Post-processing access method and device of driving track

Abstract

The embodiment of the invention provides a post-processing access method and device for a driving track. The method comprises the following steps: acquiring at least two driving tracks output by an end-to-end multi-modal model carried by a vehicle end and sent by the vehicle end through a calling service interface; for each driving track, evaluating a safety score of the driving track based on a preset evaluation index; wherein the preset evaluation indexes comprise evaluation indexes related to driving safety; according to the safety score, selecting a target driving track from the at least two driving tracks; wherein the target driving track is the driving track with the highest safety; and returning the target driving track to the vehicle end.

Description

Technical Field

[0001] This specification relates to the field of vehicle control technology, and in particular to a method and apparatus for post-processing access to driving trajectories. Background Technology

[0002] With the rapid development of intelligent assisted driving technology, end-to-end assisted driving decision-making and planning solutions have gradually become a hot topic in the industry. These solutions are usually based on models trained using deep learning technology. After training, the models can directly generate post-processing inputs of driving decisions and driving trajectories based on sensor inputs (such as cameras, LiDAR, etc.).

[0003] Based on the number of generated planned trajectories, driving models can be categorized into single-trajectory and multimodal trajectories. A single-trajectory model outputs only one optimal driving trajectory. However, in the same driving scenario, drivers with different driving habits may make different decisions, thus a single-trajectory model cannot meet practical needs. Multimodal trajectories, on the other hand, perfectly meet these practical requirements. When using end-to-end multimodal model inference, multiple predicted driving trajectories can be output. Furthermore, the model also outputs the selection probability of each driving trajectory and selects the trajectory with the highest probability as the final output.

[0004] However, whether outputting a single trajectory or a multimodal trajectory, the ultimate goal is to output an optimal driving trajectory for the vehicle controller to assist in vehicle driving. This approach has the following problems:

[0005] 1. Black-box characteristics: The inference results of the model lack interpretability and are difficult to trace and verify;

[0006] 2. Uncertainty: Due to the limitations of the training data and the complexity of the model, the output driving trajectory has significant uncertainty;

[0007] 3. Safety risks: Directly providing the output driving trajectory to the vehicle controller can easily lead to safety hazards, especially in extreme or unknown scenarios.

[0008] Therefore, relying directly on the probabilities and trajectories output by the model to select the final trajectory in existing multimodal programming models poses a significant risk in practical applications. Summary of the Invention

[0009] In view of the above, one or more embodiments of this specification provide a post-processing access method and apparatus for driving trajectory, electronic device, and storage medium to solve the problems existing in the related art.

[0010] To achieve the above objectives, one or more embodiments of this specification provide the following technical solutions:

[0011] According to a first aspect of the embodiments of this specification, a post-processing access method for driving trajectories is provided, applied to a server corresponding to a vehicle, wherein the server provides a service interface for the vehicle to call, and the method includes:

[0012] Obtain at least two driving trajectories output by the end-to-end multimodal model mounted on the vehicle, which are sent by the vehicle through a service interface call;

[0013] For each driving trajectory, a safety score is evaluated based on preset evaluation indicators; wherein, the preset evaluation indicators include evaluation indicators related to driving safety.

[0014] Based on the safety score, a target driving trajectory is selected from the at least two driving trajectories; wherein the target driving trajectory is the driving trajectory with the highest safety.

[0015] The target driving trajectory is returned to the vehicle.

[0016] Optionally, the preset evaluation indicators include at least one of the following:

[0017] No-fault collision assessment index, drivable area compliance assessment index, collision time assessment index, speed limit compliance assessment index, trip progress assessment index, comfort assessment index, and driving direction compliance assessment index.

[0018] Optionally, when there is more than one preset evaluation index, the step of evaluating the safety score of the driving trajectory based on the preset evaluation index includes:

[0019] Calculate the single-index score of the driving trajectory under each preset evaluation index;

[0020] The single-index score of each preset evaluation indicator is multiplied by the weight corresponding to the preset evaluation indicator and then summed to obtain the multi-index weight score of the driving trajectory.

[0021] The safety score of the driving trajectory is obtained by dividing the multi-indicator weight score by the sum of the weight values ​​of various preset evaluation indicators.

[0022] Optionally, the calculation of the single-index score for the driving trajectory under each preset evaluation index includes:

[0023] For no-fault collision assessment indicators, determine whether there is any overlap between the vehicle's position coordinates and the obstacle's position coordinates at the same moment in the driving trajectory;

[0024] If so, determine the collision type based on the overlapping obstacles, and evaluate the trajectory segment at the same moment based on the collision type to obtain the evaluation score of the trajectory segment;

[0025] The evaluation score of each segment in the driving trajectory is calculated, and the lowest evaluation score is taken as the no-fault collision evaluation score of the driving trajectory.

[0026] Optionally, the calculation of the single-index score for the driving trajectory under each preset evaluation index includes:

[0027] For the drivable area compliance assessment index, determine whether the vehicle has entered a non-drivable area in the driving trajectory;

[0028] If a vehicle enters a non-drivable area, the minimum score for compliance with the drivable area assessment of the driving trajectory is determined.

[0029] If the vehicle does not enter the drivable area, the score for compliance with the drivable area assessment of the driving trajectory is the maximum score.

[0030] Optionally, the calculation of the single-index score for the driving trajectory under each preset evaluation index includes:

[0031] For the collision time assessment index, the vehicle state at each future moment is obtained based on the driving trajectory; wherein, the vehicle state includes the vehicle's position coordinates, speed, and orientation angle;

[0032] Calculate the future position coordinates of the vehicle at each moment after maintaining the stated speed and orientation angle for a preset duration;

[0033] Determine whether there is any overlap between the future position coordinates and the position coordinates of the obstacle at the same moment;

[0034] If so, determine the collision type based on the overlapping obstacles, and evaluate the trajectory segment at the same moment based on the collision type to obtain the evaluation score of the trajectory segment;

[0035] The evaluation score of each segment in the driving trajectory is calculated, and the lowest evaluation score is taken as the collision time evaluation score of the driving trajectory.

[0036] Optionally, the evaluation of trajectory segments at the same moment based on the collision type to obtain an evaluation score for the trajectory segment includes at least one of the following:

[0037] If the object of collision is a traffic light and the collision type is a red light, then the evaluation score for that trajectory segment is determined as the lower limit score;

[0038] If the object of collision is not a traffic light, and the collision type is a dynamic forward collision, a static obstacle collision, or a lateral collision, then the evaluation score for that trajectory segment is determined as the lower limit score.

[0039] If the object of collision is not a traffic light, and the collision type is not a dynamic forward collision, a static obstacle collision, or a lateral collision, then the evaluation score for that trajectory segment is determined to be the median score.

[0040] Optionally, the calculation of the single-index score for the driving trajectory under each preset evaluation index includes:

[0041] For the speed limit compliance assessment index, the speed limit of each vehicle's location is obtained based on the vehicle's position coordinates in the driving trajectory;

[0042] Subtract the speed limit at the current location from the vehicle's speed corresponding to its position coordinates to obtain the overspeed difference of the vehicle at each position coordinate.

[0043] Calculate the overspeed loss score based on the overspeed difference for each location coordinate;

[0044] The speed limit compliance assessment score of the driving trajectory is obtained by subtracting the speeding loss score from the initial score of the driving trajectory.

[0045] Optionally, the calculation of the single-index score for the driving trajectory under each preset evaluation index includes:

[0046] For the trip progress evaluation indicators, the current position of the vehicle in the driving trajectory is set as the starting point;

[0047] A reference line connecting the starting point and the vehicle's destination;

[0048] The point on the reference line mapped from the starting point is taken as the starting point of the process, and the point on the reference line mapped from the last position point of the driving trajectory is taken as the ending point of the process. The trajectory process between the ending point and the starting point is calculated.

[0049] If the trajectory progress is greater than the progress threshold, the trajectory progress is divided by the target progress to obtain the trip progress evaluation score of the driving trajectory; wherein, the target progress is the largest trajectory progress among the at least two driving trajectories.

[0050] Optionally, the calculation of the single-index score for the driving trajectory under each preset evaluation index includes:

[0051] To assess comfort, the acceleration and angular velocity of the vehicle at different times along the driving trajectory are obtained.

[0052] Determine whether the acceleration and angular velocity states are within a preset range;

[0053] If both the acceleration and angular velocity states are within a preset range, the comfort evaluation score of the driving trajectory is set to the upper limit score.

[0054] If the acceleration and angular velocity states are not within the preset range, the comfort evaluation score of the driving trajectory will be set to the intermediate score.

[0055] Optionally, the calculation of the single-index score for the driving trajectory under each preset evaluation index includes:

[0056] For the driving direction compliance assessment, the drivable area is obtained based on the vehicle's position in the driving trajectory, and the cumulative length of the driving trajectory located in the non-drivable area is calculated.

[0057] The driving direction compliance score of the driving trajectory is calculated based on the cumulative length.

[0058] According to a second aspect of the embodiments of this specification, a post-processing access device for driving trajectories is provided, applied to a server corresponding to a vehicle, the server providing a service interface for the vehicle to call, the device comprising:

[0059] The acquisition unit acquires at least two driving trajectories output by the end-to-end multimodal model mounted on the vehicle terminal, which are sent by the vehicle terminal through a service interface call.

[0060] The evaluation unit assesses the safety score of each driving trajectory based on preset evaluation indicators; wherein, the preset evaluation indicators include evaluation indicators related to driving safety.

[0061] The processing unit selects a target driving trajectory from the at least two driving trajectories based on the safety score; wherein the target driving trajectory is the driving trajectory with the highest safety.

[0062] The return unit then sends the target driving trajectory back to the vehicle.

[0063] Optionally, the preset evaluation indicators include at least one of the following:

[0064] No-fault collision assessment index, drivable area compliance assessment index, collision time assessment index, speed limit compliance assessment index, trip progress assessment index, comfort assessment index, and driving direction compliance assessment index.

[0065] Optionally, when there is more than one preset evaluation index, the evaluation unit, when evaluating the safety score of the driving trajectory based on the preset evaluation index, further includes:

[0066] The scoring calculation subunit calculates the single-index score of the driving trajectory under each preset evaluation index;

[0067] The weighted calculation subunit multiplies the single-index score of each preset evaluation index with the weight corresponding to the preset evaluation index and then sums them to obtain the multi-index weight score of the driving trajectory; the multi-index weight score is divided by the sum of the weight values ​​of various preset evaluation indicators to obtain the safety score of the driving trajectory.

[0068] Optionally, the scoring calculation subunit includes:

[0069] The no-fault collision assessment index calculation subunit determines whether there is an overlap between the vehicle's position coordinates and the obstacle's position coordinates at the same moment in the driving trajectory. If so, it determines the collision type based on the overlapping obstacle and evaluates the trajectory segment at the same moment based on the collision type to obtain the evaluation score of the trajectory segment. It then calculates the evaluation score of each trajectory segment in the driving trajectory and takes the lowest evaluation score as the no-fault collision assessment score of the driving trajectory.

[0070] Optionally, the scoring calculation subunit includes:

[0071] The drivable area compliance assessment index calculation subunit calculates whether the vehicle has entered a non-drivable area in the driving trajectory. If the vehicle has entered a non-drivable area, the drivable area compliance assessment score of the driving trajectory is determined to be the lower limit score. If the vehicle has not entered a non-drivable area, the drivable area compliance assessment score of the driving trajectory is determined to be the upper limit score.

[0072] Optionally, the scoring calculation subunit includes:

[0073] The collision time assessment index calculation subunit acquires the vehicle state of the driving trajectory at each future moment for the collision time assessment index; wherein, the vehicle state includes the vehicle's position coordinates, speed, and orientation angle; calculates the future position coordinates of the vehicle after maintaining the speed and orientation angle for a preset duration at each moment; and determines whether there is any overlap between the future position coordinates and the position coordinates of the obstacle at the same moment.

[0074] If so, determine the collision type based on the overlapping obstacles, and evaluate the trajectory segment at the same moment based on the collision type to obtain the evaluation score of the trajectory segment; count the evaluation scores of each trajectory segment in the driving trajectory, and take the lowest evaluation score as the collision time evaluation score of the driving trajectory.

[0075] Optionally, the evaluation of trajectory segments at the same moment based on the collision type to obtain an evaluation score for the trajectory segment includes at least one of the following:

[0076] If the object of collision is a traffic light and the collision type is a red light, then the evaluation score for that trajectory segment is determined as the lower limit score;

[0077] If the object of collision is not a traffic light, and the collision type is a dynamic forward collision, a static obstacle collision, or a lateral collision, then the evaluation score for that trajectory segment is determined as the lower limit score.

[0078] If the object of collision is not a traffic light, and the collision type is not a dynamic forward collision, a static obstacle collision, or a lateral collision, then the evaluation score for that trajectory segment is determined to be the median score.

[0079] Optionally, the scoring calculation subunit includes:

[0080] The speed limit compliance assessment index calculation subunit, for the speed limit compliance assessment index, obtains the speed limit speed at each location of the vehicle based on the vehicle's position coordinates in the driving trajectory; subtracts the speed limit speed at each location from the vehicle's speed corresponding to the vehicle's position coordinate to obtain the speeding difference value of the vehicle at each position coordinate; calculates the speeding loss score based on the speeding difference value of each position coordinate; and subtracts the speeding loss score from the initial score of the driving trajectory to obtain the speed limit compliance assessment score of the driving trajectory.

[0081] Optionally, the scoring calculation subunit includes:

[0082] The trip progress evaluation index calculation subunit, for the trip progress evaluation index, sets the current position of the vehicle in the driving trajectory as the starting point; connects the starting point and the vehicle's destination with a reference line; maps the point on the reference line to the starting point as the process start point, and maps the last position point of the driving trajectory to the point on the reference line as the process end point; calculates the trajectory progress between the process end point and the process start point; if the trajectory progress is greater than the process threshold, divides the trajectory progress by the target progress to obtain the trip progress evaluation score of the driving trajectory; wherein, the target progress is the largest trajectory progress among the at least two driving trajectories.

[0083] Optionally, the scoring calculation subunit includes:

[0084] The comfort evaluation index calculation subunit acquires the acceleration and angular velocity states of the vehicle at different times along the driving trajectory for the comfort evaluation index; determines whether the acceleration and angular velocity states are within a preset range; if both the acceleration and angular velocity states are within the preset range, the comfort evaluation score of the driving trajectory is set to the upper limit score; if the acceleration and angular velocity states are not both within the preset range, the comfort evaluation score of the driving trajectory is set to the middle score.

[0085] Optionally, the scoring calculation subunit includes:

[0086] The driving direction compliance assessment calculation subunit, for driving direction compliance assessment, obtains the drivable area based on the vehicle's position in the driving trajectory, calculates the cumulative length of the driving trajectory located in the non-drivable area, and calculates the driving direction compliance score of the driving trajectory based on the cumulative length.

[0087] According to a third aspect of the embodiments of this specification, an electronic device is provided, including a communication interface, a processor, a memory, and a bus, wherein the communication interface, the processor, and the memory are interconnected via the bus;

[0088] The memory stores machine-readable instructions, and the processor executes the above method by invoking the machine-readable instructions.

[0089] According to a fourth aspect of the embodiments of this specification, a machine-readable storage medium is provided, the machine-readable storage medium storing machine-readable instructions, which, when invoked and executed by a processor, implement the above-described method.

[0090] The technical solutions provided in the embodiments of this specification may include the following beneficial effects:

[0091] The design incorporates a service interface that defines input and output data, facilitating easy portability to the post-processing of the end-to-end multimodal model. Specifically, the end-to-end multimodal model invokes the service interface to obtain post-processing based on preset evaluation metrics. This post-processing uses these metrics to assess the safety of each of the at least two driving trajectories output by the end-to-end multimodal model, thereby helping the model select the safest target driving trajectory from among multiple driving trajectories. Attached Figure Description

[0092] Figure 1 A schematic diagram of a service interface provided for an exemplary embodiment of this specification;

[0093] Figure 2 A flowchart illustrating a post-processing access method for driving trajectories, provided as an exemplary embodiment of this specification;

[0094] Figure 3 A schematic diagram illustrating environmental information provided for an exemplary embodiment of this specification;

[0095] Figure 4 A schematic diagram of the structure of an electronic device containing a post-processing access device for driving trajectory, provided as an exemplary embodiment of this specification;

[0096] Figure 5 This is a block diagram of a post-processing access device for driving trajectory provided as an exemplary embodiment of this specification. Detailed Implementation

[0097] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with one or more embodiments of this specification. Rather, they are merely examples of apparatuses and methods consistent with some aspects of one or more embodiments of this specification as detailed in the appended claims.

[0098] It should be noted that the steps of the corresponding methods are not necessarily performed in the order shown and described in this specification in other embodiments. In some other embodiments, the methods may include more or fewer steps than described in this specification. Furthermore, a single step described in this specification may be broken down into multiple steps in other embodiments; and multiple steps described in this specification may be combined into a single step in other embodiments.

[0099] The following section, with reference to the accompanying drawings, provides a detailed explanation of the post-processing access method for the driving trajectory described in this manual.

[0100] First, let's introduce... Figure 1 The service interface designed in the embodiments shown in this specification is illustrated. For example... Figure 1 As shown, the service interface can include two sub-interfaces: an environmental information input interface and a model output interface. The model output interface provides output services for the end-to-end multimodal model mounted on the vehicle, ensuring that the model's output information conforms to relevant data standards. Similarly, the environmental information input interface provides input services for sensors (such as cameras and LiDAR) mounted on the vehicle, ensuring that the environmental information collected by the sensors conforms to relevant data standards. In other words, the service interface defines relevant data standards to convert the driving trajectory or predicted trajectory output by the model into a driving trajectory or predicted trajectory that conforms to relevant data standards, and to convert the environmental information input by the sensors into environmental information that conforms to relevant data standards. The environmental information input interface and the model output interface are described below:

[0101] [Environmental Information Input Interface]

[0102] 1) Surrounding map information

[0103] a. Driving area

[0104] The drivable area for vehicles includes road areas and parking areas, specifically involving lane and speed limit information for each area. When drivable area information is required, it is necessary to combine the vehicle's location to obtain a target list of all road areas and parking areas within a 50-meter radius (the range can be customized according to actual conditions) of the desired type.

[0105] b. Traffic light status

[0106] Traffic light status includes four states: red, green, yellow, and unknown. To obtain the traffic light status, you need to combine this information with your vehicle's location to retrieve a list of traffic light statuses.

[0107] 2) Target information

[0108] a. Dynamic obstacles

[0109] Dynamic obstacle information includes location, heading angle, speed, size, and type. Types specifically include vehicles, pedestrians, and bicycles. When dynamic obstacle information is needed, it must be combined with the vehicle's position to obtain a list of dynamic obstacles.

[0110] b. Static obstacles

[0111] The static obstacle list includes information on location, heading angle, size, and type. Types specifically include roadblocks, guardrails, cones, and general obstacles. When static obstacle information is needed, it must be obtained by combining the vehicle's position with the list of static obstacles within a defined range.

[0112] [Model Output (Post-processing Method Input) Interface]

[0113] 1) Driving trajectory planned by multimodal model

[0114] The driving trajectory (numbered 0, 1, ..., N) and the selection probability of each driving trajectory, wherein each driving trajectory contains the vehicle's position and heading angle information at each future time (0-T).

[0115] 2) Predict the target trajectory

[0116] The model will identify K surrounding targets and predict the trajectory of each target, where each target has a predicted trajectory, and each predicted trajectory contains the target's position and heading angle information at every future time (0-T).

[0117] It is understood that the service interface can be provided by a server, which may be a service platform provided by the vehicle brand. The server exposes this service interface so that the vehicle can call it to obtain post-processing data of the driving trajectory provided by the server. This facilitates the integration of post-processing methods for driving trajectories into all end-to-end multimodal models on all vehicles, enabling the evaluation of the safety of daily driving trajectories output by the end-to-end multimodal model.

[0118] Please refer to the following for further information. Figure 2 , Figure 2 A flowchart illustrating a post-processing access method for driving trajectories, provided as an exemplary embodiment of this specification, is applied to a server corresponding to a vehicle. The server provides a service interface for the vehicle to call. The method includes:

[0119] Step 110: Obtain at least two driving trajectories output by the end-to-end multimodal model.

[0120] End-to-end multimodal models can generate driving trajectories for driver assistance or prompts based on data collected by sensors (such as cameras and lidar) on the vehicle. To meet the different driving habits of different drivers and the accuracy of responding to road conditions, at least two different driving trajectories can usually be output.

[0121] Each driving trajectory output by the end-to-end multimodal model is a predicted planned trajectory of the vehicle over a certain period of time (let's say T seconds). (The trajectory represents the vehicle's position at each future moment, and the number of trajectories can be set according to the actual needs of the model.) Each driving trajectory can also correspond to a selection probability and the predicted trajectory of surrounding objects over the next T seconds.

[0122] Taking a single inference iteration as an example, the model will output the information shown above. In addition, the model needs to obtain real-time updated information about the vehicle's surrounding environment (such as...) for inference. Figure 2 The system displays the drivable area, speed limit information, lane markings, traffic lights, etc., and target object information (types include dynamic obstacles and static obstacles). The output driving trajectory is then evaluated as follows.

[0123] Step 120: For each driving trajectory, evaluate the safety score of the driving trajectory based on preset evaluation indicators; wherein, the preset evaluation indicators include evaluation indicators related to driving safety.

[0124] It should be noted that the length and width of the vehicle are known parameters. A Cartesian coordinate system is established with the center of the rear axle of the vehicle as the origin, and the vehicle is simplified to a rectangle. Given the position of the center of the rear axle of the vehicle, the position coordinates of the four vertices of the vehicle rectangle can be obtained as polygon[(x1,y1),(x2,y2),(x3,y3),(x4,y4)].

[0125] Centered on the vehicle's position at every future moment along each driving trajectory, information such as the coordinates of drivable areas (lane lines, lane connectors) and obstacles within a 100m radius (the range can be customized based on actual conditions) is obtained from environmental information. Similar to the vehicle's polygon coordinates, obstacle polygon coordinates can also be constructed. By comparing the vehicle's polygon coordinates with each of the above drivable area location information pairwise, the overlap, number, type, and speed limit information of each drivable area can be obtained.

[0126] Next, based on the information obtained, the following judgments can be made:

[0127] (1) For each lane, whether it contains any of the four vertices of the vehicle rectangle, further count the total number of lanes that meet the above conditions. If the total number of lanes is greater than 1, then the vehicle is considered to be in multiple lanes and a bool variable is created to store the result of whether it is in multiple lanes.

[0128] (2) For each lane, check whether the four vertices of the vehicle rectangle are all within the lane. If any lane satisfies this condition, the vehicle is considered to be in a single lane and a bool variable is created to store the result of whether it is in a single lane.

[0129] (3) For each time step, determine whether the four vertices of the current vehicle rectangle are in the feasible region, count the number of vertices that meet the condition, and if <4, consider the vehicle to be in the drivable region. Create a bool variable to store the result of whether it is in the drivable region.

[0130] (4) Obtain the lane line number of the current route drivable area from the environmental information, check whether the vehicle center point is in these areas, and consider the point that is not on the lane line to be in the opposite lane. Create a bool variable to store the result of whether it is in the opposite lane.

[0131] The above method can obtain the vehicle's footprint data in the driving trajectory. This footprint data can be used for the evaluation and calculation of the preset evaluation indicators (the relevant content will be introduced in subsequent embodiments).

[0132] In this specification, the preset evaluation indicators can be flexibly configured according to business needs or user needs proposed by target vehicle users. Different preset evaluation indicators can be freely combined to meet the decision-making needs of various specific or personalized driving trajectories.

[0133] In one exemplary embodiment, such as Figure 3 As shown, the preset evaluation indicators may include at least one or a combination of the following:

[0134] No-fault collision assessment index, drivable area compliance assessment index, collision time assessment index, speed limit compliance assessment index, trip progress assessment index, comfort assessment index, and driving direction compliance assessment index.

[0135] It is understandable that the score range for each of the above preset evaluation indicators can be the same or different. Generally, to facilitate comparison between different preset evaluation indicators, the same score range can be used. For example, the score range for each preset evaluation indicator is [0,1]; where 0 is the lower limit score and 1 is the upper limit score. The lower the score, the less ideal the safety assessment of the driving trajectory, and the higher the score, the more ideal the safety assessment of the driving trajectory.

[0136] The following section will introduce the evaluation calculation process for each of the above-mentioned preset evaluation indicators.

[0137] The [No-Fault Collision Assessment Metric] may include the following steps a, b, and c:

[0138] a. For the no-fault collision assessment index, determine whether there is an overlap between the position coordinates of the vehicle and the position coordinates of the obstacle at the same moment in the driving trajectory.

[0139] In some examples, the no-fault collision assessment score of the driving trajectory can be initialized, such as by setting the initial value to 1. Next, the position coordinates of the vehicle at each future moment in the driving trajectory (such as the coordinates of the vehicle's rectangle polygon as shown above) are obtained.

[0140] For each moment, the vehicle's coordinates are used to determine the overlap relationship with the predicted obstacle's position coordinates at the same moment, and the trajectory that overlaps with the obstacle and the obstacle number are recorded.

[0141] b. If so, determine the collision type based on the existence of overlapping obstacles, and evaluate the trajectory segment at the same moment based on the collision type to obtain the evaluation score of the trajectory segment.

[0142] For example, if the obstacle is a traffic light and the collision type is red (determined based on the red light status of the traffic light), it means that the vehicle is at risk of running a red light at that moment. Therefore, the evaluation score of this trajectory segment can be determined as the lower limit score (e.g., 0 points).

[0143] If the object of collision is not a traffic light, and the collision type is dynamic forward collision, static obstacle collision, or lateral collision, then the evaluation score for that trajectory segment is set as the lower limit score (e.g., 0 points).

[0144] If the object of collision is not a traffic light, and the collision type is not a dynamic forward collision, a stationary obstacle collision, or a lateral collision, then the evaluation score for that trajectory segment will be set to the middle score (e.g., 0.5 points).

[0145] The methods for determining dynamic forward collision, stationary obstacle collision, and lateral collision can be as follows:

[0146] The collision type is determined based on the state information (speed, position) of the vehicle and overlapping obstacles:

[0147] (1) If the vehicle speed is less than the stationary speed threshold (the recommended stationary speed threshold is 0.05 km / h), the vehicle is considered stationary, and the collision type is a stationary collision.

[0148] (2) If the obstacle's speed is less than the stationary speed threshold, then the collision type is a stationary obstacle collision;

[0149] (3) Determine the relative position of the vehicle and the obstacle based on their coordinates. If the vehicle is behind the obstacle, the collision type is dynamic frontal collision.

[0150] (4) Obtain the rear edge line of the vehicle's coordinate position, i.e. the position of the rear line of the vehicle. When the obstacle overlaps with the line, the collision type is dynamic rear collision.

[0151] (5) Except for the above situations, all other collision situations are considered to be lateral collisions.

[0152] c. Calculate the evaluation score of each trajectory segment in the driving trajectory, and take the lowest evaluation score as the no-fault collision evaluation score of the driving trajectory.

[0153] In addition to recording the no-fault collision assessment score, it can also record the collision object number for each driving trajectory.

[0154] The [Drivable Area Compliance Assessment Index] is used to assess whether the driving trajectory encroaches on a non-drivable area.

[0155] For the drivable area compliance assessment index, determine whether the vehicle has entered a non-drivable area in the driving trajectory;

[0156] If a vehicle enters a non-drivable area, the minimum score for compliance with the drivable area assessment of the driving trajectory is determined.

[0157] If the vehicle does not enter the drivable area, the score for compliance with the drivable area assessment of the driving trajectory is the maximum score.

[0158] For example, the system calculates the footprints of the vehicle in each driving trajectory, obtaining the boolean value of each footprint located in a non-drivable area. A value of True indicates a footprint is in a non-drivable area, otherwise False. The driving trajectory's drivable area compliance score is initialized to 1. The system iterates through the boolean value of each footprint's non-drivable area, updating the drivable area compliance score to 0 when a boolean value of True is found, until all footprints have been traversed. Finally, the system outputs the final drivable area compliance score. If the vehicle never enters a non-drivable area, its final drivable area compliance score is 1. If any footprint enters a non-drivable area, its final drivable area compliance score is 0.

[0159] [Collision Time Assessment Metrics]

[0160] For the collision time assessment index, the vehicle state at each future moment is obtained based on the driving trajectory; wherein, the vehicle state includes the vehicle's position coordinates, speed, and orientation angle;

[0161] Calculate the future position coordinates of the vehicle at each moment after maintaining the stated speed and orientation angle for a preset duration;

[0162] Determine whether there is any overlap between the future position coordinates and the position coordinates of the obstacle at the same moment;

[0163] If so, determine the collision type based on the overlapping obstacles, and evaluate the trajectory segment at the same moment based on the collision type to obtain the evaluation score of the trajectory segment;

[0164] The evaluation score of each segment in the driving trajectory is calculated, and the lowest evaluation score is taken as the collision time evaluation score of the driving trajectory.

[0165] It should be noted that the determination of whether there is overlap, the identification of the collision type, and the evaluation score of the trajectory segment based on the collision type are all the same as those in the aforementioned [no-fault collision evaluation index]. The relevant implementation can be referred to the aforementioned embodiments, and will not be repeated here.

[0166] The [Speed ​​Limit Compliance Assessment Index] is used to assess whether a driving trajectory exceeds the speed limit.

[0167] For the speed limit compliance assessment index, the speed limit of each vehicle's location is obtained based on the vehicle's position coordinates in the driving trajectory;

[0168] Subtract the speed limit at the current location from the vehicle's speed corresponding to its position coordinates to obtain the overspeed difference of the vehicle at each position coordinate.

[0169] Calculate the overspeed loss score based on the overspeed difference for each location coordinate;

[0170] The speed limit compliance assessment score of the driving trajectory is obtained by subtracting the speeding loss score from the initial score of the driving trajectory.

[0171] For example, when the overspeed difference is positive, the difference can be retained; when the difference is negative, it can be set to 0.

[0172] Here, based on the overspeed difference at each location coordinate, the overspeed loss score is calculated, which may include:

[0173] Calculate the sum of overspeed differences for all location coordinates and take the average overspeed difference. Divide the average overspeed difference by the maximum overspeed threshold (recommended maximum overspeed threshold 2.3 m / s) to obtain the overspeed loss score.

[0174] As shown above, the initial score can be set to 1. Subtracting the speeding penalty score from the initial score yields the speed limit compliance assessment score for the driving trajectory. If the result of subtracting the speeding penalty score from the initial score is negative, then 0 points is taken as the speed limit compliance assessment score. Furthermore, if the result of subtracting the speeding penalty score from the initial score is greater than 0 but less than 0.5, then 0.5 points can be taken as the speed limit compliance assessment score.

[0175] [Trip Progress Assessment Indicators] are used to assess how close the vehicle is to its destination.

[0176] For the trip progress evaluation indicators, the current position of the vehicle in the driving trajectory is set as the starting point;

[0177] A reference line connecting the starting point and the vehicle's destination;

[0178] The point on the reference line mapped from the starting point is taken as the starting point of the process, and the point on the reference line mapped from the last position point of the driving trajectory is taken as the ending point of the process. The trajectory process between the ending point and the starting point is calculated.

[0179] If the trajectory progress is greater than the progress threshold, the trajectory progress is divided by the target progress to obtain the trip progress evaluation score of the driving trajectory; wherein, the target progress is the largest trajectory progress among the at least two driving trajectories.

[0180] For example, the process threshold can be set according to needs, such as 0.1m. Alternatively, if the trajectory progress is not greater than the process threshold, the travel progress evaluation score of the driving trajectory can be directly set to the upper limit score (e.g., 1 point).

[0181] [Comfort Assessment Index] is used to assess the comfort of a driving trajectory.

[0182] To assess comfort, the acceleration and angular velocity of the vehicle at different times along the driving trajectory are obtained.

[0183] Determine whether the acceleration and angular velocity states are within a preset range;

[0184] If both the acceleration and angular velocity states are within a preset range, the comfort evaluation score of the driving trajectory is set to the upper limit score.

[0185] If the acceleration and angular velocity states are not within the preset range, the comfort evaluation score of the driving trajectory will be set to the intermediate score.

[0186] For example, the velocity and angular velocity states may include longitudinal acceleration, lateral acceleration, jerk, longitudinal jerk, yaw angular velocity, and yaw angular acceleration; correspondingly, determining whether the acceleration and angular velocity states are within a preset range may include the following determination methods:

[0187] (1) Determine whether the longitudinal acceleration at different times is within the upper and lower boundary values ​​(the recommended upper boundary value for longitudinal acceleration is 2.40 m / s², and the lower boundary value for longitudinal acceleration is -4.05 m / s²). If the longitudinal acceleration at all times meets the range, it is True; otherwise, it is False and the bool result is saved.

[0188] (2) Determine whether the lateral acceleration of each trajectory at different times is less than the maximum absolute lateral acceleration boundary value (the recommended maximum absolute lateral acceleration boundary value is 4.89 m / s²). If the lateral acceleration at all times meets this range, it is True; otherwise, it is False and the bool result is saved.

[0189] (3) Determine whether the jerk of each trajectory at different times is less than the maximum absolute jerk boundary value (the recommended maximum absolute jerk boundary value is 8.37 m / s cubed). If the jerk at all times meets this range, it is True; otherwise, it is False and the bool result is saved.

[0190] (4) Determine whether the longitudinal jerk of each trajectory at different times is less than the maximum absolute longitudinal jerk boundary value (the recommended maximum absolute longitudinal jerk boundary value is 4.13 m / s cubed). If the longitudinal jerk at all times meets this range, it is True; otherwise, it is False and the bool result is saved.

[0191] (5) Determine whether the yaw rate of each trajectory at different times is within the upper and lower boundary values ​​(the recommended upper boundary value for yaw rate is 0.95 rad / s, and the lower boundary value for yaw rate is -0.95 rad / s). If the yaw rate at all times meets the range, it is True; otherwise, it is False and the bool result is saved.

[0192] (6) Determine whether the yaw acceleration of each trajectory at different times is within the upper and lower boundary values ​​(the recommended upper boundary value for yaw acceleration is 1.93 rad / s, and the lower boundary value for yaw acceleration is -1.93 rad / s). If the yaw acceleration at all times meets the range, it is True; otherwise, it is False and the bool result is saved.

[0193] After completing the above judgment, further determine whether the longitudinal acceleration, lateral acceleration, jerk, longitudinal jerk, yaw rate and yaw rate of the driving trajectory at different times are all True. If so, set the comfort evaluation score of the driving trajectory to the upper limit score (e.g., 1 point); otherwise, set the comfort evaluation score of the driving trajectory to the middle score (e.g., 0.5 points).

[0194] [Driving Direction Compliance Assessment Index] is used to assess whether the driving trajectory complies with the driving direction.

[0195] For the driving direction compliance evaluation index, the drivable area is obtained based on the location of the vehicle in the driving trajectory, and the cumulative length of the driving trajectory in the non-drivable area is calculated.

[0196] The driving direction compliance score of the driving trajectory is calculated based on the cumulative length.

[0197] For example, the driving direction compliance assessment score can be initialized to a lower limit (e.g., 0 points), and then the driving direction compliance assessment score can be updated and calculated based on the cumulative length mentioned above, as follows:

[0198] When the cumulative length in the non-drivable area is greater than the driving direction compliance distance threshold (the recommended driving direction compliance distance threshold is 4m) but less than the non-drivable area driving distance threshold (the recommended non-drivable area driving distance threshold is 12m), the driving direction compliance assessment score can be updated to the median score (0.5 points).

[0199] When the cumulative length in the non-drivable area is less than the distance threshold for the driving direction, the current trajectory is 1 point.

[0200] When there is more than one preset evaluation index, the evaluation of the safety score of the driving trajectory based on the preset evaluation index includes:

[0201] Calculate the single-index score of the driving trajectory under each preset evaluation index;

[0202] The single-index score of each preset evaluation indicator is multiplied by the weight corresponding to the preset evaluation indicator and then summed to obtain the multi-index weight score of the driving trajectory.

[0203] The safety score of the driving trajectory is obtained by dividing the multi-indicator weight score by the sum of the weight values ​​of various preset evaluation indicators.

[0204] It should be noted that the weights corresponding to the preset evaluation indicators can be set in advance or calculated based on machine learning models or big data technology, and there is no limitation on this.

[0205] Step 130: Select a target driving trajectory from the at least two driving trajectories based on the safety score; wherein the target driving trajectory is the driving trajectory with the highest safety.

[0206] Step 140: Return the target driving trajectory to the vehicle.

[0207] The server can return the target driving trajectory to the vehicle based on the service interface called by the vehicle, so that the end-to-end multimodal model on the vehicle can select the target driving trajectory. Since the safety score represents the safety level of the driving trajectory, the higher the safety score, the safer the assisted driving is according to the driving trajectory. Therefore, the safety score can be used to help the model select the target driving trajectory.

[0208] In summary, the above embodiments designed a service interface to define input and output data, thereby facilitating the portability of post-processing to end-to-end multimodal models. Specifically, the end-to-end multimodal model obtains post-processing based on preset evaluation metrics by calling the service interface. This post-processing uses the preset evaluation metrics to assess the safety of each of the at least two driving trajectories output by the end-to-end multimodal model, thereby helping the end-to-end multimodal model select the safest target driving trajectory from multiple driving trajectories.

[0209] In an exemplary embodiment of this specification, an apparatus capable of implementing the above-described method is also provided.

[0210] Figure 4 This is a schematic structural diagram of a device provided in an exemplary embodiment. Please refer to... Figure 4 At the hardware level, the device includes a processor, internal bus, network interface, memory, and non-volatile memory, and may also include other hardware required for business operations. One or more embodiments of this specification can be implemented in software, such as the processor reading the corresponding computer program from non-volatile memory into memory and then running it. Of course, besides software implementation, one or more embodiments of this specification do not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution entity of the following processing flow is not limited to individual logic units, but can also be hardware or logic devices.

[0211] Please refer to Figure 5 In one software implementation, a post-processing access device for driving trajectory is provided, applied to a server corresponding to a vehicle, wherein the server provides a service interface for the vehicle to call, and the device includes:

[0212] The acquisition unit 510 acquires at least two driving trajectories output by the end-to-end multimodal model mounted on the vehicle terminal, which are sent by the vehicle terminal through a service interface call.

[0213] The evaluation unit 520 evaluates the safety score of each driving trajectory based on preset evaluation indicators; wherein, the preset evaluation indicators include evaluation indicators related to driving safety.

[0214] Processing unit 530 selects a target driving trajectory from the at least two driving trajectories based on the safety score; wherein the target driving trajectory is the driving trajectory with the highest safety.

[0215] The return unit 540 returns the target driving trajectory to the vehicle.

[0216] Optionally, the preset evaluation indicators include at least one of the following:

[0217] No-fault collision assessment index, drivable area compliance assessment index, collision time assessment index, speed limit compliance assessment index, trip progress assessment index, comfort assessment index, and driving direction compliance assessment index.

[0218] Optionally, when there is more than one preset evaluation index, the evaluation unit 520, when evaluating the safety score of the driving trajectory based on the preset evaluation index, further includes:

[0219] The scoring calculation subunit calculates the single-index score of the driving trajectory under each preset evaluation index;

[0220] The weighted calculation subunit multiplies the single-index score of each preset evaluation index with the weight corresponding to the preset evaluation index and then sums them to obtain the multi-index weight score of the driving trajectory; the multi-index weight score is divided by the sum of the weight values ​​of various preset evaluation indicators to obtain the safety score of the driving trajectory.

[0221] Optionally, the scoring calculation subunit includes:

[0222] The no-fault collision assessment index calculation subunit determines whether there is an overlap between the vehicle's position coordinates and the obstacle's position coordinates at the same moment in the driving trajectory. If so, it determines the collision type based on the overlapping obstacle and evaluates the trajectory segment at the same moment based on the collision type to obtain the evaluation score of the trajectory segment. It then calculates the evaluation score of each trajectory segment in the driving trajectory and takes the lowest evaluation score as the no-fault collision assessment score of the driving trajectory.

[0223] Optionally, the scoring calculation subunit includes:

[0224] The drivable area compliance assessment index calculation subunit calculates whether the vehicle has entered a non-drivable area in the driving trajectory. If the vehicle has entered a non-drivable area, the drivable area compliance assessment score of the driving trajectory is determined to be the lower limit score. If the vehicle has not entered a non-drivable area, the drivable area compliance assessment score of the driving trajectory is determined to be the upper limit score.

[0225] Optionally, the scoring calculation subunit includes:

[0226] The collision time assessment index calculation subunit, for the collision time assessment index, obtains the vehicle state of the driving trajectory at each future moment; wherein, the vehicle state includes the vehicle's position coordinates, speed, and orientation angle; calculates the future position coordinates of the vehicle after maintaining the speed and orientation angle for a preset time at each moment; determines whether there is overlap between the future position coordinates and the position coordinates of obstacles at the same moment; if so, determines the collision type based on the overlapping obstacles, and evaluates the trajectory segment at the same moment based on the collision type to obtain the evaluation score of the trajectory segment; and counts the evaluation scores of each trajectory segment in the driving trajectory, taking the lowest evaluation score as the collision time assessment score of the driving trajectory.

[0227] Optionally, the evaluation of trajectory segments at the same moment based on the collision type to obtain an evaluation score for the trajectory segment includes at least one of the following:

[0228] If the object of collision is a traffic light and the collision type is a red light, then the evaluation score for that trajectory segment is determined as the lower limit score;

[0229] If the object of collision is not a traffic light, and the collision type is a dynamic forward collision, a static obstacle collision, or a lateral collision, then the evaluation score for that trajectory segment is determined as the lower limit score.

[0230] If the object of collision is not a traffic light, and the collision type is not a dynamic forward collision, a static obstacle collision, or a lateral collision, then the evaluation score for that trajectory segment is determined to be the median score.

[0231] Optionally, the scoring calculation subunit includes:

[0232] The speed limit compliance assessment index calculation subunit, for the speed limit compliance assessment index, obtains the speed limit speed at each location of the vehicle based on the vehicle's position coordinates in the driving trajectory; subtracts the speed limit speed at each location from the vehicle's speed corresponding to the vehicle's position coordinate to obtain the speeding difference value of the vehicle at each position coordinate; calculates the speeding loss score based on the speeding difference value of each position coordinate; and subtracts the speeding loss score from the initial score of the driving trajectory to obtain the speed limit compliance assessment score of the driving trajectory.

[0233] Optionally, the scoring calculation subunit includes:

[0234] The trip progress evaluation index calculation subunit, for the trip progress evaluation index, sets the current position of the vehicle in the driving trajectory as the starting point; connects the starting point and the vehicle's destination with a reference line; maps the point on the reference line to the starting point as the process start point, and maps the last position point of the driving trajectory to the point on the reference line as the process end point; calculates the trajectory progress between the process end point and the process start point; if the trajectory progress is greater than the process threshold, divides the trajectory progress by the target progress to obtain the trip progress evaluation score of the driving trajectory; wherein, the target progress is the largest trajectory progress among the at least two driving trajectories.

[0235] Optionally, the scoring calculation subunit includes:

[0236] The comfort evaluation index calculation subunit acquires the acceleration and angular velocity states of the vehicle at different times along the driving trajectory for the comfort evaluation index; determines whether the acceleration and angular velocity states are within a preset range; if both the acceleration and angular velocity states are within the preset range, the comfort evaluation score of the driving trajectory is set to the upper limit score; if the acceleration and angular velocity states are not both within the preset range, the comfort evaluation score of the driving trajectory is set to the middle score.

[0237] Optionally, the scoring calculation subunit includes:

[0238] The driving direction compliance assessment calculation subunit, for driving direction compliance assessment, obtains the drivable area based on the vehicle's position in the driving trajectory, calculates the cumulative length of the driving trajectory located in the non-drivable area, and calculates the driving direction compliance score of the driving trajectory based on the cumulative length.

[0239] For details on the implementation process of the functions and roles of each module in the above-mentioned device, please refer to the implementation process of the corresponding steps in the above-mentioned post-processing access method for driving trajectory. For relevant parts, please refer to the description of the method implementation method. It will not be repeated here.

[0240] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the units or modules can be selected to achieve the purpose of the solution described in this specification, depending on actual needs. Those skilled in the art can understand and implement this without any inventive effort.

[0241] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer, which can take the form of a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email sending and receiving device, game console, tablet computer, wearable device, or any combination of these devices.

[0242] In a typical configuration, a computer includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0243] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0244] Computer-readable media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage, quantum memory, graphene-based storage media or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0245] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0246] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0247] The terminology used in one or more embodiments of this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of one or more embodiments of this specification. The singular forms “a,” “described,” and “the” used in one or more embodiments of this specification and in the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0248] It should be understood that although the terms first, second, third, etc., may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first information may also be referred to as second information without departing from the scope of one or more embodiments of this specification, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "in response to a determination," or "when," or "in the event of a determination."

[0249] The above description is merely a preferred embodiment of one or more embodiments of this specification and is not intended to limit the scope of one or more embodiments of this specification. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments of this specification should be included within the protection scope of one or more embodiments of this specification.

Claims

1. A post-processing access method for driving trajectory, applied to a server corresponding to a vehicle, wherein the server provides a service interface for the vehicle to call, the method comprising: Obtain at least two driving trajectories output by the end-to-end multimodal model mounted on the vehicle, which are sent by the vehicle through a service interface call; For each driving trajectory, a safety score is evaluated based on preset evaluation indicators; wherein, the preset evaluation indicators include evaluation indicators related to driving safety. Based on the safety score, a target driving trajectory is selected from the at least two driving trajectories; wherein the target driving trajectory is the driving trajectory with the highest safety. The target driving trajectory is returned to the vehicle.

2. The method according to claim 1, wherein the preset evaluation index includes at least one of the following: No-fault collision assessment index, drivable area compliance assessment index, collision time assessment index, speed limit compliance assessment index, trip progress assessment index, comfort assessment index, and driving direction compliance assessment index.

3. The method according to claim 2, wherein when there is more than one preset evaluation index, the step of evaluating the safety score of the driving trajectory based on the preset evaluation index includes: Calculate the single-index score of the driving trajectory under each preset evaluation index; The single-index score of each preset evaluation indicator is multiplied by the weight corresponding to the preset evaluation indicator and then summed to obtain the multi-index weight score of the driving trajectory. The safety score of the driving trajectory is obtained by dividing the multi-indicator weight score by the sum of the weight values ​​of various preset evaluation indicators.

4. The method according to claim 3, wherein calculating the single-index score of the driving trajectory under each preset evaluation index includes: For no-fault collision assessment indicators, determine whether there is any overlap between the vehicle's position coordinates and the obstacle's position coordinates at the same moment in the driving trajectory; If so, determine the collision type based on the overlapping obstacles, and evaluate the trajectory segment at the same moment based on the collision type to obtain the evaluation score of the trajectory segment; The evaluation score of each segment in the driving trajectory is calculated, and the lowest evaluation score is taken as the no-fault collision evaluation score of the driving trajectory.

5. The method according to claim 3, wherein calculating the single-index score of the driving trajectory under each preset evaluation index includes: For the drivable area compliance assessment index, determine whether the vehicle has entered a non-drivable area in the driving trajectory; If a vehicle enters a non-drivable area, the minimum score for compliance with the drivable area assessment of the driving trajectory is determined. If the vehicle does not enter the drivable area, the score for compliance with the drivable area assessment of the driving trajectory is the maximum score.

6. The method according to claim 3, wherein calculating the single-index score of the driving trajectory under each preset evaluation index includes: For the collision time assessment index, the vehicle state at each future moment is obtained based on the driving trajectory; wherein, the vehicle state includes the vehicle's position coordinates, speed, and orientation angle; Calculate the future position coordinates of the vehicle at each moment after maintaining the stated speed and orientation angle for a preset duration; Determine whether there is any overlap between the future position coordinates and the position coordinates of the obstacle at the same moment; If so, determine the collision type based on the overlapping obstacles, and evaluate the trajectory segment at the same moment based on the collision type to obtain the evaluation score of the trajectory segment; The evaluation score of each segment in the driving trajectory is calculated, and the lowest evaluation score is taken as the collision time evaluation score of the driving trajectory.

7. The method according to claim 4 or 6, wherein evaluating the trajectory segment at the same moment based on the collision type to obtain an evaluation score for the trajectory segment includes at least one of the following: If the object of collision is a traffic light and the collision type is a red light, then the evaluation score for that trajectory segment is determined as the lower limit score; If the object of collision is not a traffic light, and the collision type is a dynamic forward collision, a static obstacle collision, or a lateral collision, then the evaluation score for that trajectory segment is determined as the lower limit score. If the object of collision is not a traffic light, and the collision type is not a dynamic forward collision, a static obstacle collision, or a lateral collision, then the evaluation score for that trajectory segment is determined to be the median score.

8. The method according to claim 3, wherein calculating the single-index score of the driving trajectory under each preset evaluation index includes: For the speed limit compliance assessment index, the speed limit of each vehicle's location is obtained based on the vehicle's position coordinates in the driving trajectory; Subtract the speed limit at the current location from the vehicle's speed corresponding to its position coordinates to obtain the overspeed difference of the vehicle at each position coordinate. Calculate the overspeed loss score based on the overspeed difference for each location coordinate; The speed limit compliance assessment score of the driving trajectory is obtained by subtracting the speeding loss score from the initial score of the driving trajectory.

9. The method according to claim 3, wherein calculating the single-index score of the driving trajectory under each preset evaluation index includes: For the trip progress evaluation indicators, the current position of the vehicle in the driving trajectory is set as the starting point; A reference line connecting the starting point and the vehicle's destination; The point on the reference line mapped from the starting point is taken as the starting point of the process, and the point on the reference line mapped from the last position point of the driving trajectory is taken as the ending point of the process. The trajectory process between the ending point and the starting point is calculated. If the trajectory progress is greater than the progress threshold, the trajectory progress is divided by the target progress to obtain the trip progress evaluation score of the driving trajectory; wherein, the target progress is the largest trajectory progress among the at least two driving trajectories.

10. The method according to claim 3, wherein calculating the single-index score of the driving trajectory under each preset evaluation index includes: To assess comfort, the acceleration and angular velocity of the vehicle at different times along the driving trajectory are obtained. Determine whether the acceleration and angular velocity states are within a preset range; If both the acceleration and angular velocity states are within a preset range, the comfort evaluation score of the driving trajectory is set to the upper limit score. If the acceleration and angular velocity states are not within the preset range, the comfort evaluation score of the driving trajectory will be set to the intermediate score.

11. The method according to claim 3, wherein calculating the single-index score of the driving trajectory under each preset evaluation index includes: For the driving direction compliance assessment, the drivable area is obtained based on the vehicle's position in the driving trajectory, and the cumulative length of the driving trajectory located in the non-drivable area is calculated. The driving direction compliance score of the driving trajectory is calculated based on the cumulative length.

12. A post-processing access device for driving trajectory, applied to a server corresponding to a vehicle, the server providing a service interface for the vehicle to call, the device comprising: The acquisition unit acquires at least two driving trajectories output by the end-to-end multimodal model mounted on the vehicle terminal, which are sent by the vehicle terminal through a service interface call. The evaluation unit assesses the safety score of each driving trajectory based on preset evaluation indicators; wherein, the preset evaluation indicators include evaluation indicators related to driving safety. The processing unit selects a target driving trajectory from the at least two driving trajectories based on the safety score; wherein the target driving trajectory is the driving trajectory with the highest safety. The return unit then sends the target driving trajectory back to the vehicle.

13. An electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor implements the method as described in any one of claims 1-11 by executing the executable instructions.

14. A machine-readable storage medium having stored thereon machine-readable instructions that, when executed by a processor, implement the method as described in any one of claims 1-11.

Images