Trajectory processing method of traffic reference object and electronic device

Abstract

The application discloses a traffic reference object trajectory processing method and electronic equipment, and the method comprises the following steps: acquiring a plurality of prediction trajectory data of a traffic reference object, wherein the prediction trajectory data is used for representing the trajectory of the traffic reference object in a future period; performing smoothing processing on the plurality of prediction trajectory data to obtain a plurality of smoothed trajectory data; performing motion analysis on the plurality of smoothed trajectory data to obtain a plurality of motion analysis results; performing intention analysis on the plurality of smoothed trajectory data to obtain a plurality of intention analysis results; and determining target trajectory data from the plurality of smoothed trajectory data based on the motion state and the driving intention. The application solves the technical problem of low prediction accuracy of the trajectory data of the traffic object.

Description

Technical Field

[0001] This application relates to the field of vehicles, and more specifically, to a method and electronic device for processing the trajectory of a vehicle's traffic reference object. Background Technology

[0002] Currently, when predicting the trajectory of traffic reference objects, the method of predicting the trajectory of the traffic reference objects based on preset rules is often adopted. However, the trajectory prediction method based on preset rules requires manual design, has poor flexibility, and is difficult to handle complex driving scenarios.

[0003] Furthermore, when predicting the trajectory of traffic reference objects, an end-to-end large model is often used to predict and output the trajectory of the aforementioned traffic reference objects. However, the trajectory obtained by the trajectory prediction method based on the end-to-end large model may violate kinematic constraints, resulting in the technical problem of low prediction accuracy of the trajectory data of traffic objects.

[0004] There is currently no effective solution to the aforementioned technical problems. Summary of the Invention

[0005] This invention provides a trajectory processing method and electronic device for traffic reference objects, to at least solve the technical problem of low prediction accuracy of trajectory data for traffic objects.

[0006] According to one aspect of the present invention, a trajectory processing method for a traffic reference object is provided. The method includes: acquiring multiple predicted trajectory data of the traffic reference object, wherein the predicted trajectory data is used to represent the trajectory of the traffic reference object in a future time period; smoothing the multiple predicted trajectory data to obtain multiple smoothed trajectory data; performing motion analysis on the multiple smoothed trajectory data to obtain multiple motion analysis results, wherein the motion analysis results are used to represent the motion state of the traffic reference object corresponding to each smoothed trajectory data in a future time period; performing intent analysis on the multiple smoothed trajectory data to obtain multiple intent analysis results, wherein the intent analysis results are used to represent the driving intent of the traffic reference object corresponding to each smoothed trajectory data in a future time period; and determining target trajectory data from the multiple smoothed trajectory data based on the motion state and driving intent, wherein the target trajectory data is the smoothed trajectory data in a normal state among the multiple smoothed trajectory data.

[0007] Since this invention is based on motion state and driving intention, it selects smooth trajectory data in a normal state from multiple smooth trajectory data as target trajectory data, thereby achieving the purpose of optimizing the predicted trajectory data of traffic objects and thus realizing the technical effect of improving the prediction accuracy of the trajectory data of traffic objects.

[0008] Optionally, the motion state includes at least one of velocity, acceleration, heading angle, and angular velocity; the driving intention includes longitudinal driving intention and / or lateral driving intention.

[0009] Since the aforementioned motion states include different types of states, and the aforementioned driving intentions include longitudinal driving intentions and / or lateral driving intentions, the accuracy of the motion states and the accuracy of the driving intentions are guaranteed, thereby achieving the technical effect of improving the efficiency of determining target trajectory data.

[0010] Optionally, the lateral driving intention includes at least one of U-turn, left turn, right turn, straight ahead, left lane change, and right lane change.

[0011] Since lateral driving intentions can include a single type of lateral driving intention or different types of lateral driving intentions, the accuracy of lateral driving intentions can be guaranteed, thereby achieving the technical effect of improving the efficiency of determining target trajectory data.

[0012] Optionally, the method further includes: comparing the current heading angle of the traffic reference object with the terminal heading angle of the corresponding trajectory of the smooth trajectory data to obtain a comparison result, wherein the comparison result is used to represent the deviation between the current heading angle and the terminal heading angle; and determining the lateral driving intention based on the comparison result.

[0013] Since the aforementioned lateral driving intention is determined based on the deviation between the current heading angle and the final heading angle, the accuracy of the lateral driving intention is guaranteed, thereby achieving the technical effect of improving the efficiency of determining target trajectory data.

[0014] Optionally, based on the comparison results, the lateral driving intention is determined, including: when the comparison results indicate that the final heading angle is the same as the current heading angle, the lateral driving intention is determined to be to go straight; when the comparison results indicate that the final heading angle is different from the current heading angle, the lateral driving intention is determined to be to make a U-turn, or to turn left, or to turn right.

[0015] Since different lateral driving intentions can be determined based on different comparison results, the accuracy of lateral driving intentions can be guaranteed, thereby achieving the technical effect of improving the efficiency of determining target trajectory data.

[0016] Optionally, when the lateral driving intention is to go straight, the method further includes: determining the end lane of the trajectory corresponding to the smooth trajectory data, and the predicted end heading angle of the traffic reference object at the end of the trajectory, wherein the end lane is used to represent the lane where the end trajectory point is located in the trajectory, and the end heading angle is used to represent the heading angle of the traffic reference object at the end trajectory point; determining the lane position relationship between the end lane and the current lane where the traffic reference object is located; and determining the angle difference between the end heading angle and the current heading angle of the traffic reference object; and determining the lateral driving intention based on the lane position relationship and the angle difference.

[0017] Since the lateral driving intention is determined based on the lane position relationship between the end lane and the current lane, as well as the angle difference between the end heading angle and the current heading angle, the accuracy of the lateral driving intention can be guaranteed, thereby achieving the technical effect of improving the efficiency of determining target trajectory data.

[0018] Optionally, when the traffic reference object is static, the method further includes: determining the heading angle of the traffic reference object on the non-U-turn lane and the first angle between the heading angle and the lane direction of the non-U-turn lane; filtering smooth trajectory data corresponding to the first angle that is greater than a first angle threshold from multiple smooth trajectory data corresponding to the traffic reference object to obtain target trajectory data; or, determining multiple starting trajectory data from multiple smooth trajectory data, wherein the starting trajectory data represents the smooth trajectory data corresponding to the traffic reference object in the starting state in the smooth trajectory data; determining the probability that the multiple starting trajectory data occur in a real traffic scenario; filtering starting trajectory data corresponding to probabilities less than a probability threshold from the multiple smooth trajectory data corresponding to the traffic reference object to obtain target trajectory data; or... Determine the average acceleration corresponding to multiple starting trajectory data; filter out the starting trajectory data corresponding to the average acceleration greater than the average acceleration threshold from multiple smooth trajectory data corresponding to the traffic reference object to obtain the target trajectory data; or, determine the driving intention of multiple starting trajectory data; filter out the starting trajectory data corresponding to the driving intention of U-turn from multiple smooth trajectory data corresponding to the traffic reference object to obtain the target trajectory data; or, determine the starting scenario data corresponding to multiple starting trajectory data; if there is a starting scenario data that is the same as the target scenario data among the multiple starting scenario data, filter out the starting trajectory data corresponding to the target scenario data from the multiple smooth trajectory data to obtain the target trajectory data, wherein the target scenario data is a scenario where the traffic light is red within a preset distance.

[0019] When the traffic reference object is static, the target trajectory data can be obtained by filtering the smooth trajectory data that needs to be filtered from multiple smooth trajectory data according to the trajectory evaluation scheme suitable for static types. This achieves the goal of determining the target trajectory data from multiple smooth trajectory data, thereby realizing the technical effect of improving the prediction accuracy of the trajectory data of traffic objects.

[0020] Optionally, when the traffic reference object is dynamic, the method further includes: determining a second included angle between the current heading angle corresponding to the smooth trajectory data and the lane direction of the current lane, and filtering smooth trajectory data corresponding to the second included angle that is greater than a second included angle threshold from multiple smooth trajectory data to obtain target trajectory data; or, determining the speed difference between the speed represented by the motion analysis result and the current speed of the traffic reference object, and filtering smooth trajectory data corresponding to the speed difference that is greater than a speed difference threshold from multiple smooth trajectory data to obtain target trajectory data; or, determining the trajectory length of multiple smooth trajectory data, and filtering smooth trajectory data corresponding to the lateral driving intention being a U-turn intention, the speed being greater than a first speed threshold, and the trajectory length being less than a length threshold from multiple smooth trajectory data to obtain target trajectory data; or, filtering smooth trajectory data with a speed less than a second speed threshold from multiple smooth trajectory data to obtain target trajectory data, wherein the second speed threshold is different from the first speed threshold; or, determining the obstacle distance corresponding to multiple smooth trajectory data, and filtering smooth trajectory data with a trajectory length greater than the obstacle distance from multiple smooth trajectory data to obtain target trajectory data, wherein the obstacle distance is the collision distance between the traffic reference object and the obstacle.

[0021] Since the traffic reference object is dynamic, the target trajectory data can be obtained by filtering the smooth trajectory data that needs to be filtered from multiple smooth trajectory data according to the trajectory evaluation scheme suitable for the dynamic type. This achieves the purpose of determining the target trajectory data from multiple smooth trajectory data, thereby realizing the technical effect of improving the prediction accuracy of the trajectory data of the traffic object.

[0022] Optionally, multiple predicted trajectory data are smoothed separately to obtain multiple smoothed trajectory data, including: smoothing multiple predicted trajectory data separately to obtain multiple initial smoothed trajectory data; resampling multiple initial smoothed trajectory data to obtain multiple sampling points for each of the multiple initial smoothed trajectory data; and generating multiple smoothed trajectory data based on the multiple sampling points of each initial smoothed trajectory data.

[0023] Since multiple smooth trajectory data are generated from multiple sampling points of each of the multiple initial smooth trajectory data, the smooth trajectory data can be recovered, thereby achieving the technical effect of improving the accuracy of the smooth trajectory data.

[0024] According to one aspect of the present invention, a trajectory processing apparatus for a traffic reference object is provided. The apparatus may include: an acquisition unit for acquiring multiple predicted trajectory data of the traffic reference object, wherein the predicted trajectory data represents the trajectory of the traffic reference object in a future time period; a processing unit for smoothing the multiple predicted trajectory data to obtain multiple smoothed trajectory data; a motion analysis unit for performing motion analysis on the multiple smoothed trajectory data to obtain multiple motion analysis results, wherein the motion analysis results represent the motion state of the traffic reference object corresponding to each smoothed trajectory data in a future time period; an intent analysis unit for performing intent analysis on the multiple smoothed trajectory data respectively to obtain multiple intent analysis results, wherein the intent analysis results represent the driving intent of the traffic reference object corresponding to each smoothed trajectory data in a future time period; and a determination unit for determining target trajectory data from the multiple smoothed trajectory data based on the motion state and driving intent, wherein the target trajectory data is the smoothed trajectory data in a normal state among the multiple smoothed trajectory data.

[0025] According to another aspect of the present invention, a processor is also provided. The processor is used to run a program, wherein the program, when run by the processor, performs the methods described in the embodiments of the present invention.

[0026] According to another aspect of the embodiments of the present invention, an electronic device is also provided, comprising: a memory storing a computer program; and a processor for running the computer program, wherein the computer program executes the methods of various embodiments of the present invention when it is run.

[0027] According to another aspect of the present invention, a computer-readable storage medium is also provided. The computer-readable storage medium includes a stored executable program, wherein, when the executable program is executed, it controls the device containing the computer-readable storage medium to perform the methods described in the embodiments of the present invention.

[0028] According to another aspect of the present invention, a computer program product is also provided, the computer program product comprising a computer program, wherein the computer program, when executed by a processor, implements the method of the present invention.

[0029] According to another aspect of the present invention, a computer program product is also provided, including a non-volatile computer-readable storage medium for storing a computer program, which, when executed by a processor, implements the method in the embodiments of the present invention.

[0030] According to another aspect of the embodiments of the present invention, the embodiments of this application also provide a computer program that, when executed by a processor, implements the methods described in the embodiments of the present invention.

[0031] In this embodiment of the invention, when processing the trajectory of a traffic reference object, multiple predicted trajectory data of the traffic reference object can be acquired; the multiple predicted trajectory data can be smoothed to obtain multiple smoothed trajectory data; motion analysis can be performed on the multiple smoothed trajectory data to obtain multiple motion analysis results; intent analysis can be performed on the multiple smoothed trajectory data to obtain multiple intent analysis results; and based on the motion state and driving intent, target trajectory data can be determined from the multiple smoothed trajectory data. Because this embodiment of the invention, after smoothing the acquired multiple predicted trajectory data to obtain multiple smoothed trajectory data, performs motion analysis on the aforementioned multiple smoothed trajectory data to obtain multiple motion analysis results, and performs intent analysis on the aforementioned multiple smoothed trajectory data to obtain multiple intent analysis results, and then, based on the motion state and driving intent, selects the smoothed trajectory data in a normal state from the multiple predicted trajectory data as the target trajectory data, thereby achieving the goal of optimizing the predicted trajectory data of the traffic object, thus solving the technical problem of low prediction accuracy of the trajectory data of the traffic object, and ultimately achieving the technical effect of improving the prediction accuracy of the trajectory data of the traffic object. Attached Figure Description

[0032] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0033] Figure 1 This is a flowchart of a trajectory processing method for a traffic reference object according to an embodiment of the present invention;

[0034] Figure 2 This is a flowchart of a post-processing method for predicting trajectories using a large model, according to an embodiment of the present invention.

[0035] Figure 3 This is a schematic diagram of a trajectory processing device for a traffic reference object according to an embodiment of the present invention;

[0036] Figure 4 This is a schematic diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0037] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0038] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention 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 the invention 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 a 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.

[0039] According to an embodiment of the present invention, a trajectory processing method for a traffic reference object 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.

[0040] Figure 1 This is a flowchart of a trajectory processing method for a traffic reference object according to an embodiment of the present invention, such as... Figure 1 As shown, the method may include the following steps:

[0041] Step S101: Obtain multiple predicted trajectory data of the traffic reference object, wherein the predicted trajectory data is used to represent the trajectory of the traffic reference object in the future time period.

[0042] In the technical solution provided by step S101 of the present invention, the predicted trajectory data can be used to represent the trajectory of a traffic reference object in a future time period. The future time period may include: a future moment and the moment preceding that future moment.

[0043] In this embodiment, the aforementioned traffic reference object can also be referred to as a traffic participant. For example, a traffic participant can be any other vehicle in any traffic scenario besides the vehicle itself (also referred to as the vehicle itself). The traffic scenario (hereinafter referred to as the scenario) can include at least one of the following: racetrack scenario, driving school scenario, highway scenario, urban street scenario, and rural national highway scenario, etc. This is only an example and is not specifically limited.

[0044] In this embodiment, multiple predicted trajectory data of the traffic reference object are obtained. Optionally, this embodiment can obtain multiple predicted trajectory data of the traffic reference object from a trajectory prediction model, wherein the trajectory prediction model can be constructed based on a large model, for example, the large model can be, but is not limited to, an end-to-end large model.

[0045] Optionally, multiple predicted trajectory data of the traffic reference object can be obtained from the trajectory prediction model. For example, multiple predicted trajectory data of the traffic reference object can be obtained from the output layer of the trajectory prediction model, thereby achieving the goal of obtaining multiple predicted trajectory data of the traffic reference object.

[0046] Step S102: Smooth the multiple predicted trajectory data to obtain multiple smoothed trajectory data.

[0047] In the technical solution provided by step S102 of the present invention, the smoothness of the smoothed trajectory data is higher than the smoothness of the corresponding predicted trajectory data. That is, the smoothed trajectory data can be used to represent the smoothed predicted trajectory.

[0048] In this embodiment, after acquiring multiple predicted trajectory data of the traffic reference object, the multiple predicted trajectory data are fitted to obtain multiple smooth trajectory data. Optionally, based on the acquired multiple predicted trajectory data, this embodiment uses a fifth-order polynomial least squares method to smooth the multiple predicted trajectory data, which can obtain multiple initial smooth trajectory data. Based on the multiple initial smooth trajectory data, multiple smooth trajectory data can be generated.

[0049] Optionally, from the aforementioned multiple predicted trajectory data, a predetermined number of predicted trajectory points are extracted at the midpoint of each predicted trajectory data set. Then, in each predicted trajectory data set, a polynomial is used to fit the predicted trajectory formed by the extracted predetermined number of predicted trajectory points, yielding the corresponding fifth-order polynomial coefficients. The polynomial can be two sets of polynomials (tx, ty), where t can represent a time function with respect to x and y, x can represent the abscissa of the predicted trajectory point, and y can represent the ordinate of the predicted trajectory point. Then, according to a predetermined sampling interval, the predicted trajectory formed by the predetermined number of predicted trajectory points is resampled using the fifth-order polynomial with the fifth-order polynomial coefficients substituted, respectively, to obtain smoothed trajectory data corresponding to the aforementioned predicted trajectories. This achieves the goal of obtaining multiple smoothed trajectory data sets.

[0050] It should be noted that the above-mentioned preset quantity can be set according to the traffic reference object, for example, the above-mentioned preset quantity can be 7 or 8, etc.; the above-mentioned preset sampling interval can also be set according to the traffic reference object, for example, the above-mentioned preset sampling interval can be 0.01 seconds (s) or 0.015 seconds (s), etc. The values ​​here are only for illustrative purposes and are not specifically limited.

[0051] Step S103: Perform motion analysis on multiple smooth trajectory data to obtain multiple motion analysis results.

[0052] In the technical solution provided by step S103 of the present invention, the motion analysis results can be used to represent the motion state of the traffic reference object in a future time period. The motion analysis can be a general kinematic analysis, and can include: velocity analysis, acceleration analysis, heading analysis, and angular velocity analysis.

[0053] In this embodiment, after smoothing multiple predicted trajectory data to obtain multiple smooth trajectory data, motion analysis is performed on the multiple smooth trajectory data to obtain multiple motion analysis results. Optionally, based on obtaining multiple smooth trajectory data, this embodiment inputs the multiple smooth trajectory data into a motion analysis model for motion analysis to obtain multiple motion analysis results. The aforementioned motion analysis model may include: a velocity analysis model, an acceleration analysis model, a heading analysis model, and an angular velocity analysis model.

[0054] Optionally, multiple smooth trajectory data can be input into a motion analysis model for motion analysis, resulting in multiple motion analysis results. For example, inputting multiple smooth trajectory data into a velocity analysis model for velocity analysis, inputting multiple smooth trajectory data into an acceleration analysis model for acceleration analysis, inputting multiple fitted trajectory data into a heading analysis model for heading analysis, and inputting multiple fitted trajectory data into an angular velocity analysis model for angular velocity analysis can all yield multiple motion analysis results.

[0055] Step S104: Perform intent parsing on multiple smooth trajectory data to obtain multiple intent parsing results.

[0056] In the technical solution provided by step S104 of the present invention, the intent parsing result can be used to represent the driving intent of the traffic reference object in a future time period. The intent parsing can be scene intent parsing, and can include: longitudinal intent parsing and lateral intent parsing.

[0057] In this embodiment, intent parsing is performed on multiple smooth trajectory data to obtain multiple intent parsing results. Optionally, based on obtaining multiple smooth trajectory data, this embodiment inputs the multiple smooth trajectory data into an intent parsing model for intent parsing, which can yield multiple intent parsing results. The intent parsing model may include a vertical intent parsing model and a horizontal intent parsing model.

[0058] Optionally, inputting multiple smooth trajectory data into the intent parsing model for intent parsing can yield multiple intent parsing results. For example, inputting multiple smooth trajectory data into the vertical intent parsing model for intent parsing, and inputting multiple smooth trajectory data into the horizontal intent parsing model for intent parsing, can yield multiple intent parsing results.

[0059] Step S105: Based on the motion state and driving intention, determine the target trajectory data from multiple smooth trajectory data.

[0060] In the technical solution provided by step S105 of the present invention, the above-mentioned scene data can be used to represent the scene in which the traffic reference object is located in a future time period.

[0061] In this embodiment, the target trajectory data can be the trajectory data in a normal state among multiple smooth trajectory data. That is, the target trajectory data can be reasonable trajectory data.

[0062] In this embodiment, after performing motion analysis on multiple smooth trajectory data to obtain multiple motion analysis results, and performing intent analysis on the multiple smooth trajectory data to obtain multiple intent analysis results, the target trajectory data is determined from the multiple smooth trajectory data based on the motion state and driving intent. Optionally, based on obtaining multiple motion analysis results and multiple intent analysis results, this embodiment evaluates the multiple smooth trajectory data using the motion state corresponding to the motion analysis results and the driving intent corresponding to the intent analysis results, which can be used to indicate whether there is trajectory data in an abnormal state among the multiple smooth trajectory data. Based on the above multiple evaluation results, the target trajectory data can be determined, thereby achieving the purpose of optimizing the predicted trajectory data of traffic objects.

[0063] Optionally, the target trajectory data can be determined based on the obtained multiple evaluation results. For example, if the multiple evaluation results include an evaluation result indicating that the smooth trajectory data is in an abnormal state, the smooth trajectory data in an abnormal state can be deleted from the multiple smooth trajectory data to obtain the target trajectory data; if the multiple evaluation results do not include an evaluation result indicating that the smooth trajectory data is in an abnormal state, the multiple smooth trajectory data can be directly used as the target trajectory data.

[0064] In steps S101 to S105 of this application, when processing the trajectory of a traffic reference object, multiple predicted trajectory data of the traffic reference object can be obtained; the multiple predicted trajectory data can be smoothed to obtain multiple smoothed trajectory data; motion analysis can be performed on the multiple smoothed trajectory data to obtain multiple motion analysis results; intent analysis can be performed on the multiple smoothed trajectory data to obtain multiple intent analysis results; and based on the motion state and driving intent, target trajectory data can be determined from the multiple smoothed trajectory data. Since this embodiment of the invention, after smoothing the acquired multiple predicted trajectory data to obtain multiple smoothed trajectory data, performs motion analysis on the aforementioned multiple smoothed trajectory data to obtain multiple motion analysis results, and performs intent analysis on the aforementioned multiple smoothed trajectory data to obtain multiple intent analysis results, and then, based on the motion state and driving intent, the smoothed trajectory data in a normal state can be selected as the target trajectory data from the multiple predicted trajectory data, thereby achieving the goal of optimizing the predicted trajectory data of the traffic object, thus solving the technical problem of low prediction accuracy of the trajectory data of the traffic object, and ultimately achieving the technical effect of improving the prediction accuracy of the trajectory data of the traffic object.

[0065] The following is a further description of the motion state and driving intention of this embodiment.

[0066] As an optional embodiment, the motion state includes at least one of velocity, acceleration, heading angle, and angular velocity; the driving intention includes longitudinal driving intention and / or lateral driving intention.

[0067] In this embodiment, the aforementioned motion state may include at least one of velocity, acceleration, heading angle, and angular velocity.

[0068] In this embodiment, the driving intention may include longitudinal driving intention and / or lateral driving intention. The longitudinal driving intention may include: constant speed (CONSTANT), start (START_UP), speed up (SPEED_UP), slow down (SLOW_DOWN), pull up (PULL_UP), and stand still (STATIC).

[0069] The following section will further explain the lateral driving intent described above in this embodiment.

[0070] As an alternative embodiment, the lateral driving intent includes at least one of U-turn, left turn, right turn, straight ahead, left lane change, and right lane change.

[0071] In this embodiment, the aforementioned lateral driving intention may include at least one of the following: maintaining the current direction (KEEP_CURRENT), making a U-turn, turning left (TURN_LEFT), turning right (TURN_RIGHT), going straight (NO_TURN), changing lanes to the left (LEFT_CHANGE), and changing lanes to the right (RIGHT_CHANGE).

[0072] Since lateral driving intentions can include a single type of lateral driving intention or different types of lateral driving intentions, the accuracy of lateral driving intentions can be guaranteed, thereby achieving the technical effect of improving the efficiency of determining target trajectory data.

[0073] The following section further describes the above-described motion analysis of multiple smooth trajectory data to obtain multiple motion analysis results in this embodiment.

[0074] As an optional embodiment, step S103 involves performing motion analysis on multiple smooth trajectory data to obtain multiple motion analysis results, including: inputting the multiple smooth trajectory data into a velocity analysis model to perform velocity analysis, obtaining multiple velocities, wherein the velocity analysis model is constructed based on smooth trajectory data samples and velocity samples; inputting the multiple smooth trajectory data into an acceleration analysis model to perform acceleration analysis, obtaining multiple accelerations, wherein the acceleration analysis model is constructed based on smooth trajectory data samples and acceleration samples; inputting the multiple smooth trajectory data into a heading analysis model to perform heading analysis, obtaining multiple heading angles, wherein the heading analysis model is constructed based on smooth trajectory data samples and heading angle samples; inputting the multiple smooth trajectory data into an angular velocity analysis model to perform angular velocity analysis, obtaining multiple angular velocities, wherein the angular velocity analysis model is constructed based on smooth trajectory data samples and angular velocity samples; and determining the multiple velocities, multiple accelerations, multiple heading angles, and multiple angular velocities as multiple motion analysis results.

[0075] In this embodiment, the velocity analysis model described above can be constructed based on smooth trajectory data samples and velocity samples. For example, the velocity analysis model may include a general velocity parser (VelocityIntepreter).

[0076] In this embodiment, the aforementioned speed can be represented by a trajectory speed value.

[0077] In this embodiment, multiple smooth trajectory data are input into a velocity analysis model for velocity analysis to obtain multiple velocities. Optionally, in this embodiment, based on obtaining multiple smooth trajectory data, the multiple smooth trajectory data are input into a velocity analysis model, and then a general velocity parser is used to analyze the multiple smooth trajectory data to obtain multiple velocities.

[0078] In this embodiment, the acceleration analytical model described above can be constructed based on smooth trajectory data samples and acceleration samples. For example, the acceleration analytical model may include an acceleration interpreter.

[0079] In this embodiment, the acceleration described above can be represented by a trajectory acceleration value.

[0080] In this embodiment, multiple smooth trajectory data are input into an acceleration analysis model for acceleration analysis to obtain multiple accelerations. Optionally, in this embodiment, based on obtaining multiple smooth trajectory data, the multiple smooth trajectory data are input into an acceleration analysis model, and then an acceleration general analyzer is used to perform acceleration analysis on the multiple smooth trajectory data to obtain multiple accelerations.

[0081] In this embodiment, the heading analysis model described above can be constructed based on smooth trajectory data samples and heading angle samples. For example, the heading analysis model may include a general heading parser (YawInterpreter).

[0082] In this embodiment, the heading angle can be represented by the trajectory heading value.

[0083] In this embodiment, multiple smooth trajectory data are input into a heading analysis model for heading analysis to obtain multiple heading angles. Optionally, in this embodiment, based on obtaining multiple smooth trajectory data, multiple smooth trajectory data are input into a heading analysis model, and then a general heading analyzer is used to perform heading analysis on the multiple smooth trajectory data to obtain multiple heading angles.

[0084] In this embodiment, the aforementioned angular velocity analytical model can be constructed based on smooth trajectory data samples and angular velocity samples. For example, the aforementioned angular velocity analytical model may include: a general angular velocity interpreter (YawRateInterpreter).

[0085] In this embodiment, the above-mentioned angular velocity analysis results can be represented by trajectory angular velocity values.

[0086] In this embodiment, multiple smooth trajectory data are input into a heading analysis model for heading analysis, resulting in multiple heading analysis results. Optionally, based on the obtained multiple smooth trajectory data, this embodiment inputs the multiple smooth trajectory data into an angular velocity analysis model, and then uses a general angular velocity analyzer to analyze the multiple smooth trajectory data to obtain multiple angular velocities.

[0087] In this embodiment, after obtaining multiple velocities, multiple accelerations, multiple heading angles, and multiple angular velocities, these multiple velocities, multiple accelerations, multiple heading angles, and multiple angular velocities are determined as multiple motion analysis results. Optionally, in this embodiment, based on obtaining multiple velocities, multiple accelerations, multiple heading angles, and multiple angular velocities, these multiple velocities, multiple accelerations, multiple heading angles, and multiple angular velocities are collectively used as multiple motion analysis results.

[0088] By performing different types of motion analysis on multiple smooth trajectory data, multiple motion analysis results can be obtained, thereby achieving the goal of determining the motion state of traffic reference objects in future time periods, and thus realizing the technical effect of improving the accuracy of motion analysis of smooth trajectory data.

[0089] The method described below for parsing multiple smooth trajectory data to obtain multiple intent parsing results in this embodiment will be further described below.

[0090] As an optional embodiment, step S104 involves performing intent parsing on multiple smooth trajectory data to obtain multiple intent parsing results, including: inputting the multiple smooth trajectory data into a longitudinal intent parsing model for longitudinal intent parsing to obtain multiple longitudinal driving intentions, wherein the longitudinal intent parsing model is constructed based on smooth trajectory data samples and longitudinal driving intention samples; inputting the multiple smooth trajectory data into a lateral intent parsing model for lateral intent parsing to obtain multiple lateral driving intentions, wherein the lateral intent parsing model is constructed based on smooth trajectory data samples and lateral driving intention samples; and determining the multiple longitudinal driving intentions and the multiple lateral driving intentions as multiple intent parsing results.

[0091] In this embodiment, the longitudinal intent parsing model described above can be constructed based on smooth trajectory data samples and longitudinal driving intent samples. For example, the longitudinal intent parsing model may include a longitudinal intent parser (LonIntentInterpreter).

[0092] In this embodiment, multiple smooth trajectory data are input into the longitudinal intent parsing model for longitudinal intent parsing, resulting in multiple longitudinal intent parsing results. Optionally, in this embodiment, based on obtaining multiple smooth trajectory data, the multiple smooth trajectory data are input into the longitudinal intent parsing model, and then the longitudinal intent parser performs longitudinal intent parsing on the multiple smooth trajectory data respectively, resulting in multiple longitudinal driving intentions.

[0093] In this embodiment, the lateral intent parsing model described above can be constructed based on smooth trajectory data samples and lateral driving intent samples. For example, the lateral intent parsing model may include a lateral intent parser (LatIntentInterpreter).

[0094] In this embodiment, multiple smooth trajectory data are input into the lateral intent parsing model for lateral intent parsing, resulting in multiple lateral intent parsing results. Optionally, in this embodiment, based on obtaining multiple smooth trajectory data, the multiple smooth trajectory data are input into the lateral intent parsing model, and then the lateral intent parser performs lateral intent parsing on the multiple smooth trajectory data respectively, resulting in multiple lateral driving intentions.

[0095] In this embodiment, after obtaining multiple longitudinal driving intentions and multiple lateral driving intentions, the multiple longitudinal driving intentions and multiple lateral driving intentions are determined as multiple intention parsing results. Optionally, in this embodiment, based on obtaining multiple longitudinal driving intentions and multiple lateral driving intentions, the aforementioned multiple longitudinal driving intentions and multiple lateral driving intentions are collectively used as multiple intention parsing results.

[0096] By performing different types of intent analysis on multiple motion analysis results, multiple intent analysis results can be obtained. This achieves the goal of determining the driving intent of traffic reference objects in future time periods, thereby improving the technical effect of improving the accuracy of intent analysis of smooth trajectory data.

[0097] The trajectory processing method for the traffic reference object described in this embodiment will be further described below.

[0098] As an optional embodiment, the method further includes: comparing the current heading angle of the traffic reference object with the final heading angle of the corresponding trajectory of the smooth trajectory data to obtain a comparison result, wherein the comparison result is used to represent the deviation between the current heading angle and the final heading angle; and determining the lateral driving intention based on the comparison result.

[0099] In this embodiment, the comparison result can be used to represent the deviation between the current heading angle and the final heading angle. For example, the comparison result can be used to indicate that the current heading angle and the final heading angle are the same, or the comparison result can be used to indicate that the current heading angle and the final heading angle are different. This is only an example and is not a specific limitation.

[0100] In this embodiment, the current heading angle of the traffic reference object is compared with the final heading angle of the corresponding trajectory in the smooth trajectory data to obtain a comparison result. Optionally, after determining the current heading angle of other vehicles, this embodiment compares the current heading angle and the final heading angle of the other vehicles to obtain a comparison result.

[0101] In this embodiment, after comparing the current heading angle of the traffic reference object with the terminal heading angle of the corresponding trajectory in the smooth trajectory data, and obtaining the comparison result, the lateral driving intention is determined based on the comparison result. Optionally, this embodiment performs content parsing on the comparison result obtained from the comparison, which can determine whether the lateral driving intention is to go straight, turn left, or turn right.

[0102] Since the aforementioned lateral driving intention is determined based on the deviation between the current heading angle and the final heading angle, the accuracy of the lateral driving intention is guaranteed, thereby achieving the technical effect of improving the efficiency of determining target trajectory data.

[0103] The steps for determining lateral driving intent based on the comparison results described above in this embodiment will be further explained below.

[0104] As an optional embodiment, determining the lateral driving intention based on the comparison result includes: when the comparison result indicates that the final heading angle is the same as the current heading angle, determining the lateral driving intention as going straight; when the comparison result indicates that the final heading angle is different from the current heading angle, determining the lateral driving intention as proceeding, turning around, turning left, or turning right.

[0105] In this embodiment, after comparing the current heading angle of the traffic reference object with the terminal heading angle of the corresponding trajectory in the smooth trajectory data, and obtaining the comparison result, if the comparison result indicates that the terminal heading angle is the same as the current heading angle, the lateral driving intention is determined to be straight. Optionally, in this embodiment, if the parsed comparison result indicates that the terminal heading angle is the same as the current heading angle, the lateral driving intention can be determined to be straight.

[0106] In this embodiment, after comparing the current heading angle of the traffic reference object with the terminal heading angle of the corresponding trajectory in the smooth trajectory data, and obtaining the comparison result, if the comparison result indicates that the terminal heading angle is different from the current heading angle, the lateral driving intention is determined to be a U-turn, a left turn, or a right turn. Optionally, in this embodiment, when the parsed comparison result indicates that the terminal heading angle is different from the current heading angle, the lateral driving intention is determined to be a U-turn, a left turn, or a right turn.

[0107] Since different lateral driving intentions can be determined based on different comparison results, the accuracy of lateral driving intentions can be guaranteed, thereby achieving the technical effect of improving the efficiency of determining target trajectory data.

[0108] The trajectory processing method for the traffic reference object described in this embodiment will be further described below.

[0109] As an optional embodiment, when the lateral driving intention is to go straight, the method further includes: determining the end lane of the trajectory corresponding to the smooth trajectory data, and the predicted end heading angle of the traffic reference object at the end of the trajectory, wherein the end lane is used to represent the lane where the end trajectory point is located in the trajectory, and the end heading angle is used to represent the heading angle of the traffic reference object at the end trajectory point; determining the lane position relationship between the end lane and the current lane where the traffic reference object is located; and determining the angle difference between the end heading angle and the current heading angle of the traffic reference object; and determining the lateral driving intention based on the lane position relationship and the angle difference.

[0110] In this embodiment, the aforementioned end lane can be used to represent the lane where the end trajectory point is located in the trajectory. The aforementioned end trajectory point is the last trajectory point among all trajectory points included in the trajectory.

[0111] In this embodiment, the aforementioned terminal heading angle can be used to represent the heading angle of the traffic reference object at the terminal trajectory point. For example, the aforementioned terminal heading angle can be used to represent the heading angle of other vehicles at the final trajectory point.

[0112] In this embodiment, the final lane of the smoothed trajectory data is determined, along with the predicted final heading angle of the traffic reference object at the end of the trajectory. Optionally, this embodiment determines the final trajectory point from the smoothed trajectory data, performs lane matching on the final trajectory point to obtain the final lane of the trajectory corresponding to the smoothed trajectory data, and identifies the heading angle of the traffic reference object at the final trajectory point to obtain the final heading angle of the traffic reference object.

[0113] In this embodiment, the lane position relationship described above is the topological relationship between the end lane and the current lane.

[0114] In this embodiment, the aforementioned angle difference can be represented by Δyaw.

[0115] In this embodiment, after determining the end lane of the smooth trajectory data and the predicted end heading angle of the traffic reference object at the end of the trajectory, the lane position relationship between the end lane and the current lane where the traffic reference object is located is determined; and the angle difference between the end heading angle and the current heading angle of the traffic reference object is determined; based on the lane position relationship and the angle difference, the lateral driving intention is determined. Optionally, based on determining the end lane and the end heading angle, this embodiment performs relationship matching between the end lane and the current lane to obtain the topological relationship between the end lane and the current lane, and calculates the difference between the end heading angle and the current heading angle to obtain Δyaw between the end heading angle and the current heading angle. Then, based on the matched topological relationship and the calculated Δyaw, the lateral driving intention can be determined as KEEP_CURRENT, LEFT_CHANGE, or RIGHT_CHANGE.

[0116] Since the lateral driving intention is determined based on the lane position relationship between the end lane and the current lane, as well as the angle difference between the end heading angle and the current heading angle, the accuracy of the lateral driving intention can be guaranteed, thereby achieving the technical effect of improving the efficiency of determining target trajectory data.

[0117] The trajectory processing method for the traffic reference object described in this embodiment will be further described below.

[0118] As an optional embodiment, when the traffic reference object is static, the method further includes: determining the heading angle of the traffic reference object on the non-U-turn lane and a first angle between the heading angle and the lane direction of the non-U-turn lane; filtering smooth trajectory data corresponding to the first angle that is greater than a first angle threshold from multiple smooth trajectory data corresponding to the traffic reference object to obtain target trajectory data; or, determining multiple starting trajectory data from multiple smooth trajectory data, wherein the starting trajectory data represents the smooth trajectory data corresponding to the traffic reference object in a starting state in the smooth trajectory data; determining the probability that the multiple starting trajectory data occur in a real traffic scenario; filtering starting trajectory data corresponding to probabilities less than a probability threshold from the multiple smooth trajectory data corresponding to the traffic reference object to obtain target trajectory data. Alternatively, determine the average acceleration corresponding to multiple starting trajectory data; filter the starting trajectory data corresponding to the average acceleration greater than the average acceleration threshold from multiple smooth trajectory data corresponding to the traffic reference object to obtain the target trajectory data; or, determine the driving intention of multiple starting trajectory data; filter the starting trajectory data corresponding to the driving intention of making a U-turn from multiple smooth trajectory data corresponding to the traffic reference object to obtain the target trajectory data; or, determine the starting scenario data corresponding to multiple starting trajectory data; if there is starting scenario data that is the same as the target scenario data among the multiple starting scenario data, filter the starting trajectory data corresponding to the target scenario data from the multiple smooth trajectory data to obtain the target trajectory data, wherein the target scenario data is a scenario where the traffic light is red within a preset distance.

[0119] In this embodiment, the aforementioned static traffic reference object can be other vehicles that exhibit minimal displacement and velocity changes. That is, the static traffic reference object can be understood as other vehicles moving slowly or completely stationary in a traffic scenario. Therefore, the aforementioned static traffic reference object also has velocity and acceleration. For example, if it is a slowly moving other vehicle, its velocity and acceleration are simply small values; if it is a completely stationary other vehicle, its velocity and acceleration are zero. This is merely an example and not a specific limitation.

[0120] In this embodiment, the first included angle threshold can be, but is not limited to, π / 10.

[0121] In this embodiment, the heading angle of a traffic reference object on a non-U-turn lane is determined, and a first angle is formed between this heading angle and the lane direction of the non-U-turn lane. From multiple smooth trajectory data corresponding to the traffic reference object, smooth trajectory data corresponding to the first angle exceeding a first angle threshold are filtered out to obtain target trajectory data. Optionally, this embodiment calculates the angle between the heading angle and the lane direction of the non-U-turn lane to obtain the first angle between the heading angle and the lane direction of the UTURN lane. Then, from the multiple smooth trajectory data corresponding to the traffic reference object, smooth trajectory data corresponding to the first angle exceeding the first angle threshold is determined, and smooth trajectory data corresponding to the first angle exceeding the first angle threshold is deleted from the multiple smooth trajectory data corresponding to the traffic reference object to obtain target trajectory data.

[0122] In this embodiment, the aforementioned starting trajectory data can be used to represent the smooth trajectory data corresponding to the traffic reference object in the starting state within the smooth trajectory data. That is, the aforementioned starting trajectory data can be the starting trajectory within the smooth trajectory.

[0123] In this embodiment, multiple starting trajectory data are determined from multiple smooth trajectory data; the probability of these multiple starting trajectory data occurring in a real traffic scenario is determined; and starting trajectory data with probabilities less than a probability threshold are filtered from the multiple smooth trajectory data corresponding to a traffic reference object to obtain target trajectory data. Optionally, this embodiment segments the multiple smooth trajectory data, and multiple starting trajectory data can be determined from the segmented smooth trajectory data. Then, the multiple starting trajectory data are input into a probability prediction model for probability prediction to obtain the probability that the multiple starting trajectory data occur in a real traffic scenario. The probability prediction model is constructed based on starting trajectory data samples and probability samples. The target trajectory data is obtained by determining starting trajectory data with probabilities less than a probability threshold from the multiple smooth trajectory data corresponding to the traffic reference object, and by deleting starting trajectory data with probabilities less than a probability threshold from the multiple smooth trajectory data corresponding to the traffic reference object.

[0124] In this embodiment, the average acceleration threshold can be, but is not limited to, 3 m / s^2 or 3.1 m / s^2.

[0125] In this embodiment, the average acceleration corresponding to multiple starting trajectory data is determined; from multiple smooth trajectory data corresponding to the traffic reference object, starting trajectory data corresponding to average acceleration greater than an average acceleration threshold is filtered out to obtain target trajectory data. Optionally, after determining the average acceleration corresponding to multiple starting trajectory data, this embodiment further determines starting trajectory data corresponding to average acceleration greater than an average acceleration threshold from multiple smooth trajectory data corresponding to the traffic reference object, and deletes starting trajectory data corresponding to average acceleration greater than an average acceleration threshold from multiple smooth trajectory data corresponding to the traffic reference object to obtain target trajectory data.

[0126] In this embodiment, the driving intention of multiple starting trajectory data is determined; from multiple smooth trajectory data corresponding to the traffic reference object, the starting trajectory data corresponding to the driving intention of making a U-turn is filtered out to obtain the target trajectory data. Optionally, after determining the driving intention of multiple starting trajectory data, this embodiment further determines the starting trajectory data corresponding to the driving intention of making a U-turn from the multiple smooth trajectory data corresponding to the traffic reference object, and deletes the starting trajectory data corresponding to the driving intention of making a U-turn from the multiple smooth trajectory data corresponding to the traffic reference object to obtain the target trajectory data.

[0127] In this embodiment, the target scene data can be a scene where the traffic light is red within a preset distance. For example, the preset distance can be, but is not limited to, 30m or 30.5m, etc. The values ​​here are only illustrative and are not specifically limited.

[0128] In this embodiment, starting scenario data corresponding to multiple starting trajectory data are determined. If there is starting scenario data in the multiple starting scenario data that is the same as the target scenario data, the starting trajectory data corresponding to the target scenario data is filtered from the multiple smooth trajectory data to obtain the target trajectory data. Optionally, after determining the starting scenario data corresponding to the multiple starting trajectory data, this embodiment compares the multiple starting scenario data with the target scenario data. If the comparison reveals that there is starting scenario data in the multiple starting scenario data that is the same as the target scenario data, the starting trajectory data corresponding to the target scenario data is deleted from the multiple smooth trajectory data to obtain the target trajectory data.

[0129] When the traffic reference object is static, the target trajectory data can be obtained by filtering the smooth trajectory data that needs to be filtered from multiple smooth trajectory data according to the trajectory evaluation scheme suitable for static types. This achieves the goal of determining the target trajectory data from multiple smooth trajectory data, thereby realizing the technical effect of improving the prediction accuracy of the trajectory data of traffic objects.

[0130] The trajectory processing method for the traffic reference object described in this embodiment will be further described below.

[0131] As an optional embodiment, when the traffic reference object is dynamic, the method further includes: determining a second included angle between the current heading angle corresponding to the smooth trajectory data and the lane direction of the current lane, and filtering smooth trajectory data corresponding to the second included angle that is greater than a second included angle threshold from multiple smooth trajectory data to obtain target trajectory data; or, determining the speed difference between the speed represented by the motion analysis result and the current speed of the traffic reference object, and filtering smooth trajectory data corresponding to the speed difference that is greater than a speed difference threshold from multiple smooth trajectory data to obtain target trajectory data; or, determining the trajectory length of multiple smooth trajectory data, and filtering smooth trajectory data corresponding to the lateral driving intention being a U-turn intention, the speed being greater than a first speed threshold, and the trajectory length being less than a length threshold from multiple smooth trajectory data to obtain target trajectory data; or, filtering smooth trajectory data with a speed less than a second speed threshold from multiple smooth trajectory data to obtain target trajectory data, wherein the second speed threshold is different from the first speed threshold; or, determining the obstacle distance corresponding to multiple smooth trajectory data, and filtering smooth trajectory data with a trajectory length greater than the obstacle distance from multiple smooth trajectory data to obtain target trajectory data, wherein the obstacle distance is the collision distance between the traffic reference object and the obstacle.

[0132] In this embodiment, the aforementioned dynamic type of traffic reference object is other vehicles that generate extremely large displacement and speed changes. That is, the aforementioned dynamic type of traffic reference object can be understood as other vehicles that actively move in the traffic scene, and the values ​​corresponding to the displacement and speed changes of other vehicles are so large that they cannot be ignored.

[0133] In this embodiment, the second included angle threshold can be, but is not limited to, π / 4.

[0134] In this embodiment, the second angle between the current heading angle and the lane direction of the current lane corresponding to the smooth trajectory data is determined, and smooth trajectory data corresponding to the second angle greater than a second angle threshold is filtered from multiple smooth trajectory data to obtain target trajectory data. Optionally, after determining the second angle between the current heading angle and the lane direction of the current lane corresponding to the smooth trajectory data, this embodiment determines smooth trajectory data corresponding to the second angle greater than a second angle threshold from multiple smooth trajectory data, and deletes smooth trajectory data corresponding to the second angle greater than the second angle threshold from multiple smooth trajectory data to obtain target trajectory data.

[0135] In this embodiment, the aforementioned velocity difference threshold can be, but is not limited to, 5 m / s or 5.1 m / s.

[0136] In this embodiment, the speed difference between the speed represented by the motion analysis result and the current speed of the traffic reference object is determined, and smoothed trajectory data corresponding to speed differences greater than a speed difference threshold are filtered from multiple smoothed trajectory data to obtain target trajectory data. Optionally, after determining the speed difference between the speed represented by the motion analysis result and the current speed of the traffic reference object, this embodiment determines smoothed trajectory data corresponding to speed differences greater than a speed difference threshold from multiple smoothed trajectory data, and deletes smoothed trajectory data corresponding to speed differences greater than a speed difference threshold from multiple smoothed trajectory data to obtain target trajectory data.

[0137] In this embodiment, the first speed threshold can be, but is not limited to, 6 m / s or 5.9 m / s.

[0138] In this embodiment, the aforementioned length threshold can be, but is not limited to, 5m or 5.05m.

[0139] In this embodiment, the trajectory lengths of multiple smooth trajectory data are determined, and smooth trajectory data with lateral driving intentions of U-turns, speeds greater than a first speed threshold, and trajectory lengths less than a length threshold are filtered from the multiple smooth trajectory data to obtain target trajectory data. Optionally, after determining the trajectory lengths of multiple smooth trajectory data, this embodiment determines smooth trajectory data with lateral driving intentions of U-turns, speeds greater than a first speed threshold, and trajectory lengths less than a length threshold from the multiple smooth trajectory data, and deletes smooth trajectory data with lateral driving intentions of U-turns, speeds greater than a first speed threshold, and trajectory lengths less than a length threshold from the multiple smooth trajectory data to obtain target trajectory data.

[0140] In this embodiment, the second speed threshold is different from the first speed threshold. For example, if the first speed threshold is 6 m / s, then the second speed threshold is 4.5 m / s. These values ​​are only illustrative and are not specifically limited.

[0141] In this embodiment, target trajectory data is obtained by filtering smooth trajectory data with speeds less than a second speed threshold from multiple smooth trajectory data. Optionally, this embodiment can obtain target trajectory data by determining smooth trajectory data with speeds less than the second speed threshold from multiple smooth trajectory data, and by deleting smooth trajectory data with speeds less than the second speed threshold from multiple smooth trajectory data.

[0142] In this embodiment, the obstacle distance is the collision distance between the traffic reference object and the obstacle. For example, the obstacle can be a static obstacle in front.

[0143] In this embodiment, obstacle distances corresponding to multiple smooth trajectory data are determined, and smooth trajectory data with trajectory lengths greater than obstacle distances are filtered from the multiple smooth trajectory data to obtain target trajectory data. Optionally, after determining the obstacle distances corresponding to multiple smooth trajectory data, this embodiment can obtain target trajectory data by determining smooth trajectory data with trajectory lengths greater than obstacle distances from the multiple smooth trajectory data, and deleting smooth trajectory data with trajectory lengths greater than obstacle distances from the multiple smooth trajectory data.

[0144] Since the traffic reference object is dynamic, the target trajectory data can be obtained by filtering the smooth trajectory data that needs to be filtered from multiple smooth trajectory data according to the trajectory evaluation scheme suitable for the dynamic type. This achieves the purpose of determining the target trajectory data from multiple smooth trajectory data, thereby realizing the technical effect of improving the prediction accuracy of the trajectory data of the traffic object.

[0145] The following section further describes the steps of smoothing multiple predicted trajectory data to obtain multiple smooth trajectory data in this embodiment.

[0146] As an optional embodiment, step S102 involves smoothing multiple predicted trajectory data to obtain multiple smoothed trajectory data, including: smoothing multiple predicted trajectory data to obtain multiple initial smoothed trajectory data; resampling multiple initial smoothed trajectory data to obtain multiple sampling points for each of the multiple initial smoothed trajectory data; and generating multiple smoothed trajectory data based on the multiple sampling points of each initial smoothed trajectory data.

[0147] In this embodiment, the smoothing process described above is a smoothing operation performed on the trajectory curve.

[0148] In this embodiment, after acquiring multiple predicted trajectory data of the traffic reference object, the multiple predicted trajectory data are smoothed respectively to obtain multiple initial smoothed trajectory data. Optionally, this embodiment can obtain multiple initial smoothed trajectory data by performing curve smoothing on the multiple predicted trajectory data based on the acquired multiple predicted trajectory data.

[0149] In this embodiment, after smoothing multiple predicted trajectory data to obtain multiple initial smoothed trajectory data, the multiple initial smoothed trajectory data are resampled to obtain multiple sampling points for each of the multiple initial smoothed trajectory data. Optionally, in this embodiment, a preset number of predicted trajectory points are extracted from the multiple initial smoothed trajectory data at the midpoint of each initial smoothed trajectory data, that is, multiple sampling points for each of the multiple initial smoothed trajectory data can be obtained. Next, in each initial smoothed trajectory data, a polynomial is used to fit the predicted trajectory composed of the preset number of extracted predicted trajectory points to obtain the corresponding fifth-order polynomial coefficients. Then, according to a preset sampling interval, the predicted trajectory composed of the preset number of predicted trajectory points is resampled using the fifth-order polynomial with the fifth-order polynomial coefficients substituted, to obtain smoothed trajectory data corresponding to the predicted trajectories respectively.

[0150] In this embodiment, after resampling multiple initial smooth trajectory data to obtain multiple sampling points for each initial smooth trajectory data, multiple smooth trajectory data are generated based on these sampling points. Optionally, this embodiment utilizes the multiple sampling points of each data set to generate the trajectory, thereby obtaining multiple smooth trajectory data; that is, the multiple sampling points of each data set are used to restore the smoothed predicted trajectory.

[0151] Since multiple smooth trajectory data are generated from multiple sampling points of each of the multiple initial smooth trajectory data, the smooth trajectory data can be recovered, thereby achieving the technical effect of improving the accuracy of the smooth trajectory data.

[0152] In this embodiment of the invention, when processing the trajectory of a traffic reference object, multiple predicted trajectory data of the traffic reference object can be acquired; the multiple predicted trajectory data can be smoothed to obtain multiple smoothed trajectory data; motion analysis can be performed on the multiple smoothed trajectory data to obtain multiple motion analysis results; intent analysis can be performed on the multiple smoothed trajectory data to obtain multiple intent analysis results; and based on the motion state and driving intent, target trajectory data can be determined from the multiple smoothed trajectory data. Because this embodiment of the invention, after smoothing the acquired multiple predicted trajectory data to obtain multiple smoothed trajectory data, performs motion analysis on the aforementioned multiple smoothed trajectory data to obtain multiple motion analysis results, and performs intent analysis on the aforementioned multiple smoothed trajectory data to obtain multiple intent analysis results, and then, based on the motion state and driving intent, selects the smoothed trajectory data in a normal state from the multiple predicted trajectory data as the target trajectory data, thereby achieving the goal of optimizing the predicted trajectory data of the traffic object, thus solving the technical problem of low prediction accuracy of the trajectory data of the traffic object, and ultimately achieving the technical effect of improving the prediction accuracy of the trajectory data of the traffic object.

[0153] The technical solutions of the embodiments of the present invention will be illustrated below with reference to preferred embodiments.

[0154] Currently, when predicting the trajectory of traffic reference objects, the method of predicting the trajectory of the traffic reference objects based on preset rules is often adopted. However, the trajectory prediction method based on preset rules requires manual design, has poor flexibility, and is difficult to handle complex driving scenarios.

[0155] Furthermore, when predicting the trajectory of traffic reference objects, an end-to-end large model is often used to predict and output the trajectory of the aforementioned traffic reference objects. However, the trajectory obtained by the trajectory prediction method based on the end-to-end large model may violate kinematic constraints, resulting in the technical problem of low prediction accuracy of the trajectory data of traffic objects.

[0156] To address the aforementioned technical problems, this invention proposes a trajectory processing method for traffic reference objects. After smoothing multiple acquired predicted trajectory data to obtain multiple smoothed trajectory data, motion analysis is performed on these smoothed trajectory data to obtain multiple motion analysis results. Furthermore, intent analysis is performed on these smoothed trajectory data to obtain multiple intent analysis results. Then, based on the motion state and driving intent, the smoothed trajectory data in a normal state can be selected as the target trajectory data from the multiple predicted trajectory data. This achieves the goal of optimizing the predicted trajectory data of traffic objects, thereby solving the technical problem of low prediction accuracy of traffic object trajectory data and ultimately improving the prediction accuracy of traffic object trajectory data.

[0157] In this embodiment, by executing a post-processing method for predicting trajectories based on large models, the target trajectory data can be determined from multiple predicted trajectory data based on scene data from multiple motion parsing results, multiple intent parsing results, and traffic reference objects. For example, Figure 2 This is a flowchart of a post-processing method for predicting trajectories using a large model, according to an embodiment of the present invention. Figure 2 As shown, the method may include the following steps:

[0158] Step S201: Using the large model, output the predicted trajectories of traffic participants.

[0159] After using a large model to output the predicted trajectories of traffic participants, step S202 is performed to smooth the predicted trajectories.

[0160] In the technical solution provided by step S202 of the present invention, seven predicted trajectory points are extracted from the multiple predicted trajectory data at the midpoint of each predicted trajectory. Then, in each predicted trajectory, two sets of polynomials (tx, ty) are used to fit the predicted trajectory composed of the extracted predicted trajectory points to obtain the corresponding fifth-order polynomial coefficients. Then, at a sampling interval of 0.1s, the predicted trajectory composed of the predicted trajectory points is resampled using the fifth-order polynomial with the fifth-order polynomial coefficients substituted, and the smoothed predicted trajectory can be recovered.

[0161] After smoothing the predicted trajectory, step S203 is executed to analyze the predicted trajectory.

[0162] In the technical solution provided in step S203 of the present invention, four types of general kinematic parsers and two types of scene intent parsers are used to parse the kinematic information and abstract semantic information of the trajectory from the coordinate information in the candidate trajectory.

[0163] In this embodiment, the above four types of general kinematic resolvers may include: a general velocity resolver, used to calculate the trajectory velocity value based on the displacement difference of discrete points on the trajectory, as shown in the following formula: , where V i It can be used to represent trajectory velocity values. x i It can be used to represent the x-coordinate of the location of traffic participants at future times. x i-1 It can be used to represent the x-coordinate of the position of a traffic participant at the time preceding a future time. y i It can be used to represent the ordinate of the position of traffic participants at future times. y i-1 It can be used to represent the vertical coordinate of the position of a traffic participant at the time preceding a future time. t i It can be used to represent future moments. t i-1 It can be used to represent the moment before a future moment; the universal acceleration resolver is used to calculate the trajectory acceleration value based on the velocity difference between discrete points on the trajectory, as shown in the following formula: ,in, a i It can be used to represent the trajectory acceleration value, V i-1 It can be used to represent the previous velocity value of the trajectory velocity value; the general heading resolver is used to calculate the trajectory heading value based on the displacement vector of the discrete points of the trajectory, as shown in the following formula: ,in, yaw iIt can be used to represent the trajectory heading value; the general angular velocity resolver is used to calculate the trajectory angular velocity based on the heading difference between discrete points on the trajectory, as shown in the following formula: ,in, yaw_rate i It can be used to represent the angular velocity of a trajectory. yaw i-1 It can be used to represent the previous heading value of the trajectory heading value.

[0164] In this embodiment, the two types of scene intent parsers mentioned above may include: a vertical intent parser and a horizontal intent parser.

[0165] In this embodiment, the longitudinal intent parser can determine which of the following longitudinal intents is based on Table 1: constant speed (CONSTANT) intent, start-up (START_UP) intent, speed-up (SPEED_UP) intent, slow-down (SLOW_DOWN) intent, pull-up (PULL_UP) intent, and stationary (STATIC) intent.

[0166] Table 1 Vertical Intent Determination Table

[0167]

[0168] In this embodiment, the aforementioned lateral intent parser can first determine the NO_TURN / TURN_RIGHT / TURN_LEFT / TURN_BACK intent of the traffic participant based on the trajectory end heading and the current heading. If the intent is determined to be NO_TURN, the intent is then determined based on the topological relationship between the trajectory end matching lane and the current lane and the angle difference.

[0169] Table 2 Lateral Intent Determination Table

[0170]

[0171] After parsing the predicted trajectory, step S204 is executed to evaluate the predicted trajectory.

[0172] In the technical solution provided in step S204 of the present invention, the rationality of the candidate trajectories is evaluated based on the kinematic and intent information parsed from the candidate trajectories, combined with the scenario analysis of the target traffic participants. Specifically, by evaluating the rationality of the candidate trajectories, unreasonable trajectories can be eliminated, retaining only reasonable ones. It should be noted that the evaluator differs for different scenarios and different types of traffic participants.

[0173] For example, the evaluator can be as shown in Table 3 below:

[0174] Table 3 Evaluation Scheme for Candidate Trajectories

[0175]

[0176]

[0177]

[0178] In this embodiment, when processing the trajectory of a traffic reference object, multiple predicted trajectory data of the traffic reference object can be acquired; the multiple predicted trajectory data are smoothed to obtain multiple smooth trajectory data; motion analysis is performed on the multiple smooth trajectory data to obtain multiple motion analysis results; intent analysis is performed on the multiple smooth trajectory data to obtain multiple intent analysis results; and based on the motion state and driving intent, target trajectory data is determined from the multiple smooth trajectory data. Because this embodiment of the invention, after smoothing the acquired multiple predicted trajectory data to obtain multiple smooth trajectory data, performs motion analysis on the aforementioned multiple smooth trajectory data to obtain multiple motion analysis results, and performs intent analysis on the aforementioned multiple smooth trajectory data to obtain multiple intent analysis results, and then, based on the motion state and driving intent, selects the smooth trajectory data in a normal state from the multiple predicted trajectory data as the target trajectory data, thereby achieving the goal of optimizing the predicted trajectory data of the traffic object, thus solving the technical problem of low prediction accuracy of the trajectory data of the traffic object, and ultimately achieving the technical effect of improving the prediction accuracy of the trajectory data of the traffic object.

[0179] According to an embodiment of the present invention, a trajectory processing device for a traffic reference object is also provided. It should be noted that this trajectory processing device for a vehicle's traffic reference object can be used to execute a trajectory processing method for a traffic reference object as described in the embodiment.

[0180] Figure 3 This is a schematic diagram of a trajectory processing device for a traffic reference object according to an embodiment of the present invention. Figure 3 As shown, the trajectory processing device 300 for the traffic reference object may include: an acquisition unit 301, a processing unit 302, a motion analysis unit 303, an intent analysis unit 304, and a first determination unit 305.

[0181] The acquisition unit 301 is used to acquire multiple predicted trajectory data of the traffic reference object, wherein the predicted trajectory data is used to represent the trajectory of the traffic reference object in the future time period.

[0182] The processing unit 302 is used to smooth multiple predicted trajectory data to obtain multiple smoothed trajectory data.

[0183] The motion analysis unit 303 is used to perform motion analysis on multiple smooth trajectory data to obtain multiple motion analysis results. The motion analysis results are used to represent the motion state of the traffic reference object corresponding to each smooth trajectory data in the future time period.

[0184] The intent parsing unit 304 is used to perform intent parsing on multiple smooth trajectory data to obtain multiple intent parsing results. The intent parsing results are used to represent the driving intent of the traffic reference object corresponding to each smooth trajectory data in the future time period.

[0185] The first determining unit 305 is used to determine target trajectory data from multiple smooth trajectory data based on motion state and driving intention, wherein the target trajectory data is the smooth trajectory data in a normal state among the multiple smooth trajectory data.

[0186] Optionally, the trajectory processing device 300 for the traffic reference object may further include: a comparison unit for comparing the current heading angle of the traffic reference object with the final heading angle of the corresponding trajectory of the smooth trajectory data to obtain a comparison result, wherein the comparison result is used to represent the deviation between the current heading angle and the final heading angle; and a second determination unit for determining the lateral driving intention based on the comparison result.

[0187] Optionally, the second determining unit may include: a first determining module, used to determine the lateral driving intention as going straight when the comparison result indicates that the final heading angle is the same as the current heading angle; and a second determining module, used to determine the lateral driving intention as turning around, turning left, or turning right when the comparison result indicates that the final heading angle is different from the current heading angle.

[0188] Optionally, the trajectory processing device 300 for the traffic reference object may further include: a third determining unit, configured to determine the end lane of the trajectory corresponding to the smoothed trajectory data, and the predicted end heading angle of the traffic reference object at the end of the trajectory, wherein the end lane is used to represent the lane where the end trajectory point is located in the trajectory, and the end heading angle is used to represent the heading angle of the traffic reference object at the end trajectory point; a fourth determining unit, configured to determine the lane position relationship between the end lane and the current lane where the traffic reference object is located; and to determine the angle difference between the end heading angle and the current heading angle of the traffic reference object; and a fifth determining unit, configured to determine the lateral driving intention based on the lane position relationship and the angle difference.

[0189] Optionally, the trajectory processing device 300 for the traffic reference object may further include: a first determining and filtering unit, configured to determine the heading angle of the traffic reference object on the non-U-turn lane and the first included angle between the heading angle and the lane direction of the non-U-turn lane; filter smooth trajectory data corresponding to the first included angle that is greater than a first included angle threshold from multiple smooth trajectory data corresponding to the traffic reference object to obtain target trajectory data; a second determining and filtering unit, configured to determine multiple starting trajectory data from multiple smooth trajectory data, wherein the starting trajectory data represents the smooth trajectory data corresponding to the traffic reference object in the starting state in the smooth trajectory data; determine the probability of multiple starting trajectory data occurring in a real traffic scenario; filter starting trajectory data corresponding to probabilities less than a probability threshold from multiple smooth trajectory data corresponding to the traffic reference object to obtain target trajectory data; a third determining and filtering unit, configured to determine multiple starting trajectory data that is greater than a first included angle threshold between the heading angle and the lane direction of the non-U-turn lane; and ... A filtering unit is used to determine the average acceleration corresponding to multiple starting trajectory data; from multiple smooth trajectory data corresponding to traffic reference objects, the starting trajectory data corresponding to the average acceleration exceeding the average acceleration threshold is filtered to obtain the target trajectory data; a fourth determining and filtering unit is used to determine the driving intention of multiple starting trajectory data; from multiple smooth trajectory data corresponding to traffic reference objects, the starting trajectory data corresponding to the driving intention of making a U-turn is filtered to obtain the target trajectory data; a fifth determining and filtering unit is used to determine the starting scenario data corresponding to multiple starting trajectory data; when there is starting scenario data that is the same as the target scenario data among the multiple starting scenario data, the starting trajectory data corresponding to the target scenario data is filtered from the multiple smooth trajectory data to obtain the target trajectory data, wherein the target scenario data is a scenario where the traffic light is red within a preset distance.

[0190] Optionally, the trajectory processing device 300 for the traffic reference object may further include: a sixth determining and filtering unit, configured to determine the second included angle between the current heading angle corresponding to the smooth trajectory data and the lane direction of the current lane, and to filter smooth trajectory data corresponding to the second included angle that is greater than a second included angle threshold from multiple smooth trajectory data to obtain target trajectory data; a seventh determining and filtering unit, configured to determine the speed difference between the speed represented by the motion analysis result and the current speed of the traffic reference object, and to filter smooth trajectory data corresponding to the speed difference that is greater than a speed difference threshold from multiple smooth trajectory data to obtain target trajectory data; and an eighth determining and filtering unit, configured to determine the trajectory length of multiple smooth trajectory data, so as to... The system includes a ninth determining and filtering unit, which filters smooth trajectory data from multiple smooth trajectory data to obtain target trajectory data, provided that the lateral driving intention is a U-turn intention, the speed is greater than a first speed threshold, and the trajectory length is less than a length threshold. The tenth determining and filtering unit is used to determine the obstacle distances corresponding to the multiple smooth trajectory data, and to filter smooth trajectory data from the multiple smooth trajectory data to obtain target trajectory data, provided that the second speed threshold is different from the first speed threshold. The obstacle distance is the collision distance between the traffic reference object and the obstacle.

[0191] Optionally, the processing unit 302 may include: a smoothing module for smoothing multiple predicted trajectory data respectively to obtain multiple initial smoothed trajectory data; a resampling module for resampling multiple initial smoothed trajectory data to obtain multiple sampling points for each of the multiple initial smoothed trajectory data; and a generation module for generating multiple smoothed trajectory data based on the multiple sampling points of each initial smoothed trajectory data.

[0192] In this embodiment, a trajectory processing device for a traffic reference object is provided. The device may include: an acquisition unit for acquiring multiple predicted trajectory data of the traffic reference object, wherein the predicted trajectory data represents the trajectory of the traffic reference object in a future time period; a processing unit for smoothing the multiple predicted trajectory data to obtain multiple smoothed trajectory data; a motion analysis unit for performing motion analysis on the multiple smoothed trajectory data to obtain multiple motion analysis results, wherein the motion analysis results represent the motion state of the traffic reference object corresponding to each smoothed trajectory data in a future time period; an intent analysis unit for performing intent analysis on the multiple smoothed trajectory data respectively to obtain multiple intent analysis results, wherein the intent analysis results represent the driving intent of the traffic reference object corresponding to each smoothed trajectory data in a future time period; and a determination unit for determining target trajectory data from the multiple smoothed trajectory data based on the motion state and driving intent, wherein the target trajectory data is the smoothed trajectory data in a normal state among the multiple smoothed trajectory data. This achieves the goal of optimizing the predicted trajectory data of the traffic object, thereby solving the technical problem of low prediction accuracy of the trajectory data of the traffic object, and ultimately realizing the technical effect of improving the prediction accuracy of the trajectory data of the traffic object.

[0193] According to an embodiment of the present invention, a vehicle is also provided. Figure 4 This is a schematic diagram of an electronic device according to an embodiment of the present invention, such as... Figure 4 As shown, the electronic device 400 may include a memory 410 and a processor 420, wherein the memory 410 is used to store computer programs; and the processor 420 is used to run the programs stored in the memory 410 to implement the trajectory processing method for traffic reference objects of this application.

[0194] In this application, "multiple" refers to two or more.

[0195] In this application, unless otherwise expressly defined, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0196] The terms “first,” “second,” “third,” “fourth,” etc., in this application (if present) are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0197] In this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, in this application, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0198] According to an embodiment of the present invention, a processor is also provided for running a program, wherein the program is executed by the processor to perform the trajectory processing method for the traffic reference object in the embodiment.

[0199] According to an embodiment of the present invention, a computer program product is also provided, the computer program product including a computer program, wherein when the computer program is executed by a processor, it implements the trajectory processing method for traffic reference objects in the embodiment.

[0200] According to an embodiment of the present invention, a computer program product is also provided, including a non-volatile computer-readable storage medium for storing a computer program. When the computer program is executed by a processor, it implements the trajectory processing method for traffic reference objects in the embodiment.

[0201] According to an embodiment of the present invention, a computer program is also provided, which, when executed by a processor, implements the trajectory processing method for the traffic reference object in the embodiment.

[0202] According to another aspect of the present invention, a computer-readable storage medium is also provided. The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the trajectory processing method for a traffic reference object in the embodiment.

[0203] Computer-readable storage media, also known as computer storage media, may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. These propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable storage media can transmit, propagate, or transfer programs for use by or in conjunction with an instruction execution system, apparatus, or device.

[0204] The program code contained in a computer-readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, radio frequency, or any suitable combination thereof.

[0205] According to an embodiment of the present invention, a computer program product is also provided, the computer program product including a computer program, wherein when the computer program is executed by a processor, it implements the vehicle transaction information processing method in the embodiment.

[0206] According to an embodiment of the present invention, a computer program product is also provided, including a non-volatile computer-readable storage medium for storing a computer program, which, when executed by a processor, implements the vehicle transaction information processing method in the embodiment.

[0207] According to an embodiment of the present invention, a computer program is also provided, which, when executed by a processor, implements the vehicle transaction information processing method in the embodiment.

[0208] Optionally, when the above computer program is executed by the processor, the program code implements the following steps: acquiring multiple predicted trajectory data of a traffic reference object, wherein the predicted trajectory data is used to represent the trajectory of the traffic reference object in a future time period; smoothing the multiple predicted trajectory data to obtain multiple smoothed trajectory data; performing motion analysis on the multiple smoothed trajectory data to obtain multiple motion analysis results, wherein the motion analysis results are used to represent the motion state of the traffic reference object corresponding to each smoothed trajectory data in a future time period; performing intent analysis on the multiple smoothed trajectory data to obtain multiple intent analysis results, wherein the intent analysis results are used to represent the driving intent of the traffic reference object corresponding to each smoothed trajectory data in a future time period; and determining target trajectory data from the multiple smoothed trajectory data based on the motion state and driving intent, wherein the target trajectory data is the smoothed trajectory data in a normal state among the multiple smoothed trajectory data.

[0209] Optionally, when the above computer program is executed by the processor, the program code implements the following steps: comparing the current heading angle of the traffic reference object with the end heading angle of the corresponding trajectory of the smooth trajectory data to obtain a comparison result, wherein the comparison result is used to represent the deviation between the current heading angle and the end heading angle; and determining the lateral driving intention based on the comparison result.

[0210] Optionally, when the above computer program is executed by the processor, the program code implements the following steps: when the comparison result indicates that the final heading angle is the same as the current heading angle, the lateral driving intention is determined to be straight; when the comparison result indicates that the final heading angle is different from the current heading angle, the lateral driving intention is determined to be a U-turn, or a left turn, or a right turn.

[0211] Optionally, when the above computer program is executed by the processor, the program code implements the following steps: determining the end lane of the trajectory corresponding to the smooth trajectory data, and the predicted end heading angle of the traffic reference object at the end of the trajectory, wherein the end lane is used to represent the lane where the end trajectory point is located in the trajectory, and the end heading angle is used to represent the heading angle of the traffic reference object at the end trajectory point; determining the lane position relationship between the end lane and the current lane where the traffic reference object is located; and determining the angle difference between the end heading angle and the current heading angle of the traffic reference object; and determining the lateral driving intention based on the lane position relationship and the angle difference.

[0212] Optionally, when the above computer program is executed by the processor, the program code implements the following steps: determining the heading angle of the traffic reference object in the non-U-turn lane and the first angle between the heading angle and the lane direction of the non-U-turn lane; filtering smooth trajectory data corresponding to the first angle that is greater than a threshold from multiple smooth trajectory data corresponding to the traffic reference object to obtain target trajectory data; or, determining multiple starting trajectory data from multiple smooth trajectory data, wherein the starting trajectory data represents the smooth trajectory data corresponding to the traffic reference object in the starting state in the smooth trajectory data; determining the probability of the multiple starting trajectory data occurring in a real traffic scenario; filtering starting trajectory data corresponding to probabilities less than a probability threshold from the multiple smooth trajectory data corresponding to the traffic reference object to obtain target trajectory data; or Alternatively, determine the average acceleration corresponding to multiple starting trajectory data; filter out the starting trajectory data corresponding to the average acceleration greater than the average acceleration threshold from multiple smooth trajectory data corresponding to the traffic reference object to obtain the target trajectory data; or, determine the driving intention of multiple starting trajectory data; filter out the starting trajectory data corresponding to the driving intention of making a U-turn from multiple smooth trajectory data corresponding to the traffic reference object to obtain the target trajectory data; or, determine the starting scenario data corresponding to multiple starting trajectory data; if there is a starting scenario data that is the same as the target scenario data among the multiple starting scenario data, filter out the starting trajectory data corresponding to the target scenario data from the multiple smooth trajectory data to obtain the target trajectory data, wherein the target scenario data is a scenario where the traffic light is red within a preset distance.

[0213] Optionally, when the above computer program is executed by the processor, the program code implements the following steps: determining the second included angle between the current heading angle corresponding to the smooth trajectory data and the lane direction of the current lane, and filtering smooth trajectory data corresponding to the second included angle that is greater than a second included angle threshold from multiple smooth trajectory data to obtain target trajectory data; or, determining the speed difference between the speed represented by the motion analysis result and the current speed of the traffic reference object, and filtering smooth trajectory data corresponding to the speed difference that is greater than a speed difference threshold from multiple smooth trajectory data to obtain target trajectory data; or, determining the trajectory length of multiple smooth trajectory data, and filtering smooth trajectory data corresponding to the lateral driving intention of a U-turn, the speed that is greater than a first speed threshold, and the trajectory length that is less than a length threshold from multiple smooth trajectory data to obtain target trajectory data; or, filtering smooth trajectory data with a speed less than a second speed threshold from multiple smooth trajectory data to obtain target trajectory data, wherein the second speed threshold is different from the first speed threshold; or, determining the obstacle distance corresponding to multiple smooth trajectory data, and filtering smooth trajectory data with a trajectory length greater than the obstacle distance from multiple smooth trajectory data to obtain target trajectory data, wherein the obstacle distance is the collision distance between the traffic reference object and the obstacle.

[0214] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0215] In the above embodiments of the present invention, 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.

[0216] 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 example, 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 couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0217] The units described as separate components may or may not be physically separate. Similarly, the components shown as units may or may not be physical units; 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, depending on actual needs.

[0218] Furthermore, the functional units in the various embodiments of the present invention 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.

[0219] 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 invention, in essence, or the part that contributes to related technologies, 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 of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0220] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A trajectory processing method for a traffic reference object, characterized in that, include: Acquire multiple predicted trajectory data of a traffic reference object, wherein the predicted trajectory data is used to represent the trajectory of the traffic reference object in a future time period; The multiple predicted trajectory data are smoothed to obtain multiple smoothed trajectory data; Motion analysis is performed on the multiple smooth trajectory data respectively to obtain multiple motion analysis results, wherein the motion analysis results are used to represent the motion state of the traffic reference object corresponding to each smooth trajectory data in the future time period; Intent parsing is performed on the multiple smooth trajectory data respectively to obtain multiple intent parsing results, wherein the intent parsing results are used to represent the driving intent of the traffic reference object corresponding to each smooth trajectory data in the future time period; Based on the motion state and the driving intention, target trajectory data is determined from the plurality of smooth trajectory data, wherein the target trajectory data is the smooth trajectory data in a normal state among the plurality of smooth trajectory data.

2. The method according to claim 1, characterized in that, The motion state includes at least one of velocity, acceleration, heading angle, and angular velocity; The driving intention includes longitudinal driving intention and / or lateral driving intention.

3. The method according to claim 2, characterized in that, The lateral driving intention includes at least one of U-turn, left turn, right turn, straight ahead, left lane change, and right lane change.

4. The method according to claim 3, characterized in that, The method includes: The current heading angle of the traffic reference object is compared with the final heading angle of the trajectory corresponding to the smooth trajectory data to obtain a comparison result, wherein the comparison result is used to represent the deviation between the current heading angle and the final heading angle; Based on the comparison results, the lateral driving intention is determined.

5. The method according to claim 4, characterized in that, Determining the lateral driving intention based on the comparison results includes: When the comparison result indicates that the final heading angle is the same as the current heading angle, the lateral driving intention is determined to be straight driving; When the comparison result indicates that the final heading angle is different from the current heading angle, the lateral driving intention is determined to be either a U-turn, a left turn, or a right turn.

6. The method according to claim 5, characterized in that, When the lateral driving intention is to go straight, the method further includes: Determine the end lane of the trajectory corresponding to the smooth trajectory data, and predict the end heading angle of the traffic reference object at the end of the trajectory, wherein the end lane is used to represent the lane where the end trajectory point is located in the trajectory, and the end heading angle is used to represent the heading angle of the traffic reference object at the end trajectory point; Determine the lane position relationship between the end lane and the current lane where the traffic reference object is located; and, Determine the angle difference between the final heading angle and the current heading angle of the traffic reference object; The lateral driving intention is determined based on the lane position relationship and the angle difference.

7. The method according to claim 2, characterized in that, When the traffic reference object is of a static type, the method includes: Determine the heading angle of the traffic reference object located in the non-U-turn lane, and the first angle between it and the lane orientation of the non-U-turn lane; From the plurality of smooth trajectory data corresponding to the traffic reference object, filter out the smooth trajectory data corresponding to the first included angle that is greater than the first included angle threshold to obtain the target trajectory data; or... From the plurality of smooth trajectory data, a plurality of starting trajectory data are determined, wherein the starting trajectory data represents the smooth trajectory data corresponding to the traffic reference object in the starting state; the probability of the plurality of starting trajectory data occurring in a real traffic scenario is determined; from the plurality of smooth trajectory data corresponding to the traffic reference object, the starting trajectory data corresponding to probabilities less than a probability threshold are filtered to obtain the target trajectory data; or... Determine the average acceleration corresponding to multiple starting trajectory data; from the multiple smooth trajectory data corresponding to the traffic reference object, filter out the starting trajectory data corresponding to the average acceleration that is greater than the average acceleration threshold to obtain the target trajectory data; or, Determine the driving intention of multiple starting trajectory data; filter the starting trajectory data corresponding to the driving intention of making a U-turn from the multiple smooth trajectory data corresponding to the traffic reference object to obtain the target trajectory data; or... Determine the starting scenario data corresponding to multiple starting trajectory data; if there is starting scenario data that is the same as the target scenario data among the multiple starting scenario data, filter the starting trajectory data corresponding to the target scenario data from the multiple smooth trajectory data to obtain the target trajectory data, wherein the target scenario data is a scenario where the traffic light is red within a preset distance.

8. The method according to claim 7, characterized in that, When the traffic reference object is dynamic, the method includes: Determine the second angle between the current heading angle corresponding to the smooth trajectory data and the lane direction of the current lane, and filter the smooth trajectory data corresponding to the second angle that is greater than the second angle threshold from the plurality of smooth trajectory data to obtain the target trajectory data; or, Determine the speed difference between the speed represented by the motion analysis result and the current speed of the traffic reference object, and filter the smoothed trajectory data corresponding to speed differences greater than a speed difference threshold from the plurality of smoothed trajectory data to obtain the target trajectory data; or, Determine the trajectory length of the plurality of smooth trajectory data, and filter the smooth trajectory data from the plurality of smooth trajectory data that correspond to the lateral driving intention being a U-turn intention, the speed being greater than a first speed threshold, and the trajectory length being less than a length threshold, to obtain the target trajectory data; or, From the plurality of smooth trajectory data, smooth trajectory data with speeds less than a second speed threshold are filtered out to obtain the target trajectory data, wherein the second speed threshold is different from the first speed threshold; or... The obstacle distances corresponding to the plurality of smooth trajectory data are determined, and the smooth trajectory data with a trajectory length greater than the obstacle distance are filtered from the plurality of smooth trajectory data to obtain the target trajectory data, wherein the obstacle distance is the collision distance between the traffic reference object and the obstacle.

9. The method according to claim 1, characterized in that, The smoothing process is performed on the multiple predicted trajectory data to obtain multiple smoothed trajectory data, including: The multiple predicted trajectory data are smoothed to obtain multiple initial smoothed trajectory data; The multiple initial smooth trajectory data are resampled to obtain multiple sampling points for each of the multiple initial smooth trajectory data; Based on the respective multiple sampling points, the multiple smooth trajectory data are generated.

10. An electronic device, characterized in that, The method includes a memory and a processor, the memory storing an executable program, and the processor running the program, wherein the program, when run by the processor, performs the method according to any one of claims 1 to 9.

Images