Trajectory evaluation method for vehicle, vehicle and readable storage medium

Abstract

This application provides a vehicle trajectory assessment method, a vehicle, and a readable storage medium. The method includes: determining a risk assessment area for the vehicle from the preset area based on the vehicle's driving status information within a preset area during the current time period, and the driving status information of other vehicles within the preset area; identifying at least one candidate risk vehicle from the other vehicles within the risk assessment area; predicting the driving status information of the candidate risk vehicle in a future time period based on the driving status information of the candidate risk vehicle; and assessing the risk of multiple candidate driving trajectories of the vehicle based on the vehicle's driving risk assessment function and the driving status information of the candidate risk vehicle in the future time period, thereby obtaining risk assessment results corresponding to each candidate driving trajectory. This application solves the technical problem in related technologies where it is impossible to accurately assess the driving risk of a vehicle.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and more specifically, to a vehicle trajectory evaluation method, a vehicle, and a readable storage medium. Background Technology

[0002] With the rapid development of intelligent connected and autonomous driving technologies, the safety of vehicle environmental perception and decision-making control systems has become a core bottleneck restricting the large-scale deployment of Level 3 and above autonomous driving systems.

[0003] In related technologies, vehicle driving risk assessment methods largely rely on sensor fusion technology. This involves acquiring real-time position and speed information of the vehicle and surrounding vehicles to construct a static risk model to determine potential collision or deviation risks. However, these methods generally assess risks based solely on the current observation state, failing to model the vehicle's motion evolution trend in the future prediction time domain. Therefore, the assessment results typically lag behind the actual risk development, making it difficult to accurately reflect the vehicle's true safety status in complex, highly dynamic traffic environments. This results in low accuracy in assessing vehicle driving risks, making it difficult to support real-time, reliable decision-making and control requirements.

[0004] There is currently no good solution to the aforementioned technical problem of being unable to accurately assess the driving risks of vehicles. Summary of the Invention

[0005] This application provides a method for evaluating vehicle trajectory, a vehicle, and a readable storage medium to at least solve the technical problem in related technologies that cannot accurately assess the driving risks of vehicles.

[0006] According to one aspect of the embodiments of this application, a vehicle trajectory assessment method is provided. The method includes: determining a risk assessment area for the vehicle from the preset area based on the vehicle's driving status information in a preset area during the current time period and the driving status information of other vehicles in the preset area; determining at least one candidate risk vehicle from other vehicles in the risk assessment area, wherein the candidate risk vehicle is a vehicle that has a collision risk with the vehicle; predicting the driving status information of the candidate risk vehicle in a future time period based on the driving status information of the candidate risk vehicle; and performing risk assessments on multiple candidate driving trajectories of the vehicle based on the vehicle's driving risk assessment function and the driving status information of the candidate risk vehicle in the future time period, thereby obtaining risk assessment results corresponding to the multiple candidate driving trajectories, wherein the risk assessment results are used to characterize the degree of risk when the vehicle drives according to the candidate driving trajectory.

[0007] Optionally, based on the driving status information of the candidate risk vehicle, the driving status information of the candidate risk vehicle in a future time period is predicted, including: inputting the driving status information of the candidate risk vehicle into a speed prediction algorithm for prediction to obtain the driving status information of the candidate risk vehicle in a future time period, wherein the speed prediction algorithm is a model pre-trained based on driving status information samples from different time periods.

[0008] Optionally, based on the vehicle's driving risk assessment function and the driving status information of candidate risk vehicles in the future time period, risk assessments are performed on multiple candidate driving trajectories of the vehicle to obtain risk assessment results corresponding to each candidate driving trajectory. This includes: traversing multiple candidate driving trajectories; and based on the vehicle's driving risk assessment function and the driving status information of candidate risk vehicles in the future time period, performing risk assessments on the currently traversed candidate driving trajectories to obtain risk assessment results for the candidate driving trajectories.

[0009] Optionally, the candidate driving trajectory includes first motion state parameters of the vehicle at multiple discrete time points within a future time period. Based on the vehicle's driving risk assessment function and the driving state information of the candidate risk vehicle within the future time period, a risk assessment is performed on the currently traversed candidate driving trajectory to obtain the risk assessment result of the candidate driving trajectory. This includes: determining second motion state parameters of the candidate risk vehicle at multiple discrete time points within the future time period based on the driving state information of the candidate risk vehicle within the future time period; and inputting the first motion state parameters at multiple discrete time points and the second motion state parameters at multiple discrete time points into the driving risk assessment function for risk assessment to obtain the risk assessment result of the candidate driving trajectory.

[0010] Optionally, the method further includes: constructing a driving risk assessment function based on a lane departure risk assessment function, a collision risk assessment function, and a potential field assessment function. The lane departure risk assessment function is used to assess the degree of deviation between the driving lanes of the vehicles when the vehicles are traveling along the candidate driving trajectory at a target discrete time point. The collision risk assessment function is used to assess the collision risk between the vehicle and at least one candidate risk vehicle when the vehicles are traveling along the candidate driving trajectory at a target discrete time point. The potential field assessment function is used to characterize the degree of influence of the candidate risk vehicle on the vehicles traveling along the candidate driving trajectory at a target discrete time point. The target discrete time point is any one of multiple discrete time points.

[0011] Optionally, a driving risk assessment function is constructed based on the lane departure risk assessment function, the collision risk assessment function, and the potential field assessment function. This includes: determining a first weighting coefficient corresponding to the lane departure risk assessment function, a second weighting coefficient corresponding to the collision risk assessment function, and a third weighting coefficient corresponding to the potential field assessment function, wherein the first weighting coefficient is less than the second weighting coefficient, and the first weighting coefficient is equal to the third weighting coefficient; weighting and fusing the lane departure risk assessment function with the first weighting coefficient, the collision risk assessment function with the second weighting coefficient, and the potential field assessment function with the third weighting coefficient to obtain an initial driving risk assessment function; and summing the initial driving risk assessment function at multiple discrete time points to obtain the final driving risk assessment function.

[0012] Optionally, based on the vehicle's driving status information in the preset area during the current time period, and the driving status information of other vehicles in the preset area, the risk assessment area of ​​the vehicle is determined from the preset area, including: predicting the collision risk level between the vehicle and other vehicles based on the vehicle's driving status information in the preset area during the current time period, and the driving status information of other vehicles in the preset area; and determining the area where other vehicles with a collision risk level higher than the preset collision risk level are located within the preset area as the vehicle's risk assessment area.

[0013] Optionally, after obtaining the risk assessment results corresponding to multiple candidate driving trajectories, the method further includes: based on the risk assessment results, selecting candidate driving trajectories with a risk level less than a preset threshold from the multiple candidate driving trajectories.

[0014] According to another aspect of the embodiments of this application, a vehicle trajectory evaluation device is also provided. The device includes: a first determining unit, configured to determine a risk assessment area of ​​the vehicle from the preset area based on the vehicle's driving status information in the current time period and the driving status information of other vehicles in the preset area; a second determining unit, configured to determine at least one candidate risk vehicle from other vehicles in the risk assessment area, wherein the candidate risk vehicle is a vehicle that has a collision risk with the vehicle; a prediction unit, configured to predict the driving status information of the candidate risk vehicle in a future time period based on the driving status information of the candidate risk vehicle; and an evaluation unit, configured to perform risk evaluation on multiple candidate driving trajectories of the vehicle based on the vehicle's driving risk evaluation function and the driving status information of the candidate risk vehicle in the future time period, and obtain risk evaluation results corresponding to the multiple candidate driving trajectories respectively, wherein the risk evaluation results are used to characterize the degree of risk when the vehicle drives according to the candidate driving trajectory.

[0015] 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.

[0016] 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.

[0017] 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.

[0018] 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 that, when executed by a processor, implements the methods in various embodiments of this application.

[0019] 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.

[0020] In this embodiment, the vehicle's risk assessment area is defined by analyzing the vehicle's driving status information relative to other vehicles in the vicinity. This allows the focus to be placed solely on candidate risk vehicles within the risk assessment area that pose a potential collision risk, avoiding ineffective calculations for irrelevant targets. Subsequently, the driving status information of these candidate risk vehicles is predicted for future timeframes. This breaks away from the traditional static assessment model that relies solely on current-moment information, enabling forward-looking predictions of the driving status of candidate risk vehicles over future time periods. Finally, using the vehicle's driving risk assessment function and the driving status information of candidate risk vehicles over future time periods, risk assessments are performed on multiple candidate driving trajectories for the vehicle. This accurately determines the degree of risk associated with the vehicle driving along these multiple candidate trajectories. In other words, the embodiments of this application consider the driving status information of candidate risk vehicles in the future time period and assess the degree of risk when the vehicle drives along multiple candidate driving trajectories through the vehicle's driving risk assessment function. This overcomes the lag defect of traditional static assessment and constructs a precise screening mechanism based on dynamic risk assessment area. It significantly improves the accuracy of risk assessment of vehicle driving trajectory in complex traffic scenarios, provides efficient and reliable safety decision support for the vehicle's autonomous driving system, and solves the technical problem of not being able to accurately assess the driving risk of vehicles in related technologies. Attached Figure Description

[0021] 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:

[0022] Figure 1 This is a flowchart of a vehicle trajectory evaluation method according to an embodiment of this application;

[0023] Figure 2 This is a flowchart of a vehicle trajectory risk assessment method according to an embodiment of this application;

[0024] Figure 3 This is a flowchart of a method for constructing a driving risk assessment function according to an embodiment of this application;

[0025] Figure 4 This is a flowchart of a risk assessment method for candidate driving trajectories of a vehicle according to an embodiment of this application;

[0026] Figure 5 This is a schematic diagram of a vehicle trajectory evaluation device according to an embodiment of this application;

[0027] Figure 6 This is a schematic diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0028] 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.

[0029] 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.

[0030] According to an embodiment of this application, an embodiment of a vehicle trajectory evaluation method 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.

[0031] This embodiment provides a vehicle trajectory evaluation method applicable to low-to-medium speed driving scenarios. Figure 1 This is a flowchart of a vehicle trajectory evaluation method according to an embodiment of this application, such as... Figure 1 As shown, the process includes the following steps.

[0032] Step S101: Based on the vehicle's driving status information in the preset area during the current time period, and the driving status information of other vehicles in the preset area, determine the vehicle's risk assessment area from the preset area.

[0033] In the technical solution provided in step S101 of this application, the vehicle's driving status information at the current moment can include six-dimensional dynamic parameters of the vehicle in the Frenet coordinate system. For example, the vehicle's longitudinal position, lateral offset, longitudinal velocity, lateral velocity, longitudinal acceleration, and lateral acceleration. These parameters are collected collaboratively by onboard sensors, such as LiDAR, cameras, Global Positioning System (GPS) sensors, and speed sensors, and obtained after coordinate system transformation, characterizing the vehicle's real-time motion state relative to the reference lane. Simultaneously, the driving status information of other vehicles within a preset area is acquired. This driving status information of other vehicles can include their longitudinal position, lateral offset, longitudinal velocity, lateral velocity, longitudinal acceleration, and lateral acceleration in the Frenet coordinate system, used to characterize the relative motion of other vehicles relative to the vehicle itself.

[0034] In this embodiment, after obtaining the driving status information of the vehicle in the preset area and the driving status information of other vehicles in the preset area, the risk assessment area of ​​the vehicle can be determined from the preset area based on the driving status information of the vehicle in the preset area and the driving status information of other vehicles in the preset area.

[0035] For example, using the vehicle as a baseline, the forward longitudinal boundary can be calculated by combining the vehicle's longitudinal speed, maximum deceleration, system reaction time, and minimum safe following distance within the current time period. A rear fixed boundary can then be set based on this forward longitudinal boundary. Furthermore, a lateral boundary can be set, which can be obtained by extending the current lane width by a preset distance. Subsequently, the dynamic rectangular perception domain formed by the forward longitudinal boundary, the rear fixed boundary, and the lateral boundary is determined as the candidate risk area for the vehicle.

[0036] In this step, since the original driving state information of the vehicle and other vehicles comes from sensors such as LiDAR and GPS, it is expressed in a global Cartesian coordinate system by default. This system has global absoluteness but lacks semantic relevance to the lane structure. To achieve accurate modeling of the vehicle's motion state and the road geometry, based on the lane centerline information, the driving state information of the vehicle and other vehicles can be uniformly transformed to the Frenet coordinate system using coordinate system transformation relationships. In the Frenet coordinate system, the motion state of each vehicle is represented by its longitudinal coordinate (arc length along the lane centerline), lateral coordinate (offset distance perpendicular to the lane centerline), and corresponding motion state parameters such as longitudinal velocity, lateral velocity, longitudinal acceleration, and lateral acceleration. This coordinate system fits the road structure and can intuitively reflect the relative position and motion trend of the vehicle and other vehicles relative to the reference lane. Based on this, by combining the driving state information of the vehicle with that of other surrounding vehicles, the risk assessment area of ​​the vehicle can be determined. This allows subsequent calculations to focus only on candidate risk vehicles with potential collision risks within the risk assessment area, avoiding invalid calculations for irrelevant targets.

[0037] Step S102: Identify at least one candidate risk vehicle from other vehicles within the risk assessment area.

[0038] In the technical solution provided in step S102 of this application, after determining the risk assessment area of ​​the vehicle, candidate risk vehicles that pose a real collision threat to the vehicle can be further screened from other vehicles within the risk assessment area. These candidate risk vehicles are those that pose a collision risk to the vehicle. This screening process is not simply based on spatial proximity, but rather combines the dynamic interaction characteristics between vehicles for multi-dimensional logical judgment to exclude low-risk vehicles or vehicles with no intention to interact.

[0039] In this embodiment, when identifying at least one candidate risk vehicle from other vehicles within the risk assessment area, it can first be determined whether the longitudinal relative speed of other vehicles within the risk assessment area exceeds a first preset speed threshold to identify whether there is a significant speed difference between other vehicles and the vehicle itself that could lead to a rear-end collision or sudden braking conflict. Simultaneously, it is checked whether the lateral relative speed of other vehicles is greater than a second preset speed threshold to identify whether other vehicles intend to change lanes or cut laterally. Only when other vehicles within the risk assessment area simultaneously meet the following conditions—being longitudinally located within the risk assessment area, laterally not exceeding the adjacent lane extension boundary, and having a significant relative speed difference—are they identified as candidate risk vehicles posing a potential collision risk to the vehicle.

[0040] In this step, identifying at least one candidate risk vehicle from other vehicles within the risk assessment area ensures that subsequent risk assessments focus only on vehicles with potential collision risks, thereby significantly improving the relevance and computational efficiency of the assessment.

[0041] Step S103: Based on the driving status information of the candidate risk vehicles, predict the driving status information of the candidate risk vehicles in the future time period.

[0042] In the technical solution provided in step S103 of this application, after identifying candidate risk vehicles that pose a potential collision risk to the vehicle, the driving status information of the candidate risk vehicles in the future time period can be predicted based on their driving status information to support high-precision dynamic risk assessment. The driving status information of the candidate risk vehicles in the future time period may include parameters such as the longitudinal position, lateral offset, longitudinal velocity, lateral velocity, longitudinal acceleration, and lateral acceleration of the candidate risk vehicles at various discrete time points in the future time period, all within the Frenet coordinate system. These parameters collectively constitute the trajectory trend of the candidate risk vehicles in the future time period.

[0043] In this embodiment, the driving status information of the candidate risk vehicle (e.g., the six-dimensional state parameters of the candidate risk vehicle in the Frenet coordinate system) can be input into the speed prediction algorithm model to predict the driving status information of the candidate risk vehicle in the future time period. The speed prediction algorithm model can be one of the hidden Markov algorithm model, long short-term memory network and Kalman filter algorithm model, and no specific limitation is made here.

[0044] In this step, based on the driving status information of candidate risk vehicles, the driving status information of candidate risk vehicles in the future time period is predicted. This can realize the prediction of the motion status of candidate risk vehicles, which breaks through the limitations of traditional static evaluation that only relies on the current state and provides a reliable reference for subsequent vehicle driving risk assessment.

[0045] Step S104: Based on the vehicle's driving risk assessment function and the driving status information of candidate risk vehicles in the future time period, risk assessment is performed on multiple candidate driving trajectories of the vehicle to obtain the risk assessment results corresponding to each candidate driving trajectory.

[0046] In the technical solution provided in step S104 of this application, the driving risk assessment function consists of three types of sub-functions: lane departure risk assessment function, collision risk assessment function, and potential field assessment function. The lane departure risk assessment function is used to quantify the risk of lateral collision or lane crossing caused by lane departure by calculating the lateral offset of the vehicle's center of gravity from the current lane centerline and its velocity change. The collision risk assessment function is used to assess the probability of a longitudinal collision between the vehicle and a candidate risk vehicle while traveling along a candidate driving trajectory, based on the relative distance, relative velocity angle, prediction time domain, and safe distance model between the vehicle and the candidate risk vehicle. The potential field assessment function can model each candidate risk vehicle as a potential field source with a distance attenuation effect. The potential field strength dynamically changes with the relative position between the candidate risk vehicles, reflecting the implicit interference and avoidance pressure between the vehicle and the candidate risk vehicle. The three types of sub-functions mentioned above are weighted and fused to form a driving risk assessment function. Among them, the collision risk assessment function has the largest weight coefficient, while the lane departure risk assessment function and the potential field assessment function have equal weight coefficients to ensure that the vehicle assessment focuses on collision risk.

[0047] Optionally, multiple candidate driving trajectories for the vehicle in future time periods are generated in advance through a trajectory prediction model. Each trajectory consists of a set of six-dimensional state parameters at discrete time points. These six-dimensional state parameters represent the vehicle's longitudinal position, lateral position, longitudinal velocity, lateral velocity, longitudinal acceleration, and lateral acceleration in the Frenet coordinate system.

[0048] In this embodiment, risk assessments can be performed on multiple candidate driving trajectories of the vehicle based on the vehicle's driving risk assessment function and the driving status information of candidate risk vehicles in the future time period, so as to obtain risk assessment results corresponding to each candidate driving trajectory. The risk assessment results are used to characterize the degree of risk when the vehicle drives according to the candidate driving trajectory.

[0049] For example, as described above, the driving status information of a candidate risk vehicle in the future time period can include parameters such as the longitudinal position, lateral offset, longitudinal velocity, lateral velocity, longitudinal acceleration, and lateral acceleration of the candidate risk vehicle in the Frenet coordinate system at each discrete time point in the future time period. Similarly, each of the vehicle's multiple candidate driving trajectories also includes parameters such as the longitudinal position, lateral offset, longitudinal velocity, lateral velocity, longitudinal acceleration, and lateral acceleration of the vehicle in the Frenet coordinate system in the future time period. Based on this, for each candidate driving trajectory, according to the parameter requirements of the vehicle's driving risk assessment function, the state parameters of the candidate risk vehicle at each discrete time point in the future time period, as well as the state parameters of the candidate driving trajectory at each discrete time point in the future time period, can be input into the vehicle's driving risk assessment function for calculation, resulting in a risk assessment result for each candidate driving trajectory. This risk assessment result can be represented by a risk value; that is, the larger the risk value corresponding to the candidate driving trajectory, the greater the degree of risk when the vehicle drives along that candidate driving trajectory.

[0050] Optionally, after obtaining the risk assessment results corresponding to the multiple candidate driving trajectories of the vehicle, the degree of risk of the vehicle driving according to the multiple candidate driving trajectories can be determined based on the risk assessment results corresponding to the multiple candidate driving trajectories. Then, the candidate driving trajectory with the lowest risk level can be selected from the multiple candidate driving trajectories as the basis for the vehicle's driving decision in the future time period.

[0051] In this step, the risk assessment of multiple candidate driving trajectories of the vehicle is carried out by using the vehicle's driving risk assessment function and the driving status information of candidate risk vehicles in the future time period. This can accurately determine the degree of risk when the vehicle drives according to these multiple candidate driving trajectories in the future time period, thereby providing reliable data basis for the vehicle's driving decision in the future time period.

[0052] In steps S101 to S104 of this application, the risk assessment area of ​​the vehicle is delineated based on the driving status information of the vehicle and other surrounding vehicles. This allows focusing only on candidate risk vehicles within the risk assessment area that pose a potential collision risk, avoiding invalid calculations for irrelevant targets. Next, the driving status information of candidate risk vehicles in the future is predicted, breaking through the traditional static assessment model that relies solely on current-time information. This achieves forward-looking prediction of the driving status of candidate risk vehicles in the future time period. Finally, using the vehicle's driving risk assessment function and the driving status information of candidate risk vehicles in the future time period, risk assessments are performed on multiple candidate driving trajectories of the vehicle, accurately determining the degree of risk when the vehicle travels along multiple candidate driving trajectories. In other words, this application considers the driving status information of candidate risk vehicles in the future time period and evaluates the degree of risk when a vehicle travels along multiple candidate driving trajectories through the vehicle's driving risk assessment function. This overcomes the lag defect of traditional static assessment and constructs a precise screening mechanism based on dynamic risk assessment areas. It significantly improves the accuracy of risk assessment of vehicle driving trajectories in complex traffic scenarios, provides efficient and reliable safety decision support for autonomous driving systems, and solves the technical problem of not being able to accurately assess the driving risks of vehicles in related technologies.

[0053] The trajectory evaluation method for vehicles in this application will be further described below.

[0054] As an optional implementation, step S103, based on the driving status information of the candidate risk vehicle, predicts the driving status information of the candidate risk vehicle in the future time period, including: inputting the driving status information of the candidate risk vehicle into a speed prediction algorithm for prediction, to obtain the driving status information of the candidate risk vehicle in the future time period, wherein the speed prediction algorithm is a model pre-trained based on driving status information samples from different time periods.

[0055] In this embodiment, to accurately predict the motion behavior of candidate risk vehicles in the future time period, a pre-trained speed prediction algorithm is used to predict the driving status information of candidate risk vehicles in the future time period. This speed prediction algorithm is not a simple linear extrapolation, but a time-series modeling model trained with a large amount of vehicle motion data in real traffic scenarios. For example, the speed prediction algorithm can be a Hidden Markov Model, a Long Short-Term Memory Network, and a Kalman Filter algorithm, etc., without specific limitations here.

[0056] Optionally, the driving status information of candidate risk vehicles within the current time period can be input into the speed prediction algorithm for prediction, to obtain the driving status information of candidate risk vehicles in future time periods. The driving status information of candidate risk vehicles within the current time period can include parameters such as the longitudinal position, lateral offset, longitudinal velocity, lateral velocity, longitudinal acceleration, and lateral acceleration of the candidate risk vehicle in the Frenet coordinate system. The driving status information of candidate risk vehicles in future time periods can include parameters such as the longitudinal position, lateral offset, longitudinal velocity, lateral velocity, longitudinal acceleration, and lateral acceleration of the candidate risk vehicle at each discrete time point within the future time period.

[0057] In this step, the driving status information of candidate risk vehicles is input into a trained speed prediction algorithm for prediction. This can accurately determine the driving status information of candidate risk vehicles in the future time period, thereby providing reliable data for risk assessment of subsequent vehicle candidate driving trajectories.

[0058] As an optional implementation, step S104 involves assessing the risk of multiple candidate driving trajectories of the vehicle based on the vehicle's driving risk assessment function and the driving status information of candidate risk vehicles in the future time period, and obtaining the risk assessment results corresponding to each candidate driving trajectory. This includes: traversing multiple candidate driving trajectories; and assessing the risk of the currently traversed candidate driving trajectory based on the vehicle's driving risk assessment function and the driving status information of candidate risk vehicles in the future time period, thereby obtaining the risk assessment results of the candidate driving trajectory.

[0059] In this embodiment, when risk assessments are performed on multiple candidate driving trajectories of a vehicle based on the vehicle's driving risk assessment function and the driving status information of candidate risk vehicles in the future time period, and the risk assessment results corresponding to the multiple candidate driving trajectories are obtained, multiple candidate driving trajectories can be traversed, and then each candidate driving trajectory currently traversed can be risk assessed according to the vehicle's driving risk assessment function and the driving status information of candidate risk vehicles in the future time period, so as to obtain the risk assessment result of the candidate driving trajectory.

[0060] Optionally, if the risk assessment result indicates that the risk level of the currently traversed candidate driving trajectory is lower than a preset threshold, it means that the risk of the vehicle driving along the candidate driving trajectory is within an acceptable range, and the candidate driving trajectory can be retained. Conversely, if the risk assessment result indicates that the risk level of the currently traversed candidate driving trajectory is greater than or equal to the preset threshold, it means that the vehicle driving along the candidate driving trajectory poses a high risk. In this case, the candidate driving trajectory can be deleted from the multiple driving trajectories.

[0061] Optionally, during the traversal of multiple candidate driving trajectories, if the currently traversed candidate driving trajectory is not the last one, the traversal continues, and the risk assessment process described above is repeated. If the currently traversed candidate driving trajectory is the last candidate trajectory, then the risk assessment process for the entire candidate driving trajectory ends, and the remaining candidate driving trajectory will serve as a reference for subsequent vehicle decisions and control.

[0062] In this step, by going through multiple candidate driving trajectories for the vehicle in the future time period one by one, the risk assessment result of each candidate driving trajectory can be determined. Then, candidate driving trajectories with acceptable risk can be selected from multiple candidate driving trajectories, providing a reliable reference for subsequent vehicle decision-making and control.

[0063] The following section further describes the process of using a vehicle-based driving risk assessment function and the driving status information of candidate risk vehicles in the future time period to assess the risk of the currently traversed candidate driving trajectories and obtain the risk assessment results of the candidate driving trajectories.

[0064] As an optional implementation, the candidate driving trajectory includes first motion state parameters of the vehicle at multiple discrete time points within a future time period. Based on the vehicle's driving risk assessment function and the driving state information of the candidate risk vehicle within the future time period, a risk assessment is performed on the currently traversed candidate driving trajectory to obtain the risk assessment result of the candidate driving trajectory. This includes: determining second motion state parameters of the candidate risk vehicle at multiple discrete time points within the future time period based on the driving state information of the candidate risk vehicle within the future time period; and inputting the first motion state parameters at multiple discrete time points and the second motion state parameters at multiple discrete time points into the driving risk assessment function for risk assessment to obtain the risk assessment result of the candidate driving trajectory.

[0065] In this embodiment, as described above, each candidate driving trajectory includes motion state parameters at multiple discrete time points within a future time period of the vehicle. For ease of explanation, these motion state parameters can be referred to as the first motion state parameters. The first motion state parameters can represent the six-dimensional dynamic state of the vehicle in the Frenet coordinate system, including longitudinal position, lateral offset, longitudinal velocity, lateral velocity, longitudinal acceleration, and lateral acceleration, reflecting the vehicle's motion posture at each discrete time point within a future time period.

[0066] Optionally, as described above, the driving status information of the candidate risk vehicle in the future time period includes the motion state parameters of the candidate risk vehicle at each discrete time point in the future time period. For ease of explanation, these motion state parameters can be referred to as the second motion state parameters. The second motion state parameters can represent the six-dimensional dynamic state of the candidate risk vehicle in the Frenet coordinate system, including longitudinal position, lateral offset, longitudinal velocity, lateral velocity, longitudinal acceleration, and lateral acceleration, reflecting the motion posture of the candidate risk vehicle at each discrete time point in the future time period.

[0067] Optionally, for the currently traversed candidate driving trajectory, after obtaining the first motion state parameters and the second motion state parameters at each discrete time point in the future time period, the first motion state parameters of the candidate driving trajectory at each discrete time point in the future time period and the second motion state parameters of the candidate risk vehicle at each discrete time point in the future time period can be input into the vehicle's driving risk assessment function for risk assessment according to the parameter requirements in the vehicle's driving risk assessment function, so as to obtain the risk assessment result of the currently traversed candidate driving trajectory.

[0068] Optionally, when conducting risk assessment, the vehicle's driving risk assessment function can calculate three types of risk components at each discrete time point based on the lane departure risk assessment function, collision risk assessment function, and potential field assessment function. These three risk components are then weighted and fused to form the risk value for that discrete time point. Subsequently, the risk values ​​for each discrete time point over a future time period are integrated or the maximum value is taken to generate the final risk assessment result for that candidate trajectory. This process achieves refined modeling of the spatiotemporal coupling between "trajectory point and target state," ensuring that the risk assessment result accurately reflects the overall safety threat faced by the vehicle when interacting with the dynamic environment while driving along the candidate trajectory. This provides a scientific, quantitative, and traceable decision-making basis for subsequent trajectory selection.

[0069] The following section will further explain the vehicle's driving risk function.

[0070] As an optional implementation, the vehicle trajectory evaluation method further includes: constructing a driving risk evaluation function based on a lane departure risk evaluation function, a collision risk evaluation function, and a potential field evaluation function. The lane departure risk evaluation function is used to evaluate the degree of deviation between the vehicle and its driving lane at a target discrete time point when the vehicle is traveling along a candidate driving trajectory. The collision risk evaluation function is used to evaluate the collision risk between the vehicle and at least one candidate risk vehicle at a target discrete time point when the vehicle is traveling along a candidate driving trajectory. The potential field evaluation function is used to characterize the influence of the candidate risk vehicle on the vehicle traveling along the candidate driving trajectory at a target discrete time point. The target discrete time point is any one of multiple discrete time points.

[0071] In this embodiment, as described above, the vehicle's driving risk assessment function is constructed from a lane departure risk assessment function, a collision risk assessment function, and a potential field assessment function. The lane departure risk assessment function measures the potential danger of the vehicle deviating from the centerline of its current lane at a target discrete time point. This function is based on the vehicle's lateral offset and lateral velocity in the Frenet coordinate system, comprehensively considering the physical distance between the vehicle and the lane line and the lateral movement trend. It uses an exponential decay model, where the greater the deviation and the higher the lateral velocity, the risk value of the lane departure increases non-linearly. This mechanism not only reflects static position deviation but also embodies dynamic loss-of-control tendencies, avoiding excessive warnings caused by minor drifts, while providing a sensitive response to rapid lane changes or loss-of-control behavior, thus accurately characterizing the risk of lateral collisions or lane violations caused by lane departure.

[0072] For example, the lane departure risk assessment function can be represented by the following formula (1).

[0073] (1)

[0074] in, It can be used to represent a lane departure risk assessment function. It can be used to represent the centroid coordinates of a vehicle at time k. It can be used to represent the lane center point corresponding to the shortest distance between the centroid coordinate position of the vehicle at time k and the lane centerline of the vehicle's current driving lane. Can be used to represent The exponential decay function of lane departure at any given time. Can be used to represent The speed deviation term between the vehicle at a given time and the candidate risk vehicle. , These are the weighting coefficients. It can be used to represent the number of candidate risk vehicles.

[0075] Optionally, the above The calculation expression can be represented by the following formula (2).

[0076] (2)

[0077] in, It can be used to represent the attenuation coefficient. It can be used to represent exponential functions. Can be used to represent The distance from the vehicle's center of gravity to the nearest lane centerline at any given time (e.g., the vehicle's lateral offset).

[0078] Optionally, the above The calculation expression can be represented by the following formula (3).

[0079] (3)

[0080] in, It can be used to represent vehicles and the first The relative distance between each candidate risk vehicle It can be used to indicate the speed of a vehicle. It can be used to represent vehicles and the first The relative speeds between the candidate risk vehicles.

[0081] Optionally, the collision risk assessment function focuses on the longitudinal interaction threat between the vehicle and each candidate risk vehicle at a target discrete time point. This function uses the relative distance between the two vehicles, the relative speed angle, and the predicted safe distance of the vehicle as its core inputs. The safe distance is dynamically calculated from the vehicle's speed, maximum deceleration, and system reaction time. When the actual relative distance between the two vehicles approaches or falls below the safe distance, the collision risk increases exponentially, and the risk intensity significantly increases as the speed angle decreases (i.e., head-on or forward approach). This collision risk assessment function directly quantifies the probability of a collision through a physical model, embedding vehicle dynamics constraints and collision avoidance logic into the assessment process, ensuring high sensitivity and strong interpretability for fatal scenarios such as rear-end collisions and cut-in collisions.

[0082] For example, the collision risk assessment function can be represented by the following formula (4).

[0083] (4)

[0084] in, It can be used to represent a collision risk assessment function. It can be used to represent the first Candidate driving trajectories It can be used to indicate the vehicle number The coefficient corresponding to the safe distance between each candidate risk vehicle. It can be used to represent vehicles and the first Safe distance between candidate risk vehicles It can be used to represent vehicles and the first The speed angle between the candidate risk vehicles It can be used to represent vehicles and the first The relative distance between the candidate risk vehicles.

[0085] Optionally, the vehicles in the above formula and Safety distance between candidate risk vehicles It can be expressed by the following formula (5).

[0086] (5)

[0087] in, It can be used to indicate the vehicle in the first... The first candidate trajectory The speed of time It can be used to represent the first The candidate risk vehicles in the Predicting speed at any given moment It can be used to represent the maximum deceleration of a vehicle. It can be used to represent the reaction time of the decision-making system of autonomous vehicles. It can be used to represent vehicles and the first The minimum distance between candidate risk vehicles when they are stationary.

[0088] Optionally, a potential field evaluation function is used to simulate the implicit interference of candidate risk vehicles on vehicle driving behavior. This function constructs a potential energy field that decays with distance based on the relative position (longitudinal and lateral) of each candidate risk vehicle and the vehicle itself. When the distance between the two vehicles is less than a certain threshold, the potential field strength rises rapidly, forming a repulsive force, simulating the driver's instinct to avoid adjacent vehicles. This potential field value changes dynamically with relative position and is not limited by speed direction. It can effectively capture non-collision-related but significantly impactful interactive behaviors such as lane-changing intentions, following pressure, and lateral approach, thus compensating for the blind spots of traditional assessments that only focus on collisions.

[0089] Alternatively, the potential field evaluation function can be expressed by the following formula (6).

[0090] (6)

[0091] in, It can be used to represent the first One candidate risk vehicle The adaptive potential field value at time t. It can be used to represent the maximum potential field value. It can be used to represent the natural exponential function. , They are used to indicate the vehicle in At any given moment, the vertical and horizontal positions in the Frenet coordinate system. , They are used to represent the first One candidate risk vehicle At any given moment, in the Frenet coordinate system, the longitudinal and lateral positions of the vehicle and the candidate risk vehicle will change as the vehicle travels along the candidate trajectory. The size of the adaptive potential field of the candidate risk vehicle will adaptively adjust as the relative position between the vehicle and the candidate risk vehicle changes.

[0092] The following section will further explain the process of constructing a vehicle driving risk assessment function based on the three types of risk assessment functions mentioned above.

[0093] As an optional implementation, a driving risk assessment function is constructed based on the lane departure risk assessment function, the collision risk assessment function, and the potential field assessment function. This includes: determining a first weighting coefficient corresponding to the lane departure risk assessment function, a second weighting coefficient corresponding to the collision risk assessment function, and a third weighting coefficient corresponding to the potential field assessment function, wherein the first weighting coefficient is less than the second weighting coefficient, and the first weighting coefficient is equal to the third weighting coefficient; weighting and fusing the lane departure risk assessment function with the first weighting coefficient, the collision risk assessment function with the second weighting coefficient, and the potential field assessment function with the third weighting coefficient to obtain an initial driving risk assessment function; and summing the initial driving risk assessment function at multiple discrete time points to obtain the final driving risk assessment function.

[0094] In this embodiment, during the construction of the driving risk assessment function, it is necessary to first clarify the relative importance of the three types of risk assessment functions, that is, to determine the weight coefficients corresponding to the three types of risk assessment functions respectively. Among them, collision risk, as a core factor that directly threatens life and vehicle safety, can be assigned the highest weight coefficient, denoted as the second weight coefficient. Lane departure risk and the influence of the adaptive potential field, while not necessarily leading to a collision, do affect the stability of driving behavior and the conservatism of decision-making. Both are of equal importance in the risk level and are therefore assigned the first weight coefficient. With the third weighting coefficient Among them, the aforementioned first weighting coefficient Second weighting coefficient Third weighting coefficient The size relationship between them can be expressed by the following formula (7).

[0095] (7)

[0096] Optionally, after determining the first weight coefficient corresponding to the lane departure risk assessment function, the second weight coefficient corresponding to the collision risk assessment function, and the third weight coefficient corresponding to the potential field assessment function, the lane departure risk assessment function can be weighted and fused with the first weight coefficient, the collision risk assessment function with the second weight coefficient, and the potential field assessment function with the third weight coefficient to obtain the initial driving risk assessment function, which can be expressed as the following formula (8).

[0097] (8)

[0098] Optionally, after obtaining the initial driving risk assessment function, the initial driving risk assessment function can be summed at multiple discrete time points to obtain the driving risk assessment function, which can be expressed by the following formula (9).

[0099] (9)

[0100] Optionally, after obtaining the driving risk assessment function for the aforementioned vehicle, for each candidate driving trajectory, the first motion state parameters of the vehicle at each discrete time point in the future time period determined by the aforementioned candidate driving trajectory, and the second motion state parameters of the candidate risk vehicle at each discrete time point in the future time period, are calculated and processed according to the input parameters required by the aforementioned driving risk assessment function, and then input into the aforementioned driving risk assessment function to obtain the risk value of each candidate driving trajectory. The risk value is then determined as the risk assessment result of the vehicle driving according to the candidate driving trajectory. The larger the risk value, the higher the degree of risk when the vehicle drives according to the candidate driving trajectory.

[0101] As an optional implementation, step S101 involves determining the vehicle's risk assessment area from the preset area based on the vehicle's driving status information within the current time period and the driving status information of other vehicles within the preset area. This includes: predicting the collision risk level between the vehicle and other vehicles based on the vehicle's driving status information within the preset area and the driving status information of other vehicles within the preset area; and determining the area where other vehicles within the preset area have a collision risk level higher than the preset collision risk level as the vehicle's risk assessment area.

[0102] In this embodiment, the aforementioned risk assessment area is used to represent the dynamic interactive area that poses a real threat to the vehicle. Determining the risk assessment area of ​​the vehicle from the preset area can avoid invalid calculations for irrelevant targets and improve assessment efficiency and real-time performance.

[0103] Optionally, based on the real-time driving status information of the vehicle and all other vehicles within a preset area at the current moment, including dynamic parameters in the Frenet coordinate system such as position, speed, acceleration, and heading angle, a preliminary prediction of the potential collision risk level between the vehicle and each other vehicle can be made. This prediction is not based on static distance judgment, but rather comprehensively considers the relative longitudinal speed, lateral approach rate, velocity vector angle, and current distance between the two vehicles, combined with a preset safety distance model (e.g., a braking safety buffer zone calculated based on the vehicle's deceleration and reaction time), to dynamically estimate the probability or threat level of a collision between the vehicle and other vehicles within the preset area in the near future (e.g., 1-2 seconds). This risk level is a continuous numerical value, ranging from low (no interaction) to high (high collision probability). Its calculation process is entirely based on kinematic and dynamic parameters, requiring no semantic information, and has strong robustness and universality.

[0104] Optionally, after obtaining the collision risk level between the vehicle and other vehicles within a preset area, the collision risk levels of all other vehicles within the preset area can be compared with a preset threshold. This threshold is preset based on the vehicle's operating environment (e.g., highway, city, ramp) and safety level, and is used to distinguish between "potentially threatening targets" and "low-risk irrelevant targets." Only when the predicted collision risk level between another vehicle and the vehicle is higher than this preset threshold is that other vehicle identified as a high-risk vehicle. After evaluating all other vehicles within the preset area, the spatial range of the identified high-risk vehicles (usually a rectangular or elliptical area centered on the vehicle and extending a certain longitudinal and lateral safety distance) can be defined as the vehicle's risk assessment area.

[0105] Optionally, by defining the risk assessment area of ​​the vehicle, the calculation scope of the subsequent risk assessment can be significantly reduced, so that the comprehensive assessment function only needs to be applied to vehicles that actually have a potential collision risk with the vehicle. This significantly reduces the computational overhead and improves the real-time response capability of the system while ensuring safety.

[0106] As an optional implementation, after obtaining the risk assessment results corresponding to multiple candidate driving trajectories, the method further includes: based on the risk assessment results, selecting candidate driving trajectories with a risk level less than a preset threshold from the multiple candidate driving trajectories.

[0107] In this embodiment, after obtaining the risk assessment results corresponding to each of the multiple candidate driving trajectories, a trajectory screening process can be further performed to ensure that the final output candidate driving trajectories meet the hard constraints of safe operation.

[0108] Optionally, the risk level of each candidate driving trajectory can be determined using the risk value calculated in the aforementioned steps. This risk level is then compared with a preset threshold to obtain a comparison result. If the comparison result indicates that the risk level of the candidate driving trajectory is greater than or equal to the preset threshold, it means that the risk of the vehicle driving along that candidate driving trajectory is high. In this case, the candidate driving trajectory can be deleted. Conversely, if the comparison result indicates that the risk level of the candidate driving trajectory is less than the preset threshold, it means that the risk of the vehicle driving along that candidate driving trajectory is within an acceptable range. In this case, the candidate driving trajectory can be retained.

[0109] In this step, multiple candidate driving trajectories are screened using a preset threshold as a reference. This allows for the selection of candidate driving trajectories with a risk level lower than the preset threshold, thereby providing a reliable reference for subsequent vehicle decisions.

[0110] The above technical solutions of the embodiments of this application will be further illustrated below with reference to preferred embodiments.

[0111] Figure 2 This is a flowchart of a vehicle trajectory risk assessment method according to an embodiment of this application, such as... Figure 2 As shown, the method includes the following steps.

[0112] Step S201: Information acquisition and preliminary processing.

[0113] In this embodiment, the vehicle can collect real-time environmental and vehicle status information using onboard multi-source sensors. These sensors include LiDAR, a high-resolution camera, a GPS positioning module, an inertial measurement unit, and wheel speedometers, used to acquire data such as the position, speed, and heading angle of surrounding vehicles, as well as the vehicle's absolute position, longitudinal / lateral velocity, acceleration, and attitude angles. This raw data is initially in a global Cartesian coordinate system, which is subject to noise interference and coordinate system inconsistency.

[0114] Optionally, to improve data consistency and semantic usability, sensor data can be time-synchronized and spatially registered to remove outliers and isolated points. Combined with the lane topology provided by high-precision maps, lane-level matching of vehicle positions can be performed. Then, the vehicle's driving state information is converted to the Frenet coordinate system, establishing a local reference frame based on the lane centerline as the longitudinal reference axis and the longitudinal arc length and lateral offset distance. Similarly, the driving state information of other vehicles within a preset area surrounding the vehicle can also be converted to the Frenet coordinate system. This conversion process transforms the vehicle state from "global absolute coordinates" to "road semantic coordinates," laying a structured and interpretable mathematical foundation for subsequent risk area delineation and risk modeling.

[0115] Step S202: Based on the processed information, construct the risk assessment area for the vehicle.

[0116] In this embodiment, in the Frenet coordinate system, the influence range of other vehicles that pose a potential interactive threat to the vehicle is defined by dynamically identifying them, forming a variable-size, adaptive driving risk assessment area. This risk assessment area is not a fixed geometric boundary, but is intelligently determined based on the relative longitudinal distance, lateral offset, relative speed, and speed difference trend between the vehicle and surrounding vehicles.

[0117] For example, vehicles located within a certain longitudinal distance (e.g., 50 meters in front, 20 meters behind) of the vehicle and with a lateral offset less than a lane width threshold (e.g., ±1.5 meters) can be selected as candidate risk vehicles. Further analysis of relative speed and acceleration information is then conducted to determine if the vehicle exhibits "intended approach" or "lane change conflict" characteristics. For instance, if a vehicle's longitudinal distance is less than the safe distance and its lateral speed is significantly positive (approaching its own lane), that vehicle is included in the high-priority risk targets. Ultimately, the spatial area encompassing all identified candidate risk vehicles constitutes the vehicle's risk assessment area. This area has dynamic shrinking / expanding capabilities, focusing only on local spaces where genuine interaction risks exist, significantly reducing computational redundancy and improving system real-time performance.

[0118] Step S203: Based on the vehicle's driving status information and the driving status information of other vehicles within the risk assessment area, construct a driving risk assessment function.

[0119] In this embodiment, the driving risk assessment function is used to quantify the comprehensive safety risks faced by a vehicle executing an arbitrary trajectory in the current environment. The driving risk assessment function consists of three sub-risk functions: lane departure risk function, collision risk assessment function, and potential field assessment function. Among them, the lane departure risk assessment function can be referred to in the aforementioned formula (1), the collision risk assessment function can be referred to in the aforementioned formula (4), and the potential field assessment function can be referred to in the aforementioned formula (6).

[0120] After normalization, the three sub-risk functions are merged into a comprehensive driving risk assessment function based on their weights. Collision risk is given the primary weight of 50%, while lane departure and potential field influence each account for 25%, forming an assessment logic of "safety first, stability second." Ultimately, this driving risk assessment function can output a continuous and comparable comprehensive risk scalar value for any candidate trajectory, achieving a precise mapping from "state information" to "risk quantification."

[0121] Step S204: Based on the driving risk assessment function, perform a risk assessment on the candidate driving trajectories of the vehicle.

[0122] In this step, the autonomous driving decision module has generated a set of candidate driving trajectories. Each candidate trajectory is sampled as a six-dimensional state sequence at multiple discrete time points within a future time period. Each candidate trajectory is sequentially input into the aforementioned driving risk assessment function to calculate its total risk value in the Frenet coordinate system. If the average or maximum risk value of a candidate trajectory within the future time period is lower than a preset risk threshold, the candidate trajectory is classified as a "low-risk trajectory" and retained. Conversely, if the average or maximum risk value of a candidate trajectory within the future time period exceeds the risk threshold, it is classified as a "high-risk trajectory" and discarded. This process rapidly filters the candidate trajectory set, retaining only those that meet safety constraints as input for subsequent trajectory smoothing, control execution, and path tracking. The entire evaluation process can be executed in parallel, ensuring multi-trajectory evaluation is completed within milliseconds, meeting real-time control requirements.

[0123] In steps S201 to S204 above, road semantic modeling is achieved by introducing the Frenet coordinate system, significantly improving the rationality and interpretability of state representation; by constructing a dynamic risk assessment region, the focus is on vehicles that actually pose a collision risk, reducing computational load; by fusing lane departure risk assessment functions, collision risk assessment functions, and potential field assessment functions, a high-fidelity, weighted comprehensive driving risk assessment function is constructed, achieving accurate modeling of complex traffic intentions. Finally, through a risk threshold-driven trajectory selection mechanism, an end-to-end closed loop of "assessment as decision" is achieved, greatly improving the safety, real-time performance, and robustness of trajectory selection.

[0124] Figure 3 This is a flowchart of a method for constructing a driving risk assessment function according to an embodiment of this application, such as... Figure 3 As shown, the method includes the following steps.

[0125] Step S301: Construct a lane departure risk assessment function, construct a potential field assessment function for other vehicles within the risk assessment area, and construct a collision risk assessment function between the vehicle and other vehicles.

[0126] In this embodiment, to accurately quantify the multi-dimensional safety risks of autonomous vehicles in complex and dynamic traffic environments, three independent but interconnected sub-risk assessment functions are constructed as the basic modules for comprehensive risk assessment. All functions take the vehicle's six-dimensional state parameters in the Frenet coordinate system as input, ensuring that the modeling process is highly coupled with the road geometry and avoiding nonlinear coupling errors caused by road curvature in the Cartesian coordinate system.

[0127] Optionally, a lane departure risk assessment function is constructed. This function is used to quantify the potential safety risks caused by the vehicle deviating from the center line of the current driving lane, and can be expressed by the aforementioned formula (1). A collision risk assessment function between the vehicle and other vehicles is constructed. This function is used to assess the probability of the vehicle colliding with each candidate risk vehicle in the future time period, and can be expressed by the aforementioned formula (4). An adaptive potential field assessment function for other vehicles within the risk assessment area is constructed. This function draws on the concept of "potential field" in physics, treating each candidate risk vehicle as a dynamic potential source with a potential field. Its influence decreases with distance and increases with relative motion. This function can be expressed by the aforementioned formula (6).

[0128] Step S302: Construct the driving risk assessment function.

[0129] In this embodiment, after constructing the three independent risk sub-functions, the three risk sub-functions can be weighted and fused to obtain the driving risk assessment function, which comprehensively represents the overall safety status of the vehicle in the current environment. This driving risk assessment function can be expressed as the aforementioned formula (9), which will not be elaborated further here.

[0130] In steps S301 to S302 above, lane departure risk assessment function, collision risk assessment function, and potential field assessment function are synchronously integrated within the same framework. This avoids the one-sidedness of traditional methods that only focus on a single distance or speed threshold, and comprehensively covers typical safety threat scenarios in autonomous driving (such as lane departure, being cut off, rear-end collision, oblique collision, etc.), which is a reliable basis for subsequent assessment of vehicle driving risks.

[0131] Figure 4 This is a flowchart of a risk assessment method for candidate driving trajectories of a vehicle according to an embodiment of this application, such as... Figure 4 As shown, the method includes the following steps.

[0132] Step S401: Select a candidate driving trajectory from the candidate trajectory set.

[0133] In this embodiment, the autonomous driving decision-making module has pre-generated a set of candidate driving trajectories, which includes multiple candidate driving trajectories. A candidate driving trajectory can be extracted from the set of candidate driving trajectories for trajectory risk assessment.

[0134] Step S402: Calculate the risk value of the candidate driving trajectory based on the driving risk assessment function.

[0135] In this embodiment, after selecting a candidate driving trajectory, the comprehensive driving risk assessment function constructed in the preceding step S302 is called to assess the risk of the candidate driving trajectory and calculate the risk value of the candidate driving trajectory. The risk value is used to characterize the degree of risk when the vehicle drives along the candidate trajectory.

[0136] Step S403: Determine whether the risk value is greater than the preset risk threshold.

[0137] In this embodiment, after obtaining the risk value of a candidate driving trajectory, the risk value can be compared with a preset risk threshold to obtain a comparison result. If the comparison result indicates that the risk value of the candidate driving trajectory is less than the preset risk threshold, it means that the risk of the candidate driving trajectory is within an acceptable range, and in this case, step S404 is executed. Conversely, if the comparison result indicates that the risk value of the candidate driving trajectory is greater than or equal to the preset risk threshold, it means that the risk corresponding to the candidate driving trajectory is high, and in this case, step S405 is executed.

[0138] Step S404: Retain the candidate driving trajectory.

[0139] In this embodiment, when the comparison result shows that the risk value of the candidate driving trajectory is less than the preset risk threshold, the candidate driving trajectory is marked as an "executable trajectory" and stored in a temporary "safe trajectory set" for use by the subsequent trajectory smoothing, optimization and control modules.

[0140] Step S405: Delete the candidate driving trajectory.

[0141] In this embodiment, if the comparison result shows that the risk value of a candidate driving trajectory is greater than or equal to a preset risk threshold, the trajectory is immediately removed from the candidate trajectory set and marked as a "high-risk eliminated trajectory." This trajectory will no longer participate in subsequent control outputs, avoiding emergency braking, lane change conflicts, or collisions caused by selecting this path. This effectively reduces invalid or dangerous decision options, significantly lowering the computational burden and execution risk of the control system.

[0142] Step S406: Determine whether the candidate driving trajectory is the last candidate driving trajectory in the candidate trajectory set.

[0143] In this embodiment, it is checked whether the currently evaluated candidate driving trajectory is the last one in the candidate driving trajectory set. If not, the process returns to step S401, continues to select the next unevaluated candidate driving trajectory, and repeats the evaluation process from S402 to S405; if yes, the process terminates and proceeds to step S407. This loop structure ensures that all candidate driving trajectories are evaluated completely and without omission, achieving "full coverage and no blind spots" for safe screening.

[0144] Step S407: Risk assessment of candidate driving trajectories ends.

[0145] In this embodiment, once all candidate driving trajectories have been evaluated, the system enters the evaluation completion state. At this point, the trajectories retained in the safe trajectory set are the feasible trajectories that meet the safety constraints. This safe trajectory set will be passed to the downstream trajectory optimization module or directly executed by the trajectory tracking controller to guide the vehicle in making safe, stable, and compliant driving decisions. This process is completed within milliseconds, meeting real-time control requirements and ensuring efficient and reliable closed-loop response from perception to execution.

[0146] In steps S401 to S407 above, the high-fidelity, multi-dimensional comprehensive risk assessment function constructed in the previous steps is directly applied to the trajectory-level assessment, so that the "safety score" of each candidate driving trajectory has physical interpretability and engineering quantifiability. This breaks through the traditional coarse-grained screening method that only relies on geometric distance or empirical rules, improves the accuracy of risk assessment of candidate driving trajectories of vehicles, and provides reliable data basis for subsequent vehicle driving decisions.

[0147] 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.

[0148] According to an embodiment of this application, a vehicle trajectory evaluation device is provided. It should be noted that the device can be used to perform the above-described vehicle trajectory evaluation method.

[0149] Figure 5 This is a schematic diagram of a vehicle trajectory evaluation device according to an embodiment of this application, as shown below. Figure 5 As shown, the trajectory evaluation device 500 for the vehicle includes: a first determination unit 501, a second determination unit 502, a prediction unit 503, and an evaluation unit 504.

[0150] The first determining unit 501 determines the risk assessment area of ​​the vehicle from the preset area based on the vehicle's driving status information in the preset area during the current time period and the driving status information of other vehicles in the preset area.

[0151] The second determining unit 502 is used to determine at least one candidate risk vehicle from other vehicles within the risk assessment area, wherein the candidate risk vehicle is a vehicle that has a collision risk with the vehicle.

[0152] The prediction unit 503 is used to predict the driving status information of the candidate risk vehicle in a future time period based on the driving status information of the candidate risk vehicle.

[0153] The evaluation unit 504 is used to evaluate the risk of multiple candidate driving trajectories of the vehicle based on the vehicle's driving risk evaluation function and the driving status information of the candidate risk vehicles in the future time period, and to obtain the risk evaluation results corresponding to the multiple candidate driving trajectories. The risk evaluation results are used to characterize the degree of risk when the vehicle drives according to the candidate driving trajectory.

[0154] Optionally, the prediction unit 503 is further configured to: input the driving status information of the candidate risk vehicle into the speed prediction algorithm for prediction, and obtain the driving status information of the candidate risk vehicle in the future time period, wherein the speed prediction algorithm is a model pre-trained based on driving status information samples of different time periods.

[0155] Optionally, the evaluation unit 504 is also used to: traverse multiple candidate driving trajectories; and perform risk assessment on the currently traversed candidate driving trajectories based on the vehicle's driving risk assessment function and the driving status information of the candidate risk vehicles in the future time period, so as to obtain the risk assessment result of the candidate driving trajectories.

[0156] Optionally, the evaluation unit 504 is further configured to: determine the second motion state parameters of the candidate risk vehicle at multiple discrete time points in the future time period based on the driving state information of the candidate risk vehicle in the future time period; input the first motion state parameters at multiple discrete time points and the second motion state parameters at multiple discrete time points into the driving risk evaluation function for risk evaluation, and obtain the risk evaluation result of the candidate driving trajectory.

[0157] Optionally, the device 500 is further configured to: construct a driving risk assessment function based on a lane departure risk assessment function, a collision risk assessment function, and a potential field assessment function, wherein the lane departure risk assessment function is used to assess the degree of deviation between the driving lanes of the vehicles when the vehicles are traveling along the candidate driving trajectory at a target discrete time point; the collision risk assessment function is used to assess the collision risk between the vehicles and at least one candidate risk vehicle when the vehicles are traveling along the candidate driving trajectory at a target discrete time point; and the potential field assessment function is used to characterize the degree of influence of the candidate risk vehicle on the vehicles traveling along the candidate driving trajectory at a target discrete time point, wherein the target discrete time point is any one of a plurality of discrete time points.

[0158] Optionally, the device 500 is further configured to: determine a first weighting coefficient corresponding to the lane departure risk assessment function, a second weighting coefficient corresponding to the collision risk assessment function, and a third weighting coefficient corresponding to the potential field assessment function, wherein the first weighting coefficient is less than the second weighting coefficient and the first weighting coefficient is equal to the third weighting coefficient; perform weighted fusion of the lane departure risk assessment function with the first weighting coefficient, the collision risk assessment function with the second weighting coefficient, and the potential field assessment function with the third weighting coefficient to obtain an initial driving risk assessment function; and sum the initial driving risk assessment function at multiple discrete time points to obtain the driving risk assessment function.

[0159] Optionally, the first determining unit 501 is further configured to: predict the collision risk level between the vehicle and other vehicles based on the vehicle's driving status information within the preset area during the current time period and the driving status information of other vehicles within the preset area; and determine the area where other vehicles within the preset area are located, whose collision risk level with the vehicle is higher than the preset collision risk level, as the vehicle's risk assessment area.

[0160] Optionally, the device 500 is also used to: based on the risk assessment results, select candidate driving trajectories from multiple candidate driving trajectories whose risk level is less than a preset threshold.

[0161] In the trajectory evaluation device for the vehicle described in this application, the risk assessment area of ​​the vehicle is delineated by using the driving status information of the vehicle and other surrounding vehicles. This allows the focus to be placed solely on candidate risk vehicles within the risk assessment area that pose a potential collision risk, avoiding invalid calculations for irrelevant targets. Subsequently, the driving status information of the candidate risk vehicles in the future is predicted, breaking through the traditional static evaluation model that relies solely on current-time information. This achieves a forward-looking prediction of the driving status of candidate risk vehicles in the future time period. Finally, using the vehicle's driving risk assessment function and the driving status information of the candidate risk vehicles in the future time period, risk assessments are performed on multiple candidate driving trajectories of the vehicle, accurately determining the degree of risk when the vehicle travels along multiple candidate driving trajectories. In other words, this application considers the driving status information of candidate risk vehicles in the future time period and evaluates the degree of risk when a vehicle travels along multiple candidate driving trajectories through the vehicle's driving risk assessment function. This overcomes the lag defect of traditional static assessment and constructs a precise screening mechanism based on dynamic risk assessment areas. It significantly improves the accuracy of risk assessment of vehicle driving trajectories in complex traffic scenarios, provides efficient and reliable safety decision support for autonomous driving systems, and solves the technical problem of not being able to accurately assess the driving risks of vehicles in related technologies.

[0162] Embodiments of this application also provide an electronic device. Figure 6 This is a schematic diagram of an electronic device according to an embodiment of this application, such as... Figure 6 As shown, the electronic device 600 may include a memory 610 and a processor 620, wherein the memory 610 is used to store an executable program; the processor 620 is used to run the executable program stored in the memory 610, wherein the program executes the vehicle driving control method in various embodiments of this application when it runs.

[0163] 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.

[0164] 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.

[0165] 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.

[0166] 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.

[0167] 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.

[0168] 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.

[0169] 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.

[0170] 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.

[0171] 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.

[0172] 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 a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0173] 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 trajectory evaluation method of a vehicle, characterized by, include: Based on the vehicle's driving status information in the preset area during the current time period, and the driving status information of other vehicles in the preset area, the risk assessment area of ​​the vehicle is determined from the preset area. From other vehicles within the risk assessment area, at least one candidate risk vehicle is identified, wherein the candidate risk vehicle is a vehicle that poses a collision risk to the vehicle in question; Based on the driving status information of the candidate risk vehicles, predict the driving status information of the candidate risk vehicles in a future time period. Based on the vehicle's driving risk assessment function and the driving status information of the candidate risk vehicles in the future time period, risk assessments are performed on multiple candidate driving trajectories of the vehicle to obtain risk assessment results corresponding to each candidate driving trajectory. The risk assessment results are used to characterize the degree of risk that exists when the vehicle drives according to the candidate driving trajectory.

2. The method of claim 1, wherein, Based on the driving status information of the candidate risk vehicles, predict the driving status information of the candidate risk vehicles in a future time period, including: The driving status information of the candidate risk vehicle is input into the speed prediction algorithm for prediction to obtain the driving status information of the candidate risk vehicle in the future time period. The speed prediction algorithm is a model pre-trained based on driving status information samples from different time periods.

3. The method of claim 1, wherein, Based on the vehicle's driving risk assessment function and the driving status information of the candidate risk vehicles in the future time period, risk assessments are performed on multiple candidate driving trajectories of the vehicle to obtain risk assessment results corresponding to each candidate driving trajectory, including: Traverse multiple candidate driving trajectories; Based on the vehicle's driving risk assessment function and the driving status information of the candidate risk vehicle in the future time period, a risk assessment is performed on the currently traversed candidate driving trajectory to obtain the risk assessment result of the candidate driving trajectory.

4. The method of claim 3, wherein, The candidate driving trajectory includes first motion state parameters of the vehicle at multiple discrete time points within the future time period. Based on the vehicle's driving risk assessment function and the driving state information of the candidate risk vehicles within the future time period, a risk assessment is performed on the currently traversed candidate driving trajectory to obtain the risk assessment result of the candidate driving trajectory, including: Based on the driving status information of the candidate risk vehicle in the future time period, determine the second motion state parameters of the candidate risk vehicle at multiple discrete time points in the future time period. The first motion state parameters at multiple discrete time points and the second motion state parameters at multiple discrete time points are input into the driving risk assessment function to perform risk assessment and obtain the risk assessment result of the candidate driving trajectory.

5. The method of claim 4, wherein, The method further includes: The driving risk assessment function is constructed based on the lane departure risk assessment function, the collision risk assessment function, and the potential field assessment function. The lane departure risk assessment function assesses the degree of deviation between the vehicle and its driving lane at a target discrete time point when the vehicle is traveling along the candidate driving trajectory. The collision risk assessment function assesses the collision risk between the vehicle and at least one of the candidate risk vehicles at the target discrete time point when the vehicle is traveling along the candidate driving trajectory. The potential field assessment function characterizes the influence of the candidate risk vehicles on the vehicle traveling along the candidate driving trajectory at the target discrete time point. The target discrete time point is any one of the plurality of discrete time points.

6. The method of claim 5, wherein, Based on the lane departure risk assessment function, collision risk assessment function, and potential field assessment function, the driving risk assessment function is constructed, including: Determine the first weighting coefficient corresponding to the lane departure risk assessment function, the second weighting coefficient corresponding to the collision risk assessment function, and the third weighting coefficient corresponding to the potential field assessment function, wherein the first weighting coefficient is less than the second weighting coefficient, and the first weighting coefficient is equal to the third weighting coefficient; The lane departure risk assessment function is weighted and fused with the first weight coefficient, the collision risk assessment function with the second weight coefficient, and the potential field assessment function with the third weight coefficient to obtain the initial driving risk assessment function. The initial driving risk assessment function is summed at the multiple discrete time points to obtain the driving risk assessment function.

7. The method according to claim 1, characterized in that, Based on the vehicle's driving status information within a preset area during the current time period, and the driving status information of other vehicles within the preset area, a risk assessment area for the vehicle is determined from the preset area, including: Based on the vehicle's driving status information within the preset area during the current time period, and the driving status information of other vehicles within the preset area, the collision risk level between the vehicle and the other vehicles is predicted. The area within the preset area where other vehicles with a collision risk level higher than the preset collision risk level are located is defined as the risk assessment area of ​​the vehicle.

8. The method according to any one of claims 1 to 7, characterized in that, After obtaining the risk assessment results corresponding to each of the multiple candidate driving trajectories, the method further includes: Based on the risk assessment results, candidate driving trajectories with a risk level less than a preset threshold are selected from multiple candidate driving trajectories.

9. A vehicle, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program, when running, performs the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored executable program, wherein, when the executable program is executed, it controls the device on which the storage medium is located to perform the method according to any one of claims 1 to 8.

Images