metrics, the interaction type of the interaction behavior is determined, including: in response to the lateral collision time metric showing a decreasing trend in a continuous time window, the interaction type is determined to be a fourth interaction type, where the fourth interaction type represents the vehicle and obstacle moving closer to each other in the lateral direction; in response to the lateral collision time metric showing an increasing trend in a continuous time window, the interaction type is determined to be a fifth interaction type, where the fifth interaction type represents the vehicle and obstacle moving away from each other in the lateral direction.

[0012] Furthermore, based on the interaction type, the driving scenario in which the vehicle is located is determined, including: in response to the interaction type being the fourth interaction type, the driving scenario is determined to be an entry scenario; in response to the interaction type being the fifth interaction type, the driving scenario is determined to be an exit scenario.

[0013] Furthermore, the interaction metrics also include the longitudinal distance metric, which is used to represent the evolution of the distance between the vehicle and the obstacle in the longitudinal direction.

[0014] According to another aspect of the embodiments of this application, a vehicle driving scene determination system is also provided, comprising: a multi-time window construction unit, configured to acquire a target time period after the interaction conditions between the vehicle and at least one obstacle are met, and to divide the target time period into multiple time windows, wherein the target time period is used to represent the time range in which the interaction behavior occurs between the vehicle and the obstacle; a future state acquisition unit, configured to acquire state information of the vehicle and the obstacle within the multiple time windows respectively, wherein the state information is used to represent the motion state of the vehicle and the obstacle within the time window; an interaction index calculation unit, configured to determine at least one interaction index of the vehicle and the obstacle within the time window based on the state information, wherein the interaction index is used to characterize the dynamic evolution relationship of the interaction behavior in the spatiotemporal dimension; a behavior recognition unit, configured to determine the interaction type to which the interaction behavior belongs based on the interaction index; and a scene mining unit, configured to determine the driving scene in which the vehicle is located based on the interaction type.

[0015] Furthermore, the vehicle driving scenario determination system also includes a multi-time window feature construction unit, which combines multiple interaction indicators in chronological order to obtain a feature sequence, wherein the feature sequence is used to reflect the trend of interaction behavior over time.

[0016] According to one aspect of the embodiments of this application, a vehicle driving scenario determination device is provided. The device includes: a first acquisition unit, configured to acquire a target time period after the interaction conditions between the vehicle and at least one obstacle are met, and to divide the target time period into multiple time windows, wherein the target time period is used to represent the time range in which the interaction behavior between the vehicle and the obstacle occurs; a second acquisition unit, configured to acquire state information of the vehicle and the obstacle within the multiple time windows, wherein the state information is used to represent the motion state of the vehicle and the obstacle within the time windows; a first determination unit, configured to determine at least one interaction index between the vehicle and the obstacle within the time windows based on the state information, wherein the interaction index is used to characterize the dynamic evolution relationship of the interaction behavior in the spatiotemporal dimension; a second determination unit, configured to determine the interaction type to which the interaction behavior belongs based on the interaction index; and a third determination unit, configured to determine the driving scenario in which the vehicle is located based on the interaction type.

[0017] According to another aspect of the embodiments of this application, a vehicle is also provided, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods in various embodiments of this application when it runs.

[0018] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0019] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0020] According to another aspect of the embodiments of this application, a computer program product is also provided, including a non-volatile computer-readable storage medium storing a computer program, which, when executed by a processor, implements the methods in various embodiments of this application.

[0021] According to another aspect of the embodiments of this application, a computer program is also provided, which, when executed by a processor, implements the methods of the various embodiments of this application.

[0022] In this embodiment, multiple time windows are obtained by dividing the target time period after the interaction conditions between the vehicle and at least one obstacle are met. Then, the state information of the vehicle and obstacle within each of the multiple time windows is acquired to determine the dynamic evolution relationship of the interaction behavior between the vehicle and obstacle in the spatiotemporal dimension. Based on interaction indicators, the interaction type of the interaction behavior is determined, and based on the interaction type, the driving scenario in which the vehicle is located is determined. This overcomes the limitations of relying on instantaneous thresholds and single-point rules in related technologies, achieving the goal of improving from static judgment to dynamic determination. This solves the technical problem of low accuracy in determining the vehicle's driving scenario and achieves the technical effect of improving the accuracy of determining the vehicle's driving scenario. Attached Figure Description

[0023] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0024] Figure 1 This is a flowchart of a method for determining a vehicle driving scenario according to an embodiment of this application;

[0025] Figure 2This is a flowchart of a vehicle behavior and scene mining method based on multiple time windows according to an embodiment of this application;

[0026] Figure 3 This is a schematic diagram of a vehicle behavior and scene mining system based on multiple time windows according to an embodiment of this application;

[0027] Figure 4 This is a schematic diagram of a typical scenario with multiple time windows according to an embodiment of this application;

[0028] Figure 5 This is a schematic diagram of a vehicle driving scenario determination system according to an embodiment of this application;

[0029] Figure 6 This is a schematic diagram of a vehicle driving scenario determination device according to an embodiment of this application. Detailed Implementation

[0030] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0031] It should be noted that the terms "first," "second," etc., 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 of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises 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.

[0032] According to an embodiment of this application, an embodiment of a method for determining a vehicle's driving scenario is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0033] This embodiment provides a method for determining the driving scenario of a vehicle. Figure 1This is a flowchart of a method for determining a vehicle driving scenario according to an embodiment of this application, such as... Figure 1 As shown, the process includes the following steps.

[0034] Step S102: Obtain the target time period after the interaction conditions between the vehicle and at least one obstacle are met, and divide the target time period into multiple time windows.

[0035] In the technical solution provided by step S102 of this application, the target time period can be used to represent the time range in which interaction occurs between the vehicle and the obstacle.

[0036] In this embodiment, the vehicle can be a self-propelled vehicle. The obstacle can be an interactive vehicle.

[0037] Optionally, to provide an evolution window for the interaction behavior in subsequent multi-time-window analysis, a target time period can be obtained after the interaction conditions between the vehicle and at least one obstacle are met. The target time period is dynamic and adaptive. For example, a typical "approach then detour" interaction might last 12 seconds, while an "emergency lane change" interaction might only last 6 seconds. By identifying the start and end times of the interaction, the complete lifecycle of the interaction can be extracted. This ensures that each analyzed target time period contains a complete interaction motivation and response process, avoiding the problem of incomplete interaction fragments caused by artificial truncation.

[0038] Optionally, after obtaining the target time period, a non-uniform, phased, semantically aligned segmentation strategy can be adopted to divide the target time period into multiple time windows. For example, a first time window, a second time window, and a third time window.

[0039] Optionally, the first time window (e.g., 0–3 seconds) can represent the period when the intention of the interaction becomes apparent, used to capture the initial trend of the interaction. For example, the vehicle begins to approach laterally or decelerate longitudinally. The second time window (e.g., 3–6 seconds) can represent the critical period of the interaction's evolution, used to observe whether the interaction has entered a stable standoff or cooperative state. For example, whether the vehicle and the obstacle are running parallel or whether speed synchronization occurs. The third time window (e.g., 6–10 seconds) can represent the period for confirming the outcome of the interaction, used to determine whether the final result of the interaction is to cut in, detour, avoid, or restore a safe distance.

[0040] Optionally, the length of each of the above time windows can be engineered to align with the reaction cycle (e.g., the driver's 2–5 second decision cycle) based on the understanding of typical driving behavior of the vehicle, ensuring that the time window is aligned with the natural transition point of the interaction behavior phase.

[0041] Optionally, the aforementioned interaction conditions can be triggered based on a continuous, non-random coupling relationship between the vehicle and at least one obstacle in terms of spatial and motion states. For example, the lateral distance of the vehicle is less than 4 meters within 5 consecutive frames and the absolute value of the vehicle's relative lateral velocity is greater than 0.5 m / s, or the longitudinal distance of the vehicle is continuously decreasing and the longitudinal collision time of the vehicle shows a decreasing trend, thereby ensuring that the captured target time period is a truly behaviorally meaningful interaction segment, rather than a brief noise jitter or irrelevant parallel passage.

[0042] In this embodiment of the application, the target time period after the interaction conditions between the vehicle and at least one obstacle are met is obtained, and the target time period is divided into multiple time windows. This overcomes the limitation that obtaining only instantaneous data within a single frame or a very short window (e.g., 100ms) leads to low accuracy in determining the subsequent state information.

[0043] Step S104: Obtain the status information of the vehicle and obstacles within multiple time windows.

[0044] In the technical solution provided by step S104 of this application, the state information can be used to represent the motion state of the vehicle and the obstacle within the time window.

[0045] In this embodiment, the aforementioned state information can be future state information. The state information of the vehicle and obstacles within multiple time windows can include predicted values ​​within multiple future time windows, such as the vehicle's position, speed, acceleration, and heading angle at a certain moment.

[0046] Optionally, the aforementioned status information can be obtained through vehicle-side sensors (such as LiDAR and cameras) or through joint prediction using high-precision maps and historical trajectory databases in the cloud. This is merely an example, and no specific restrictions are placed on the method of obtaining status information here.

[0047] In this embodiment of the application, the status information of the vehicle and the obstacle in multiple time windows is obtained respectively, so as to provide accurate and comprehensive information for the subsequent determination of at least one interaction indicator of the vehicle and the obstacle in multiple time windows.

[0048] Step S106: Based on the state information, determine at least one interaction indicator between the vehicle and the obstacle within the time window.

[0049] In the technical solution provided by step S106 of this application, the interaction index can be used to characterize the dynamic evolution relationship of interaction behavior in the spatiotemporal dimension.

[0050] In this embodiment, the aforementioned interaction metrics may include, but are not limited to: minimum lateral distance, minimum longitudinal distance, lateral collision time, longitudinal collision time, relative speed, heading difference, etc.

[0051] Optionally, minimum lateral distance and lateral collision time can jointly reflect the lateral approach trend between the vehicle and the obstacle. Minimum longitudinal distance and longitudinal collision time can jointly reflect the following or cutting-in intentions between the vehicle and the obstacle. Relative speed can reflect the motion coordination between the vehicle and the obstacle. Heading difference can reveal the consistency or divergence of directional intentions between the vehicle and the obstacle.

[0052] Optionally, the aforementioned interactive indicators can be calculated independently within their respective time windows, forming an indicator sequence that evolves over time.

[0053] Alternatively, taking the lateral collision time as an example, the determination of the lateral collision time is not directly calculated by dividing the current lateral distance by the lateral relative velocity. Instead, it can be based on predictions of future trajectories. For instance, within each time window, the remaining time at which the lateral centerline between the vehicle and the obstacle is closest can be calculated through linear extrapolation or minimum distance search. This method avoids jumps in lateral collision time caused by instantaneous speed changes, and more realistically reflects the evolution trend of lateral collision risk within that stage.

[0054] In this embodiment of the application, the completeness of the interaction behavior is restored by combining multiple interaction indicators and multiple time windows, overcoming the limitations of traditional methods that rely on a single threshold and ignore the phased and continuous nature of the interaction behavior.

[0055] Step S108: Based on the interaction metrics, determine the interaction type to which the interaction behavior belongs.

[0056] In the technical solution provided by step S108 of this application, after determining at least one interaction index between the vehicle and the obstacle within the time window based on the state information, the interaction type of the interaction behavior can be determined based on the interaction index, thereby transforming the originally discrete, fuzzy, and easily misjudged motion state into an interaction behavior category with clear driving semantics.

[0057] In this embodiment, since the interaction type is jointly determined by the evolutionary trends of multiple time windows and multiple interaction indicators, it is possible to effectively distinguish interactive behaviors that are similar in appearance but have completely different intentions.

[0058] For example, "another vehicle cutting in" and "another vehicle changing lanes after parallel driving" may have almost the same distance and speed parameters at a certain moment, but "another vehicle cutting in" shows a continuous decrease in lateral collision time and no significant change in longitudinal speed, while "another vehicle changing lanes after parallel driving" shows a decrease followed by an increase in lateral collision time and a continuous increase in heading difference. By capturing this subtle but crucial difference, "another vehicle cutting in" and "another vehicle changing lanes after parallel driving" can be accurately separated.

[0059] Optionally, the interaction type can be "parallel approach", "intent to cut in", "defensive detour", "lane change coordination", "cut in blocked", etc., which are only examples and are not specifically limited here.

[0060] In this embodiment of the application, the interaction type of an interaction behavior is determined based on interaction metrics. Each interaction behavior can be abstracted into a dynamic trajectory composed of interaction metrics from multiple time windows, thereby enabling the judgment of the interaction type rather than a simple threshold trigger.

[0061] Step S110: Determine the driving scenario in which the vehicle is located based on the interaction type.

[0062] In the technical solution provided in step S110 of this application, after determining the interaction type of the interaction behavior based on interaction indicators, the driving scenario in which the vehicle is located can be determined based on the interaction type. For example, multiple continuous or parallel interaction types can be mapped into a complete driving scenario with practical engineering significance and autonomous driving training value.

[0063] In this embodiment, based on the interaction type, feature sequences within multiple time windows can be determined, thereby determining the driving scenario in which the vehicle is located based on the feature sequences.

[0064] Optionally, if the feature sequence is parallel approach → lateral collision time continuously decreases → heading difference steadily increases → longitudinal relative speed approaches zero, then the "lane change coordination successful" driving scenario is triggered.

[0065] Optionally, if the feature sequence is parallel approach → lateral collision time continuously decreases → heading difference continuously increases → longitudinal distance does not shorten and relative speed remains negative, it can be determined as a driving scenario of "cutting intention is blocked and other vehicles actively give up changing lanes".

[0066] In this embodiment, the driving scenario of the vehicle is determined based on the interaction type within multiple time windows, overcoming the limitation of low accuracy in determining the driving scenario caused by determining the driving scenario only at an instant in related technologies.

[0067] It should be noted that in the vehicle driving scenario determination method of this embodiment, the vehicle can also interact with roadside equipment and terminal equipment. Optionally, the vehicle can send an information subscription request to the roadside equipment. This message subscription request can include specific types of information that the vehicle needs to receive, such as road conditions, traffic signal status, and obstacle warnings ahead. In response to the information subscription request, the roadside equipment can send roadside perception information to the vehicle. For example, the roadside equipment will filter out roadside perception information that meets the vehicle's needs based on its own perception capabilities and stored information, and send it to the vehicle at a certain frequency. In addition to communicating with the roadside equipment, the vehicle can also receive driving scenario switching instructions transmitted by the terminal equipment through the network. For example, the driving scenario switching instructions can be used to switch the vehicle to energy-saving mode, sport mode, autonomous driving mode, etc., so that the vehicle can adapt to new driving scenarios.

[0068] Through steps S102 to S110, the target time period after the interaction conditions between the vehicle and at least one obstacle are met is divided into multiple time windows. Then, the state information of the vehicle and the obstacle within each of these multiple time windows is acquired to determine the dynamic evolution relationship of the interaction behavior between the vehicle and the obstacle in the spatiotemporal dimension. Based on interaction indicators, the interaction type of the interaction behavior is determined, and based on the interaction type, the driving scenario in which the vehicle is located is determined. This overcomes the limitations of related technologies that rely on instantaneous thresholds and single-point rules, achieving the goal of improving from static judgment to dynamic determination. This solves the technical problem of low accuracy in determining the vehicle's driving scenario and achieves the technical effect of improving the accuracy of determining the vehicle's driving scenario.

[0069] The above-mentioned method of this application will be further described below.

[0070] As an optional implementation, step S108, determining the interaction type of the interaction behavior based on interaction indicators, includes: combining multiple interaction indicators in chronological order to obtain a feature sequence, wherein the feature sequence is used to reflect the trend of interaction behavior changing over time; comparing the same type of feature sequences within multiple time windows to obtain comparison results; and determining the interaction type based on the comparison results.

[0071] In this embodiment, by combining multiple interaction indicators in chronological order to form a feature sequence, instead of treating each interaction indicator as an independent instantaneous value, the values ​​of multiple interaction indicators within multiple time windows are concatenated in chronological order to form a feature sequence with a temporal structure. Then, by performing a horizontal comparison of the evolution patterns of the feature sequence, accurate identification of the interaction type can be achieved.

[0072] For example, multiple interaction metrics can be combined in chronological order to obtain feature sequences (e.g., minimum lateral distance sequence, minimum longitudinal distance sequence, lateral collision time series, longitudinal collision time series, relative velocity sequence, and heading difference sequence). Subsequently, the nature of the behavior can be revealed by comparing patterns in similar feature sequences within multiple time windows.

[0073] Optionally, the comparison described above is not a numerical comparison, but a structural identification of the shape of the feature sequence. For example, when the minimum lateral distance sequence shows a pattern of "high-low-medium", the lateral collision time sequence shows a pattern of "long-short-long", and the heading difference sequence remains stable, the interaction type can be determined as "parallel approach followed by active departure" by matching the pattern with a preset template. When the longitudinal collision time sequence shows a pattern of "stable-sudden drop-continuously low value" and the relative velocity sequence changes from positive to negative, it is identified as "high-risk intention to enter". If the heading difference sequence shows a trend of "slight increase-continuous acceleration-tends to stabilize", and the minimum lateral distance sequence does not decrease significantly, it can be determined as "clear intention to coordinate lane change".

[0074] In existing technologies, each interaction type requires a complex set of conditions to be manually defined as rules. Once the driving scenario changes slightly (e.g., different vehicle speeds or different road curvatures), the rules become invalid. However, the embodiments of this application define the interaction type as a learnable and generalizable morphological template by comparing the patterns of feature sequences, thereby realizing the intelligent evolution of interaction behavior.

[0075] As an optional implementation, the interaction index includes a lateral distance index, which represents the evolution of the distance between the vehicle and the obstacle in the lateral direction. Based on the comparison results, the interaction type is determined, including: in response to the number of time windows being three, determining the feature sequences in the first time window, the second time window, and the third time window within the three time windows respectively; in response to the comparison result that the lateral distance index in the first time window is greater than the lateral distance index in the second time window, and the lateral distance index in the second time window is less than the lateral distance index in the third time window, determining the interaction type as the first interaction type, wherein the first interaction type is used to indicate that the vehicle and the obstacle are close to each other in the first time window, run parallel in the second time window, and move away from each other in the third time window.

[0076] In this embodiment, the aforementioned lateral distance index can be the minimum lateral distance.

[0077] Optionally, if there are three time windows, a multi-time window construction unit deployed in the vehicle can be used to obtain the feature sequence within the first time window (e.g., 0-3 seconds), the feature sequence within the second time window (e.g., 3-6 seconds), and the feature sequence within the third time window (e.g., 6-10 seconds) of the three time windows.

[0078] Optionally, if the horizontal distance indicator within the first time window (e.g., ) is greater than the horizontal distance indicator within the second time window (e.g., ), and the horizontal distance index in the second time window is smaller than the horizontal distance index in the third time window (for example, This indicates that the interaction type is the first interaction type. That is, the vehicle and the obstacle approach each other in the first time window, run parallel to each other in the second time window, and move away from each other in the third time window.

[0079] In this embodiment, by comparing lateral distance indicators across three time windows, "single-point anomalies" are elevated to "global trends." Only when a complete rhythm is determined—where elements are close together in the first time window, parallel in the second time window, and far apart in the third time window—is it classified as the first interaction type. This first interaction type is a low-risk, predictable interaction category that aligns with human driving habits.

[0080] As an optional implementation, step S110, based on the interaction type, determines the driving scenario in which the vehicle is located, including: in response to the interaction type being a first interaction type and the vehicle being in a deceleration state, determining the driving scenario as a defensive deceleration scenario.

[0081] In this embodiment, the vehicle being in a deceleration state can be confirmed by comprehensively considering multiple dimensions of information, such as the vehicle's longitudinal acceleration being continuously negative, the vehicle speed decreasing steadily, and the throttle opening returning to zero.

[0082] Optionally, when the minimum lateral distance sequence is represented as This can represent "approaching - parallel - moving away". At this time, if the vehicle is decelerating, that is, if the vehicle's deceleration action is detected, the driving scenario can be identified as a defensive deceleration scenario.

[0083] In this embodiment, the vehicle's proactive response is incorporated into the driving scenario definition system. Defensive deceleration scenarios are not due to inherent danger, but rather to the perception of potential danger and proactive avoidance, thus accurately mapping typical driver behavior patterns in complex traffic. That is, when a driver perceives an approaching obstacle without a clear intention to change lanes, they can proactively reduce speed and increase longitudinal distance to allow for safety margins. Through the joint judgment of interaction type and vehicle driving status, accurate determination of driving scenarios is achieved.

[0084] As an optional implementation, the interaction metrics include a longitudinal collision time metric, a direction metric, and a speed metric. The longitudinal collision time metric represents the time required for the vehicle and the obstacle to first collide in the longitudinal direction. The speed metric represents the closing velocity or discrete velocity between the vehicle and the obstacle. The direction metric represents the degree of consistency or divergence in the vehicle's driving direction. Based on the comparison results, the interaction type is determined, including: if the longitudinal collision time metric in the second time window is less than the longitudinal collision time metric in the first time window, and the longitudinal collision time metric in the second time window is less than the longitudinal collision time metric in the third time window, and the speed metric is negative, the interaction type is determined to be a second interaction type, where the second interaction type indicates a potential cutting-in behavior between the vehicle and the obstacle; if the direction metric gradually increases in the order of the first time window, the second time window, and the third time window, the interaction type is determined to be a third interaction type, where the third interaction type indicates a detour or lane-changing behavior between the vehicle and the obstacle.

[0085] In this embodiment, the aforementioned longitudinal collision time index can be the longitudinal collision time. The aforementioned directional index can be the heading difference. The aforementioned speed index can be the relative speed.

[0086] Alternatively, the longitudinal collision time index reflects the remaining time a vehicle has to travel in the longitudinal direction from the potential point of impact. A smaller longitudinal collision time index indicates a more imminent collision risk.

[0087] Optionally, a negative speed index indicates that the relative motion between the vehicle and the obstacle in the longitudinal direction is a closed loop, meaning that the vehicle and the obstacle are moving closer to each other.

[0088] Optionally, the direction indicator describes the degree of deviation of the obstacle from the vehicle's heading. A continuous increase in the direction indicator means that the obstacle is actively deviating from its original driving direction.

[0089] Optionally, if the longitudinal collision time index in the second time window is less than that in the first time window, and the longitudinal collision time index in the second time window is less than that in the third time window—that is, when the longitudinal collision time index is minimum in the second window and the speed index (e.g., relative speed) is negative—it means that the vehicle has not fully revealed its intention in the first time window (0–3 seconds). However, in the second time window (3–6 seconds), the longitudinal approach speed increases, and the collision risk accumulates rapidly, reaching its peak. Upon entering the third time window (6–10 seconds), although the risk has not been eliminated, the approach trend has not intensified but instead begins to stabilize, which indicates that the vehicle has a potential cutting-in behavior with the obstacle.

[0090] Optionally, when the directional indicator gradually increases in the order of the first, second, and third time windows—that is, when the difference in heading between the vehicle and the obstacle increases continuously and monotonically in the three time windows—the interaction type can be determined to be the third interaction type. This is a detour or lane-changing behavior, meaning the vehicle no longer maintains parallel or cooperative driving with the obstacle, but actively adjusts its heading, gradually deviating from its original lane.

[0091] For example, if a vehicle drives alongside another vehicle for a period of time and then gradually veers 5° to the left, then 8°, then 12°, with the heading difference continuously increasing, even if the lateral distance indicator doesn't change drastically or even increases slightly, it's enough to indicate that the intention is not to "approach" but to "leave the current path." This behavior occurs during high-speed lane changes, merging zone avoidance, or obstacle evasion. Unlike cutting in, the typical characteristic of evasive or lane-changing behavior is that "heading moves first, distance follows," and the risk lies not in collision, but in path conflict.

[0092] Optionally, by observing the continuous upward trend of the direction indicator, it is possible to effectively distinguish between intentional lane changes and simple approach, avoiding misjudging normal lane-changing behavior as a threat, thereby achieving more accurate route planning and more natural traffic coordination.

[0093] In this embodiment, by constructing the timing logic of three interaction indicators, an intelligent response is achieved that triggers when necessary and yields when necessary, thus solving the dual dilemma of false triggering and missed triggering in the autonomous driving system of the vehicle in complex interaction scenarios.

[0094] As an optional implementation, the interaction index includes a lateral collision time index, which represents the time required for the vehicle and the obstacle to first collide in the lateral direction. Step S108 involves determining the interaction type of the interaction behavior based on the interaction index, including: determining the interaction type as a fourth interaction type in response to the lateral collision time index showing a decreasing trend in a continuous time window, wherein the fourth interaction type represents the vehicle and the obstacle approaching each other in the lateral direction; and determining the interaction type as a fifth interaction type in response to the lateral collision time index showing an increasing trend in a continuous time window, wherein the fifth interaction type represents the vehicle and the obstacle moving away from each other in the lateral direction.

[0095] In this embodiment, the aforementioned lateral collision time index can be the lateral collision time. The lateral collision time is not a rough estimate based on a single frame in the traditional sense, but rather, it is based on predictions of future trajectories. Within each time window, the theoretical remaining time required for the vehicle and obstacle to first make contact in the lateral direction is dynamically calculated. This measures the rate of change in the spatial relationship between the vehicle and obstacle in the lateral dimension, providing a safety warning.

[0096] Optionally, if the lateral collision time index shows a decreasing trend in consecutive time windows, for example, a continuous decreasing trend in the first, second, and third time windows, then the interaction type can be determined as the fourth interaction type, that is, the vehicle and the obstacle are approaching each other in the lateral direction.

[0097] Optionally, if the lateral collision time index shows an increasing trend in consecutive time windows—for example, a continuous increasing trend in the first, second, and third time windows—then the interaction type can be determined as the fifth interaction type, meaning the vehicle and the obstacle are moving away from each other in the lateral direction.

[0098] In this embodiment of the application, the above steps enable intent-level recognition of the interaction behavior between vehicles and obstacles in the lateral direction, thereby enhancing the accuracy of interaction type determination.

[0099] As an optional implementation, step S110, based on the interaction type, determines the driving scenario in which the vehicle is located, including: in response to the interaction type being the fourth interaction type, determining the driving scenario as an entry scenario; in response to the interaction type being the fifth interaction type, determining the driving scenario as an exit scenario.

[0100] In this embodiment, if the interaction type is the fourth interaction type, it means that the obstacle is approaching the vehicle in the lateral direction at a continuously accelerating pace, and the spatial relationship is rapidly evolving from a potential threat to an imminent contact. At this time, the driving scenario can be determined as the entry scenario.

[0101] Optionally, if the interaction type is the fifth interaction type, it means that the vehicle has actively or passively begun to move laterally away, which may be due to reversing after a failed lane change, exiting after a failed overtaking maneuver, or actively increasing the distance to avoid an obstacle. The fifth interaction type can be characterized by a continuous increase in longitudinal collision time within multiple windows, a discrete relative speed, and a stable lateral distance, indicating that the obstacle has completed a lane change and actively increased the distance, rather than a temporary avoidance or brief deviation. At this point, the driving scenario can be determined to be a cut-out scenario.

[0102] In this embodiment of the application, the above steps complete the accurate mapping of interactive behavior features to driving scene semantics, thereby enhancing the accuracy of driving scene determination.

[0103] As an optional implementation, the interaction metrics also include a longitudinal distance metric, which is used to represent the evolution of the distance between the vehicle and the obstacle in the longitudinal direction.

[0104] In this embodiment, the longitudinal distance index can be the minimum longitudinal distance, which can be used to characterize the continuous evolution of the distance between the vehicle and the obstacle in the longitudinal direction. It is a supplement and enhancement to the original dynamic index system based on lateral collision time index, longitudinal collision time index, direction index and speed index.

[0105] Optionally, the longitudinal distance index is essentially a dynamic record of the longitudinal spatial relationship between vehicles and obstacles, not an absolute distance value at a single moment, but rather a characteristic sequence change of longitudinal spacing over multiple time windows. For example, This represents the minimum longitudinal distance between the vehicle and the obstacle within the first time window (e.g., 0–3 seconds). This refers to the minimum distance between the vehicle and the obstacle within the second time window (e.g., 3–6 seconds). This is the minimum distance between the vehicle and the obstacle within the third time window (e.g., 6–10 seconds).

[0106] Optionally, the aforementioned feature sequence directly reflects the longitudinal "closer" or "separation" process between the vehicle and the obstacle. If the longitudinal distance index continues to shrink within multiple time windows, it can be confirmed that "continuous approaching behavior exists".

[0107] Optionally, the role of the vertical distance metric in interaction type determination is to provide decisive spatial evidence for "potential intrusion behavior." For example, when the vertical collision time of the second time window is detected to be less than the vertical collision time of the first time window and the vertical collision time of the third time window, and the relative velocity is negative, it can be initially determined as "potential intrusion." However, if the vertical distance sequence shows "..." > < The phrase "the gap first shrinks and then expands" indicates that although the obstacle intended to cut in, it actively gave up during the approach process due to the perception of risk or the vehicle's reaction, forming a "tentative but unsuccessful attempt to cut in".

[0108] Optionally, if the longitudinal distance sequence continues to decrease, i.e., " > > If this is confirmed as "firm intention to engage", then the vehicle's response level can be increased.

[0109] Optionally, longitudinal distance indicators also play a crucial role in identifying detour or lane-changing behaviors. For example, when the directional indicator gradually increases, suggesting an intention to change course, if the longitudinal distance sequence shows a "slowly increasing" trend (…), it indicates a similar effect. < < This indicates that the obstacle is not parallel to the vehicle's lane, but is actively increasing its longitudinal distance while shifting laterally to complete a lane change or bypass maneuver. This is a typical example of proactive avoidance behavior.

[0110] Alternatively, if the longitudinal distance remains stable or even decreases while the direction changes, it can be considered a "forced entry" rather than a "safe detour".

[0111] In the embodiments of this application, the introduction of the longitudinal distance index evolves the multi-time window analysis system from a dynamic trend model to a spatiotemporal behavior map, thereby truly realizing a concrete understanding of the intentions of traffic participants.

[0112] The vehicle driving scenario determination method of this application divides the target time period after the interaction conditions between the vehicle and at least one obstacle are met, resulting in multiple time windows. Then, the state information of the vehicle and the obstacle within each of the multiple time windows is acquired to determine the dynamic evolution relationship of the interaction behavior between the vehicle and the obstacle in the spatiotemporal dimension. Based on interaction indicators, the interaction type of the interaction behavior is determined, and based on the interaction type, the driving scenario in which the vehicle is located is determined. This overcomes the limitations of related technologies that rely on instantaneous thresholds and single-point rules, achieving the goal of improving from static judgment to dynamic determination. This solves the technical problem of low accuracy in determining the vehicle driving scenario and achieves the technical effect of improving the accuracy of determining the vehicle driving scenario.

[0113] The above technical solutions of the embodiments of this application will be further illustrated below with reference to preferred embodiments.

[0114] Currently, with the rapid evolution of autonomous driving technology, data-driven model training places higher demands on high-quality, fine-grained scenario data that covers real-world driving behavior. Examples include defensive deceleration caused by lateral parallel vehicles, detour requirements due to approaching vehicles, lane change coordination and lane change obstruction scenarios, other vehicles cutting in / out, and potential collision risks caused by failed lane changes.

[0115] These scenarios often exhibit strong temporal and multi-agent interaction characteristics, making them a crucial data source for improving the decision-making capabilities of autonomous driving systems. Currently, judgment methods based on single-moment threshold features are commonly used, assessing collision risk or vehicle motion characteristics through instantaneous motion states such as lateral and longitudinal distances, relative speeds, and accelerations.

[0116] In related technologies, existing driving scenario mining mainly relies on threshold judgments at a single moment or within a single time window, which are effective in simple scenarios. However, because threshold judgments at a single moment or within a single time window lack the ability to model the evolution of behavior, they struggle to reflect the changing trends of vehicle behavior over time in complex traffic processes (e.g., multi-vehicle interactions). Furthermore, these methods depend on the design of evaluation metrics, making it difficult to cover the diverse behavioral patterns in real-world roads. In engineering practice, they generally suffer from high false negative rates, high false positive rates, and blurred scene boundaries. Therefore, they are insufficient to meet the demands of large-scale data mining and high-precision behavior recognition.

[0117] Meanwhile, in related technologies, different scenarios often rely on fixed rules for recognition, lacking a description of the behavioral evolution process, making it difficult to extend to other scenarios, resulting in poor generalization ability and insufficient scalability.

[0118] To address the aforementioned issues, this application provides a vehicle behavior and scene mining method based on multiple time windows. By introducing a multi-time window feature extraction and temporal relationship analysis mechanism, the changing trends of vehicle behavior and the interaction process between vehicles are characterized in stages. This solves the problem that existing technologies based on single frames or single time segments cannot reflect the continuous changing patterns of vehicle behavior at different times and are difficult to accurately identify the behavioral evolution process.

[0119] Figure 2 This is a flowchart of a vehicle behavior and scene mining method based on multiple time windows according to an embodiment of this application, such as... Figure 2 As shown, the method may include the following steps.

[0120] Step S201: Divide the time after the target time into multiple future time windows.

[0121] In this embodiment, the time after the target time can be divided into multiple future time windows by using a multi-time window construction unit.

[0122] Step S202: Within each time window, obtain the future position, speed, or acceleration of the vehicle and the target vehicle.

[0123] In this embodiment, the future state acquisition unit can be connected to the multi-time window construction unit to acquire future position, speed or acceleration and other state information of the vehicle and the target vehicle in each time window.

[0124] Step S203: Calculate the interaction metrics between vehicles within each time window.

[0125] In this embodiment, the input terminal of the index calculation unit is connected to the future state acquisition unit, and is used to calculate the interaction index between vehicles within each time window. The interaction index may include at least one of the following: minimum lateral distance, minimum longitudinal distance, lateral collision time, longitudinal collision time, relative speed, or heading difference.

[0126] Step S204: Combine the interaction indicators within multiple time windows in chronological order to form a multi-time window feature sequence that can characterize the trend of vehicle interaction behavior over time.

[0127] In this embodiment, the input end of the feature construction unit is connected to the indicator calculation unit, which is used to combine the interaction indicators in multiple time windows in chronological order and form a multi-time window feature sequence that can characterize the trend of vehicle interaction behavior over time.

[0128] Step S205: Identify vehicle interaction behavior based on indicator change patterns in multi-time window feature sequences.

[0129] In this embodiment, the behavior mining unit is connected to the feature construction unit and is used to identify vehicle interaction behavior based on the indicator change pattern in the feature sequence of multiple time windows, including at least one of parallel approach, cutting in, detouring or lane change coordination.

[0130] Step S206: Determine the corresponding driving scenario based on the identified interactive behaviors.

[0131] In this embodiment, the behavior mining unit can be further used to determine the corresponding driving scenario based on the identified interactive behaviors.

[0132] Figure 3 This is a schematic diagram of a vehicle behavior and scene mining system based on multiple time windows according to an embodiment of this application, such as... Figure 3 As shown, the system may include: a multi-time window construction unit 302, a future state acquisition unit 304, an indicator calculation unit 306, a feature construction unit 308, and a behavior and scene mining unit 310.

[0133] The multi-time window construction unit 302 can be used to divide the future time after the target time into multiple time windows according to a preset time range. The future time can be divided into three time windows, namely the first time window, the second time window, and the third time window. The first time window can be 0 to 3 seconds, the second time window can be 3 to 6 seconds, and the third time window can be 6 to 10 seconds.

[0134] The future state acquisition unit 304 can be connected to the multi-time window construction unit 302 to acquire future state information of the autonomous vehicle and the interactive vehicle within each time window. This future state information includes position, speed, acceleration, and heading angle. In actual data processing, the future states of the autonomous vehicle and the interactive vehicle can be obtained through sensor measurement data, trajectory extrapolation results, or data generated by a simulation system.

[0135] The indicator calculation unit 306, also known as the interaction indicator calculation unit, can be connected to the future state acquisition unit 304 to calculate the interaction indicators between vehicles within each time window. These interaction indicators include minimum lateral distance, minimum longitudinal distance, lateral collision time, longitudinal collision time, relative speed, and heading difference. Minimum lateral distance and minimum longitudinal distance characterize the closest contact trend between the vehicle and interacting vehicles in the lateral and longitudinal directions, respectively. Lateral collision time and longitudinal collision time are calculated based on predicted trajectories and reflect the time required for vehicles to reach a potential collision state in the lateral and longitudinal directions, respectively. Relative speed reflects the closed or discrete speeds between vehicles. Heading difference reflects the degree of consistency or divergence in the vehicles' travel directions.

[0136] Figure 4 This is a schematic diagram of a typical scenario with multiple time windows according to an embodiment of this application, such as... Figure 4 As shown, within the first time window, the interacting vehicles gradually approach the driver vehicle. The minimum lateral distance, minimum longitudinal distance, and corresponding lateral and longitudinal collision times within this window can be calculated. In the second time window, the interacting vehicles may enter a parallel or parallel state, and the lateral and longitudinal collision times change. In the third time window, the interacting vehicles gradually move away, and the relative speed and heading difference between the vehicles change accordingly. By calculating the above interaction indicators in different windows, the quantitative state of vehicle interaction at each time stage can be obtained. The interacting vehicles approach in the first time window, move parallel in the second time window, and move away in the third time window. This multi-time-window feature sequence accurately describes the evolution of the interaction between the driver vehicle and the interacting vehicles and identifies the corresponding scenarios.

[0137] The feature construction unit 308, also known as the multi-time-window feature construction unit, can be connected to the indicator calculation unit 306. It is used to combine interactive indicators in different time windows in chronological order to form a multi-time-window feature sequence that reflects the trend of interactive behavior over time. Table 1 is a feature sequence table according to an embodiment of this application. As shown in Table 1, taking three sets of time windows as an example, the following feature sequences can be obtained respectively.

[0138] Table 1 Feature Sequence List

[0139]

[0140] The behavior and scenario mining unit 310 can construct trend curves for interactive behaviors by comparing the changes in different time window sequences. For example, when the minimum lateral distance sequence is as shown, it can represent "approaching—parallel—moving away," which, combined with the vehicle's deceleration action, can be identified as a defensive deceleration scenario. When the lateral collision time decreases or increases window by window, it can represent a "lateral approaching or moving away trend," which, combined with the lane status, can be identified as a cutting-in or cutting-out scenario.

[0141] Optionally, when the longitudinal collision time is minimized in the second window and the relative velocity is negative, it can be identified as a "potential intrusion behavior".

[0142] Optionally, when the heading difference gradually increases, it can be identified as "other vehicle detouring or changing lanes".

[0143] Compared to traditional single-frame-based vehicle behavior mining methods, this application's embodiments calculate interaction metrics between vehicles within multiple future time windows and construct cross-window metric change sequences, thereby simultaneously reflecting the early, middle, and final stages of vehicle behavior. This allows for accurate differentiation of complex interaction behaviors such as approaching, paralleling, cutting in, and detouring, thus improving the accuracy and completeness of scene mining.

[0144] Furthermore, traditional methods rely solely on instantaneous information such as minimum distance or speed difference at a single moment, making them susceptible to misjudgments due to short-term noise. In contrast, this application's embodiments utilize trend features across multiple time windows for comprehensive judgment, significantly improving the stability and robustness of behavior recognition. Simultaneously, by introducing cross-window trend features, this application's embodiments can uncover phased behavior patterns that traditional methods cannot identify. For example, "approaching first then moving away" or "parallel first then cutting in," thereby expanding the types of scenarios that can be mined and improving the coverage of scene mining.

[0145] In summary, the embodiments of this application not only improve the recognition accuracy of vehicle interaction behavior, but also reduce false positives and false negatives, and make vehicle scene mining applicable to more types of driving scenarios, thereby enhancing the overall value of data mining results for the training and verification of autonomous driving models.

[0146] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0147] According to an embodiment of this application, a vehicle driving scenario determination system embodiment is provided. Figure 5 This is a schematic diagram of a vehicle driving scenario determination system according to an embodiment of this application, such as... Figure 5 As shown, the driving scene determination system 500 for this vehicle may include: a multi-time window construction unit 502, a future state acquisition unit 504, an interaction index calculation unit 506, a behavior recognition unit 508, and a scene mining unit 510.

[0148] The multi-time window construction unit 502 is used to obtain the target time period after the interaction conditions between the vehicle and at least one obstacle are met, and to divide the target time period into multiple time windows, wherein the target time period is used to represent the time range in which the interaction behavior occurs between the vehicle and the obstacle.

[0149] The future state acquisition unit 504 is used to acquire the state information of the vehicle and the obstacle in multiple time windows, wherein the state information is used to represent the motion state of the vehicle and the obstacle in the time window.

[0150] The interaction index calculation unit 506 is used to determine at least one interaction index between the vehicle and the obstacle within a time window based on state information. The interaction index is used to characterize the dynamic evolution relationship of the interaction behavior in the spatiotemporal dimension.

[0151] The behavior recognition unit 508 is used to determine the interaction type of the interaction behavior based on interaction indicators.

[0152] Scene mining unit 510 is used to determine the driving scene in which the vehicle is located based on the interaction type.

[0153] Furthermore, the vehicle driving scenario determination system also includes a multi-time window feature construction unit, which combines multiple interaction indicators in chronological order to obtain a feature sequence, wherein the feature sequence is used to reflect the trend of interaction behavior over time.

[0154] In this embodiment, the interaction index calculation unit can also be called the index calculation unit. The multi-time window feature construction unit can also be called the feature construction unit. The behavior recognition unit and scene mining unit can be called the behavior and scene mining unit.

[0155] In this embodiment, a multi-time window construction unit 502 acquires the target time period after the interaction conditions between the vehicle and at least one obstacle are met, and divides the target time period into multiple time windows, where the target time period represents the time range in which the interaction occurs between the vehicle and the obstacle; a future state acquisition unit 504 acquires the state information of the vehicle and the obstacle within the multiple time windows, where the state information represents the motion state of the vehicle and the obstacle within the time window; an interaction index calculation unit 506 determines at least one interaction index between the vehicle and the obstacle within the time window based on the state information, where the interaction index characterizes the dynamic evolution relationship of the interaction behavior in the spatiotemporal dimension; a behavior recognition unit 508 determines the interaction type of the interaction behavior based on the interaction index; and a scene mining unit 510 determines the driving scene in which the vehicle is located based on the interaction type. This solves the technical problem of low accuracy in determining the driving scene of the vehicle and achieves the technical effect of improving the accuracy of determining the driving scene of the vehicle.

[0156] According to an embodiment of this application, a vehicle driving scenario determination device is provided. It should be noted that the vehicle driving scenario determination device can be used to execute the above-described vehicle driving scenario determination method.

[0157] Figure 6 This is a schematic diagram of a vehicle driving scenario determination device according to an embodiment of this application, such as... Figure 6 As shown, the driving scenario determination device 600 for the vehicle may include: a first acquisition unit 602, a second acquisition unit 604, a first determination unit 606, a second determination unit 608, and a third determination unit 610.

[0158] The first acquisition unit 602 is used to acquire the target time period after the interaction conditions between the vehicle and at least one obstacle are met, and to divide the target time period into multiple time windows, wherein the target time period is used to represent the time range in which the interaction behavior occurs between the vehicle and the obstacle.

[0159] The second acquisition unit 604 is used to acquire the status information of the vehicle and the obstacle in multiple time windows, wherein the status information is used to represent the motion status of the vehicle and the obstacle in the time window.

[0160] The first determining unit 606 is used to determine at least one interaction index between the vehicle and the obstacle within a time window based on state information, wherein the interaction index is used to characterize the dynamic evolution relationship of the interaction behavior in the spatiotemporal dimension.

[0161] The second determining unit 608 is used to determine the interaction type to which the interaction behavior belongs based on the interaction indicators.

[0162] The third determining unit 610 is used to determine the driving scenario in which the vehicle is located based on the interaction type.

[0163] Optionally, the second determining unit 608 includes: a combination subunit, used to combine multiple interaction indicators in chronological order to obtain a feature sequence, wherein the feature sequence is used to reflect the trend of interaction behavior changing over time; a comparison subunit, used to compare the same type of feature sequences within multiple time windows to obtain comparison results; and a determining subunit, used to determine the interaction type based on the comparison results.

[0164] Optionally, the interaction metrics include a lateral distance metric, which represents the evolution of the distance between the vehicle and the obstacle in the lateral direction. The determining sub-unit includes: a first determining sub-unit, which, in response to the number of time windows being three, determines the feature sequences within the first time window, the feature sequences within the second time window, and the feature sequences within the third time window in the three time windows; and a second determining sub-unit, which, in response to the comparison result that the lateral distance metric within the first time window is greater than the lateral distance metric within the second time window, and the lateral distance metric within the second time window is less than the lateral distance metric within the third time window, determines the interaction type as a first interaction type, wherein the first interaction type indicates that the vehicle and the obstacle are close to each other in the first time window, run parallel in the second time window, and move away from each other in the third time window.

[0165] Optionally, the third determining unit 610 includes: a third determining subunit, used to determine the driving scenario as a defensive deceleration scenario in response to the interaction type being the first interaction type and the vehicle being in a deceleration state.

[0166] Optionally, the interaction metrics include a longitudinal collision time metric, a direction metric, and a speed metric. The longitudinal collision time metric represents the time required for the vehicle and the obstacle to first collide in the longitudinal direction. The speed metric represents the closing velocity or discrete velocity between the vehicle and the obstacle. The direction metric represents the degree of consistency or divergence in the vehicle's driving direction. The determining sub-unit includes: a fourth determining sub-unit, which determines the interaction type as a second interaction type in response to the longitudinal collision time metric in the second time window being less than the longitudinal collision time metric in the first time window, the longitudinal collision time metric in the second time window being less than the longitudinal collision time metric in the third time window, and the speed metric being negative. The second interaction type indicates that there is a potential cutting-in behavior between the vehicle and the obstacle. A fifth determining sub-unit, which determines the interaction type as a third interaction type in response to the direction metric gradually increasing in the order of the first time window, the second time window, and the third time window. The third interaction type indicates that there is a bypass behavior or lane-changing behavior between the vehicle and the obstacle.

[0167] Optionally, the interaction metrics include a lateral collision time metric, which represents the time required for the vehicle and the obstacle to first collide in the lateral direction. The second determining unit 608 includes: a sixth determining subunit, which determines the interaction type as a fourth interaction type in response to the lateral collision time metric showing a decreasing trend in a continuous time window, wherein the fourth interaction type represents the vehicle and the obstacle approaching each other in the lateral direction; and a seventh determining subunit, which determines the interaction type as a fifth interaction type in response to the lateral collision time metric showing an increasing trend in a continuous time window, wherein the fifth interaction type represents the vehicle and the obstacle moving away from each other in the lateral direction.

[0168] Optionally, the third determining unit 610 includes: an eighth determining subunit, used to determine the driving scenario as an entry scenario in response to the interaction type being the fourth interaction type; and a ninth determining subunit, used to determine the driving scenario as an exit scenario in response to the interaction type being the fifth interaction type.

[0169] Optionally, the interaction metrics also include a longitudinal distance metric, which is used to represent the evolution of the distance between the vehicle and the obstacle in the longitudinal direction.

[0170] In the vehicle driving scenario determination device of this embodiment, the first acquisition unit 602 acquires the target time period after the interaction conditions between the vehicle and at least one obstacle are met, and divides the target time period into multiple time windows, wherein the target time period is used to represent the time range in which the interaction behavior occurs between the vehicle and the obstacle; the second acquisition unit 604 acquires the state information of the vehicle and the obstacle within the multiple time windows, wherein the state information is used to represent the motion state of the vehicle and the obstacle within the time window; the first determination unit 606 determines at least one interaction index between the vehicle and the obstacle within the time window based on the state information, wherein the interaction index is used to characterize the dynamic evolution relationship of the interaction behavior in the spatiotemporal dimension; the second determination unit 608 determines the interaction type to which the interaction behavior belongs based on the interaction index; and the third determination unit 610 determines the driving scenario in which the vehicle is located based on the interaction type, thereby solving the technical problem of low accuracy in determining the vehicle driving scenario, and thus achieving the technical effect of improving the accuracy of determining the vehicle driving scenario.

[0171] Embodiments of this application also provide a vehicle, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods described in various embodiments of this application when it runs.

[0172] Embodiments of this application also provide a computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0173] Embodiments of this application also provide a computer program product, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0174] Embodiments of this application also provide a computer program product, including a non-volatile computer-readable storage medium for storing a computer program that, when executed by a processor, implements the methods in various embodiments of this application.

[0175] Embodiments of this application also provide a computer program that, when executed by a processor, implements the methods described in the various embodiments of this application.

[0176] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0177] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0178] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0179] 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 unit can be implemented in hardware or as a software functional unit.

[0180] 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 (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0181] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for determining a vehicle's driving scenario, characterized in that, include: The system obtains a target time period after the interaction conditions between the vehicle and at least one obstacle are met, and divides the target time period into multiple time windows, wherein the target time period is used to represent the time range in which the interaction behavior occurs between the vehicle and the obstacle. The state information of the vehicle and the obstacle within multiple time windows is obtained respectively, wherein the state information is used to represent the motion state of the vehicle and the obstacle within the time window; Based on the state information, at least one interaction index between the vehicle and the obstacle is determined within the time window, wherein the interaction index is used to characterize the dynamic evolution relationship of the interaction behavior in the spatiotemporal dimension. Based on the interaction metrics, determine the interaction type to which the interaction behavior belongs; Based on the interaction type, the driving scenario in which the vehicle is located is determined.

2. The method according to claim 1, characterized in that, Based on the interaction metrics, the interaction type to which the interaction behavior belongs is determined, including: Multiple interaction indicators are combined in chronological order to obtain a feature sequence, wherein the feature sequence is used to reflect the trend of the interaction behavior over time. The same type of feature sequences within multiple time windows are compared to obtain comparison results. Based on the comparison results, the interaction type is determined.

3. The method according to claim 2, characterized in that, The interaction metrics include a lateral distance metric, which represents the evolution of the distance between the vehicle and the obstacle in the lateral direction. Based on the comparison results, the interaction type is determined, including: In response to the fact that the number of time windows is three, the feature sequence in the first time window, the feature sequence in the second time window, and the feature sequence in the third time window are determined respectively. In response to the comparison result that the lateral distance index in the first time window is greater than the lateral distance index in the second time window, and the lateral distance index in the second time window is less than the lateral distance index in the third time window, the interaction type is determined to be a first interaction type, wherein the first interaction type is used to indicate that the vehicle and the obstacle are close to each other in the first time window, run parallel to each other in the second time window, and move away from each other in the third time window.

4. The method according to claim 3, characterized in that, Based on the interaction type, the driving scenario of the vehicle is determined, including: In response to the interaction type being the first interaction type and the vehicle being in a deceleration state, the driving scenario is determined to be a defensive deceleration scenario.

5. The method according to claim 3, characterized in that, The interaction metrics include a longitudinal collision time metric, a direction metric, and a speed metric. The longitudinal collision time metric represents the time required for the vehicle and the obstacle to first collide in the longitudinal direction. The speed metric represents the closing velocity or discrete velocity between the vehicle and the obstacle. The direction metric represents the degree of consistency or divergence in the vehicle's travel direction. Based on the comparison results, the interaction type is determined, including: In response to the longitudinal collision time index in the second time window being less than the longitudinal collision time index in the first time window, the longitudinal collision time index in the second time window being less than the longitudinal collision time index in the third time window, and the speed index being negative, the interaction type is determined to be the second interaction type, wherein the second interaction type is used to indicate that the vehicle has a potential cutting-in behavior with the obstacle. In response to the direction indicator gradually increasing in the order of the first time window, the second time window, and the third time window, the interaction type is determined to be the third interaction type, wherein the third interaction type is used to indicate that there is a detour or lane change behavior between the vehicle and the obstacle.

6. The method according to claim 1, characterized in that, The interaction metrics include a lateral collision time metric, which represents the time required for the vehicle and the obstacle to first collide in the lateral direction. Based on the interaction metrics, the interaction type of the interaction behavior is determined, including: In response to the lateral collision time index showing a decreasing trend in consecutive time windows, the interaction type is determined to be a fourth interaction type, wherein the fourth interaction type is used to indicate that the vehicle and the obstacle are approaching each other in the lateral direction; In response to the lateral collision time index showing an increasing trend in consecutive time windows, the interaction type is determined to be the fifth interaction type, wherein the fifth interaction type is used to indicate that the vehicle and the obstacle are moving away from each other in the lateral direction.

7. The method according to claim 6, characterized in that, Based on the interaction type, the driving scenario of the vehicle is determined, including: In response to the interaction type being the fourth interaction type, the driving scenario is determined to be the entry scenario; In response to the interaction type being the fifth interaction type, the driving scenario is determined to be a cut-out scenario.

8. The method according to any one of claims 1 to 7, characterized in that, The interaction metrics also include a longitudinal distance metric, which is used to represent the evolution of the distance between the vehicle and the obstacle in the longitudinal direction.

9. A vehicle driving scenario determination system, characterized in that, include: A multi-time window construction unit is used to obtain a target time period after the interaction conditions between the vehicle and at least one obstacle are met, and to divide the target time period into multiple time windows, wherein the target time period is used to represent the time range in which the interaction behavior occurs between the vehicle and the obstacle. The future state acquisition unit is used to acquire the state information of the vehicle and the obstacle within multiple time windows, wherein the state information is used to represent the motion state of the vehicle and the obstacle within the time window; An interaction index calculation unit is used to determine at least one interaction index between the vehicle and the obstacle within the time window based on the state information, wherein the interaction index is used to characterize the dynamic evolution relationship of the interaction behavior in the spatiotemporal dimension. A behavior recognition unit is used to determine the interaction type to which the interaction behavior belongs based on the interaction indicators; The scene mining unit is used to determine the driving scene in which the vehicle is located based on the interaction type.

10. The system according to claim 9, characterized in that, The system also includes: A multi-time-window feature construction unit is used to combine multiple interaction indicators in chronological order to obtain a feature sequence, wherein the feature sequence is used to reflect the trend of the interaction behavior over time.

Images