Method and device for determining track confidence and sensing system

Abstract

The invention provides a method and equipment for determining track confidence and a sensing system. The method comprises the following steps: acquiring a first relationship between a target object and a first environment according to first track information of the target object and the first environment; and based on the first relationship, determining the airspace confidence of the first track information. In the embodiment of the invention, the spatial interaction relationship between the target object and the surrounding environment is introduced to determine the airspace confidence coefficient of the trajectory information of the target object. According to the scheme, on one hand, the possible influence of the surrounding environment on the motion behavior of the target object is considered, and the precision of the track confidence coefficient is improved; and on the other hand, the scheme can be suitable for various traffic driving scenes, such as scenes conforming to the inertia priori hypothesis and scenes not conforming to the inertia priori hypothesis, and the universality and robustness of the track confidence coefficient can be improved.

Description

Technical Field

[0001] This application relates to the field of intelligent transportation technology, and more specifically, to methods and devices for determining trajectory confidence and sensing systems. Background Technology

[0002] A commonly used method for trajectory confidence analysis is the chi-square test based on the laws of inertial motion. Specifically, the perceived object (or target object) typically follows the laws of inertial motion during its movement; that is, the motion state of the target object in its historical trajectory (including but not limited to direction, velocity, acceleration, etc.) tends to be consistent with the motion state of the newly perceived trajectory. Therefore, the chi-square test can determine the trajectory confidence based on the chi-square value of the deviation between the actual observed value and the theoretically inferred value of the target object's motion state.

[0003] However, due to the diversity of traffic scenarios, the motion state of the target object may not always follow the laws of inertial motion. In this case, if the confidence level of the target object's trajectory is determined based on the chi-square test, there may be some error. Summary of the Invention

[0004] This application provides a method, device, and sensing system for determining trajectory confidence, which helps to improve the accuracy of trajectory confidence by considering the interaction between the target object and the surrounding environment.

[0005] In a first aspect, a method for determining trajectory confidence is provided, the method comprising: acquiring first trajectory information of a target object, the first trajectory information being trajectory information generated by the target object while driving in a first environment, the first trajectory information being obtained by sensing the target object through a sensing device; acquiring a first relationship between the target object and the first environment based on the first trajectory information and the first environment; and determining the spatial confidence of the first trajectory information based on the first relationship.

[0006] Because related technologies, such as the chi-square test, only consider the trajectory of the target object itself within a historical time interval, that is, only consider the temporal movement pattern of a single object, the confidence level determined by the methods in related technologies can be called the temporal confidence level. Since the embodiments of this application consider the spatial interaction between the target object and its surrounding environment, the trajectory confidence level determined by the methods provided in the embodiments of this application can be called the spatial confidence level, or the spatial domain confidence level.

[0007] For example, the first environment can be the driving environment or driving environment of the target object. When the target object is driving in the first environment, the trajectory information obtained by the sensing device from the target object is the first trajectory information.

[0008] For example, the first trajectory information can be the trajectory information generated when the target object is driving in the first environment. The first environment can be the driving environment or driving environment of the target object. When the target object is driving in the first environment, the trajectory information obtained by the sensing device from sensing the target object is the first trajectory information.

[0009] For example, the probability of the first relationship occurring is determined, and then the spatial confidence of the first trajectory information is determined based on the characteristics of the driving scenario, such as general driving habits or prior rules, and the probability of the first relationship occurring.

[0010] It should be understood that the methods provided in the embodiments of this application can be implemented alone or in combination with time-domain confidence (such as chi-square test) mentioned in related technologies.

[0011] In this embodiment, the spatial confidence level of the target object's trajectory information is determined by introducing the spatial interaction relationship between the target object and its surrounding environment. This approach considers the potential influence of the surrounding environment on the target object's motion behavior, thus improving the accuracy of the trajectory confidence level. Furthermore, this approach is applicable to various traffic and driving scenarios, including those conforming to and those not conforming to the inertial prior assumption, thereby enhancing the universality and robustness of the trajectory confidence level.

[0012] In some embodiments, the first environment includes one or more of the following: the position of lane lines; the sensing range of the sensing device; the position of road guardrails; the position of gantry; the position of signs; the geographical environment; the movement speed of other objects around the target object; the trajectory information of the other objects; or the position of the other objects.

[0013] For example, the first environment may include road-related information, information related to other objects around the target object, geographical information about the road's surroundings, weather information, etc. For instance, road-related information may include information about road infrastructure, such as lane markings, guardrails, gantries, signs, and sensing devices. As another example, information related to other objects around the target object may include information about the distribution, location, speed, and changes in motion of other objects. As yet another example, geographical information about the road's surroundings may include one or more of flat areas, mountains, plains, plateaus, or cliffs. As yet another example, weather information may include one or more of rainy, snowy, windy, foggy, sunny, cloudy, and overcast weather.

[0014] The embodiments of this application help improve the accuracy of trajectory confidence by considering the influence of the surrounding environment on the target object.

[0015] In some embodiments, the first relationship includes one or more of the following: the relationship between the position of the target object and the position of the lane line; the relationship between the starting position of the first trajectory information within the sensing range of the sensing device and the sensing range; the relationship between the ending position of the first trajectory information within the sensing range and the sensing range; the relationship between the position of the target object and the position of the gantry; the relationship between the position of the target object and the position of the sign; the relationship between the position of the target object and the position of the road guardrail; the distance between the other object and the target object; or the difference between the movement speed of the other object and the movement speed of the target object.

[0016] For example, the first relationship may include the distance between the target object and the lane line, the distance between the target object and the lane center line, etc. Alternatively, the first relationship may include the distribution relationship between the target object and the lane center line, such as the target object traveling along the lane center line and the target object not traveling along the lane center line (or the target object traveling on the line).

[0017] For example, the first relationship may include the type of lane in which the target object is located, such as an overtaking lane, a ramp, or other lanes.

[0018] For example, the first relationship may include the distance between the target object and the gantry.

[0019] For example, the first relationship may include the distance between the target object and the sign.

[0020] For example, the first relationship may include the distance between the location of the target object and the location of the road guardrail.

[0021] For example, the first relation may include the distribution of other objects around the target object, such as dense (congested roads) or sparse (smooth roads).

[0022] For example, the distance between the other objects and the target object may include front-to-back distance, left-to-right distance, etc.

[0023] For example, the first relationship may include the geographical environment in which the target object is traveling as one or more of flat areas, mountains, plains, plateaus, or cliffs.

[0024] For example, the first relationship may include the target object's driving weather being one or more of the following: rainy, snowy, windy, foggy, sunny, cloudy, and overcast.

[0025] There are several ways to determine the primary relationship. For example, the interaction relationship between the target object and its surrounding environment can be obtained for multiple time periods, i.e., the primary relationship. For example, the interaction relationship between the target object and its surrounding environment can be obtained for different lanes. In other words, the spatial confidence of the primary trajectory information can be analyzed for different time periods, or the spatial confidence of the primary trajectory information can be analyzed for target objects in different lanes.

[0026] This application's embodiments expand the confidence analysis dimension by considering the interaction between the target object and its surrounding environment, which helps to solve the problem that traditional trajectory confidence analysis algorithms have a single prior assumption and cannot fully cover the distribution patterns of vehicle running trajectories in intelligent transportation scenarios.

[0027] In some embodiments, determining the spatial confidence level of the first trajectory information based on the first relationship includes: obtaining the probability of the first relationship occurring based on a first model; and determining the spatial confidence level of the first trajectory information based on the probability of the first relationship occurring.

[0028] In some scenarios, the probability of the first relationship occurring is directly or inversely proportional to the spatial confidence level of the first trajectory information. The relationship between the probability of the first relationship occurring and the spatial confidence level of the first trajectory information can be determined based on prior experience and the driving habits mentioned earlier.

[0029] Optionally, the relationship between the probability of the first relationship occurring and the spatial confidence of the first trajectory information can be stored in advance.

[0030] The probability of the first relationship occurring is determined by the first model, which eliminates the need to collect all cases of the first relationship in advance, making it simple to implement.

[0031] In some scenarios, when the probability of the interaction between the target object and its surrounding environment falls within different ranges, the relationship between this probability value and the spatial confidence level may differ. Considering this issue, the first model can be post-processed, or modified, based on prior rules or driving habits to determine the trajectory confidence level, thereby helping to improve the accuracy of the trajectory confidence level.

[0032] In some embodiments, artificial intelligence or machine learning techniques can be used to directly learn the relationship between spatial interaction relationships and trajectory confidence, resulting in a second model. That is, the input to the second model is the first relationship, and the output of the second model is the spatial confidence of the first trajectory information. This method is relatively simple to implement during operation.

[0033] In some embodiments, the method further includes: acquiring second trajectory information of a plurality of objects, wherein the second trajectory information is trajectory information generated by the plurality of objects when they travel in a test environment, wherein the plurality of objects include the target object, or the plurality of objects do not include the target object; determining a second relationship between the plurality of objects and the test environment based on the second trajectory information and the test environment; and fitting a probability distribution to the probability of the second relationship occurring to obtain the first model.

[0034] For example, statistics on the relationships between vehicles and their surroundings, and the probabilities of these relationships occurring, can be collected in a test environment to obtain a first model. For instance, statistical collection of interactions between multiple objects and their environment can be performed to improve the generality of the data.

[0035] For example, the second trajectory information is a reliable trajectory, such as one that can be obtained through manual measurement, or one that can be obtained through highly reliable sensing devices such as drones, in order to improve the reliability of the first model.

[0036] For example, the interaction between the target object and its surrounding static environment can be calculated over various time periods. The static environment referred to here broadly encompasses all static factors that may affect the vehicle's trajectory during operation, including but not limited to those mentioned above: lane markings, the sensing range of the sensing device, road guardrails, and the geographical environment. It should be understood that static environment information requires additional file input as support (e.g., high-precision map information for highways).

[0037] For example, the interaction between the target object and its surrounding dynamic environment can be calculated within each time period. The dynamic environment mentioned here refers to all motion factors that may affect the vehicle's trajectory during its movement, including but not limited to those mentioned above: motor vehicles, non-motorized two-wheeled vehicles, pedestrians, etc.

[0038] In some embodiments, the second relationship described above can be stored in a storage device. For example, the establishment of the first model can be triggered when the amount of data in the stored second relationship reaches a certain threshold. This is because if the amount of data in the second relationship is small, the accuracy of the first model obtained based on the second relationship may be low.

[0039] For example, the first model can be obtained based on the statistical results of the second relationship, or by fitting a probability distribution to the probability of the second relationship occurring. For instance, probability distribution models such as Gaussian, Poisson, and Gamma distributions can be used to fit the probability distribution of the second relationship occurring. The above probability distribution models are merely illustrative and are not intended to limit the scope of this application.

[0040] It should be understood that a single initial model can be built for one object or for multiple objects separately. For example, it can be built for all objects within the area sensed by the sensing device. As an example, it can be built for all vehicles sensed by the sensing device within a specific lane.

[0041] In some embodiments, the test environment is the first environment. In this way, the probability of the first relationship obtained through the first model matching the first environment is higher, and the output accuracy of the first model is higher. From the perspective of the sensing device, sensing analysis module, or sensing system, the first environment can also be referred to as the operating environment of the sensing device, sensing analysis module, or sensing system. High consistency between the test environment and the operating environment helps improve the accuracy of the obtained trajectory spatial confidence.

[0042] In some embodiments, the first relationship is used to update the first model.

[0043] The embodiments of this application help improve the accuracy of the first model by updating the first model.

[0044] For example, the probability statistics of the occurrence of the first relationship can be updated based on the first relationship, and the first model can be updated based on the results.

[0045] For example, the first relation can be stored in a storage unit for updating the first model. As an example, when the amount of data in the stored first relation reaches a preset threshold, the update of the first model can be triggered, thereby helping to avoid the overhead caused by frequent updates to the first model.

[0046] In some embodiments, the first trajectory information includes multiple trajectory information, which are obtained by multiple sensing devices sensing the target object respectively. The method further includes: obtaining third trajectory information of the target object based on the spatial confidence scores corresponding to the multiple trajectory information respectively, wherein the third trajectory information is the fusion result of the multiple trajectory information.

[0047] In some embodiments, trajectory confidence can be used to determine whether the data in the trajectory information is a true value or a false alarm, thereby eliminating false alarms in the trajectory information and improving the reliability of the trajectory information.

[0048] It should be understood that when performing trajectory fusion, trajectory fusion can be performed using only the spatial confidence provided in this application, or it can be performed by combining the spatial confidence provided in this application with the temporal confidence in related technologies.

[0049] In some embodiments, the method further includes: identifying traffic events related to the target object based on the spatial confidence level of the first trajectory information, wherein the traffic events include one or more of parking, speeding, driving in the wrong direction, and lane changing. This approach expands the application scope of confidence analysis results in trajectory post-processing scenarios.

[0050] In some embodiments, the first environment is determined based on map information associated with the first trajectory information, and / or the first environment is determined based on the perception results of other objects around the target object.

[0051] For example, the location of lane lines, the location of road guardrails, and environmental information such as the geographical environment around the road are obtained based on the map information associated with the first trajectory information.

[0052] The map information associated with the first trajectory information includes map information of the roads where the first trajectory information is located. Obtaining the first environmental information based on the map information helps reduce the complexity and processing overhead of obtaining the first environmental information.

[0053] Based on the perception results, the location of lane lines, road guardrails, signs, and other elements of the primary environment can be obtained. Similarly, based on the perception results of other objects around the target object, such as their speed, trajectory information, and location, the primary environment can be determined. Obtaining the primary environment based on the perception results of the sensing device helps improve the accuracy of the acquired primary environment.

[0054] Secondly, a device for determining trajectory confidence is provided. The device includes: a first acquisition module for acquiring first trajectory information of a target object, wherein the first trajectory information is trajectory information generated by the target object while driving in a first environment, and the first trajectory information is obtained by sensing the target object through a sensing device; a second acquisition module for acquiring a first relationship between the target object and the first environment based on the first trajectory information and the first environment; and a determination module for determining the spatial confidence of the first trajectory information based on the first relationship.

[0055] In this embodiment, the spatial confidence level of the target object's trajectory information is determined by introducing the spatial interaction relationship between the target object and its surrounding environment. This approach considers the potential influence of the surrounding environment on the target object's motion behavior, thus improving the accuracy of the trajectory confidence level. Furthermore, this approach is applicable to various traffic and driving scenarios, including those conforming to and those not conforming to the inertial prior assumption, thereby enhancing the universality and robustness of the trajectory confidence level.

[0056] Thirdly, an apparatus for determining trajectory confidence is provided, the apparatus comprising: at least one memory for storing computer programs or instructions; and at least one processor for executing some or all of the computer programs or instructions in the at least one memory to cause the method described in the first aspect to be performed.

[0057] Fourthly, a perception system is provided, comprising: a plurality of sensing devices for acquiring first trajectory information of a target object, the first trajectory information including a plurality of trajectory information obtained by the plurality of sensing devices respectively sensing the target object, the first trajectory information being trajectory information generated by the target object traveling in a first environment; a device for determining trajectory confidence as described in the second or third aspect, for determining the spatial confidence of the first trajectory information based on the first trajectory information and the first environment; and a processing device for acquiring third trajectory information of the target object according to the spatial confidence, the third trajectory information being a fusion result of the plurality of trajectory information.

[0058] Fifthly, a computer program product is provided, comprising a computer program, which, when executed by a processor, performs the methods in the possible implementations of the first aspect above.

[0059] In a sixth aspect, a computer-readable storage medium is provided, which stores a computer program that, when part or all of the computer program is executed, performs the methods in the possible implementations of the first aspect above.

[0060] In a seventh aspect, a chip is provided, comprising: a processor for retrieving and running part or all of a computer program from a memory, causing a device equipped with the chip to perform the methods in the possible implementations of the first aspect above. Attached Figure Description

[0061] Figure 1 This is a schematic diagram of a roadside sensing scenario;

[0062] Figure 2 This is a schematic diagram of the perception trajectory analysis and processing flow based on the perception engine 120;

[0063] Figure 3 A flowchart illustrating a method for determining trajectory confidence provided in an embodiment of this application;

[0064] Figure 4 An example diagram illustrating the distribution relationship between the target object and the lane centerline provided in this application embodiment;

[0065] Figure 5An example diagram illustrating the relationship between the target object and the sensing range of the sensing device provided in this application embodiment;

[0066] Figure 6 This is an example diagram illustrating the distribution relationship between the target object and other surrounding objects provided in an embodiment of this application.

[0067] Figure 7 This is an example diagram illustrating the velocity difference distribution between the target object and other surrounding objects provided in an embodiment of this application.

[0068] Figure 8 A flowchart illustrating the method for establishing a first model provided in an embodiment of this application;

[0069] Figure 9 A flowchart illustrating another method for determining trajectory confidence provided in an embodiment of this application;

[0070] Figure 10 A flowchart illustrating another method for determining trajectory confidence provided in an embodiment of this application;

[0071] Figure 11 A comparative diagram of temporal domain confidence and spatial domain confidence provided in an embodiment of this application;

[0072] Figure 12 Another comparative schematic diagram of temporal domain confidence and spatial domain confidence provided for embodiments of this application;

[0073] Figure 13 This is another comparative diagram of temporal confidence and spatial confidence provided in the embodiments of this application;

[0074] Figure 14 This is another comparative diagram of temporal confidence and spatial confidence provided in the embodiments of this application;

[0075] Figure 15 A schematic diagram of the structure of a device for determining trajectory confidence provided in an embodiment of this application;

[0076] Figure 16 A schematic diagram of another device for determining trajectory confidence provided in an embodiment of this application;

[0077] Figure 17 A schematic diagram of another device for determining trajectory confidence provided in an embodiment of this application;

[0078] Figure 18 A schematic diagram of the structure of a sensing system provided in an embodiment of this application;

[0079] Figure 19 This is a schematic diagram of another sensing system provided in an embodiment of this application;

[0080] Figure 20 This is a schematic diagram of the structure of another sensing system provided in an embodiment of this application. Detailed Implementation

[0081] The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.

[0082] In the description of the embodiments of this application, unless otherwise stated, " / " indicates that the objects before and after are in an "or" relationship. For example, A / B can represent A or B. "And / or" in the embodiments of this application 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 alone, A and B simultaneously, and B alone, where A and B can be singular or plural. Furthermore, in the description of the embodiments of this application, unless otherwise stated, "multiple" refers to two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple. Additionally, to facilitate a clear description of the technical solutions of the embodiments of this application, the terms "first" and "second" are used in the embodiments of this application to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the words "first" and "second" do not limit the quantity or the order of execution, and that the words "first" and "second" do not necessarily imply that they are different.

[0083] In the various method embodiments of this application, the order of the sequence numbers does not imply the order of execution. The execution order should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0084] It is understood that in the embodiments of this application, descriptions such as "under the circumstances," "if," "when," and "if..." can be used interchangeably. Furthermore, these descriptions all refer to the corresponding processing that will be carried out under certain objective circumstances, and are not limited to a specific time, nor do they require any judgment action during implementation, nor do they imply any other limitations.

[0085] It is understood that some optional features in the embodiments of this application can be implemented independently in certain scenarios without relying on other features, such as the current solution on which they are based, to solve the corresponding technical problems and achieve the corresponding effects. Alternatively, they can be combined with other features as needed in certain scenarios. Correspondingly, the apparatus given in the embodiments of this application can also implement these features or functions, which will not be elaborated here.

[0086] In the embodiments of this application, unless otherwise specified, the same or similar parts between the various embodiments can be referred to each other. In the various embodiments of this application, and in the various implementation methods / methods / implementations within each embodiment, unless otherwise specified or logically conflicting, the terminology and / or descriptions between different embodiments and between the various implementation methods / methods / implementations within each embodiment are consistent and can be mutually referenced. The technical features in different embodiments and the various implementation methods / methods / implementations within each embodiment can be combined according to their inherent logical relationships to form new embodiments, implementation methods, methods, or implementation approaches. The embodiments of this application described below do not constitute a limitation on the scope of protection of this application.

[0087] Typically, trajectory confidence is used to measure the reliability of a trajectory. Specifically, trajectory confidence can be a scalar value that measures the credibility of a perceived trajectory, characterizing its reliability and probability of accuracy. Generally, the range of confidence is (0,1), where a higher confidence value indicates higher confidence and greater accuracy. It should be understood that a trajectory can refer to the movement or motion trajectory of a target object. For example, a trajectory can consist of dots.

[0088] A commonly used method for trajectory confidence analysis is the chi-square test. This method can determine the trajectory confidence by statistically analyzing the deviation between the actual observed values ​​and the theoretical inferred values ​​of the sample, and then determining the trajectory confidence based on the chi-square value of the deviation.

[0089] The sensed object (or target object or target sensed object) typically follows the law of inertial motion during its movement, meaning that the motion state of the sensed object in its historical trajectory (including but not limited to direction, velocity, acceleration, etc.) tends to remain consistent with the motion state of the newly sensed trajectory. Therefore, based on the object's motion state in a known trajectory, the position of the target object in the next time period can be predicted, obtaining a theoretical inference value of the target object's position. Additionally, the actual observation value of the target object is obtained through the positioning output reported by the sensing device in the new time period. Finally, the deviation between the actual observation value and the theoretical inference value is compared. A greater deviation results in a larger chi-square variance and a lower confidence level; a smaller deviation results in a smaller chi-square variance and a higher confidence level; if the two are completely equal (chi-square value 0), the confidence level can be set to the highest value of 1.

[0090] To facilitate understanding, the chi-square test will be introduced below with specific examples.

[0091] In vehicle-mounted sensing scenarios, the motion state of the sensed object, such as vehicle speed, acceleration, and position information, can be acquired through onboard sensor devices. Furthermore, based on the sensing results from the onboard sensor devices, the trajectory confidence level of the vehicle can be obtained using the chi-square test method.

[0092] In related technologies, considering the characteristics of onboard perception devices in autonomous driving scenarios, detection data from vehicle wheel speedometers is introduced. By replacing the low-precision sensor data with high-precision detection results from vehicle mechanical hardware, the confidence analysis accuracy of the chi-square test is improved. This method may include, for example, steps 1 to 5 as follows.

[0093] In step 1, the first measurement position of the positioning component on the target vehicle is determined at a first time, and the second measurement position of the positioning component is determined at a second time, wherein the first time is earlier than the second time.

[0094] In step 2, the first distance information is obtained using the first measurement value position and the second measurement value position.

[0095] In step 3, the inertial measurement information and wheel speed sensor information of the target vehicle from the first moment to the second moment are determined. The wheel speed sensor information may include tire speed, tire angle, and other information.

[0096] In step 4, second distance information is obtained based on inertial measurement information and wheel speedometer information. The vehicle's speed and angle are obtained through wheel speedometer information analysis, which has higher accuracy compared to traditional methods based on onboard sensing devices for analyzing motion status.

[0097] In step 5, the confidence level of the target measurement value at the second time moment is obtained by using the first distance information and the second distance information.

[0098] However, the above method requires access to relevant data from the vehicle's mechanical hardware, so the technology scenario that forcibly binds to vehicle-mounted perception is not suitable for other technology scenarios such as roadside perception.

[0099] As mentioned earlier, the chi-square test is based on the laws of inertial motion. Traditional methods obtain the inertial information of the target object based on its velocity, direction, acceleration, and position. Related technologies, by introducing the heading angle, enrich the dimensions of inertial information and improve the accuracy of the chi-square test.

[0100] Specifically, based on the predicted heading angle of the target object at the predicted time point and the known heading angle corresponding to the last position point in the target's existing trajectory, the difference between the predicted heading angle and the known heading angle can be determined. If this difference is large, it indicates a significant change in the target object's direction of motion. Based on the laws of inertial motion, this situation is unlikely to occur; therefore, the prediction confidence of the prediction result can be considered low. In other words, the difference between the predicted heading angle and the heading angle corresponding to the last position point in the target's existing trajectory is negatively correlated with the prediction confidence of the prediction result. This method may include, for example, steps 1 to 4 as follows.

[0101] In step 1, multiple existing trajectories prior to the target time point and multiple first targets observed at the target time point are correlated and matched to obtain trajectory correlation and matching results.

[0102] In step 2, if it is determined from the trajectory association matching results that there is a target existing trajectory among multiple existing trajectories that is not associated with the first target, prediction is performed based on the target existing trajectory to determine the prediction result of the second target associated with the target existing trajectory before the target time at the prediction time point. The prediction result includes the prediction position point and prediction heading angle of the second target at the prediction time point.

[0103] For example, there are three primary targets and four existing trajectories. The four existing trajectories and the three primary targets are matched and associated. For each primary target, an existing trajectory that meets the association conditions can be found. This yields three association results for the three primary targets. At this point, the existing trajectory s2 that is not associated with any of the three primary targets is determined as the target existing trajectory. The existing target a5 associated with target existing trajectory s2 before the target time point is determined as the secondary target. Then, the prediction result of a5 at the prediction time point is predicted based on the target existing trajectory s2.

[0104] In step 3, the prediction confidence of the prediction result is determined based on the predicted heading angle of the second target at the predicted time point, the heading angle of the last position point in the target's existing trajectory, and the number of predicted position points of the target in the target's existing trajectory; the predicted position point of the target refers to the predicted position point located after the last observed position point in the target's existing trajectory.

[0105] In step 4, if the prediction confidence is greater than or equal to the confidence threshold, the predicted location point of the second target at the prediction time point is associated with the target's existing trajectory.

[0106] The principle behind the above scheme is that the direction of motion of the same target at two adjacent time points (with a short interval) cannot change abruptly. Based on the predicted heading angle of the second target at the predicted time point and the heading angle corresponding to the last position point in the target's existing trajectory, the difference between the predicted heading angle and the heading angle corresponding to the last position point in the target's existing trajectory can be determined. If this difference is large, it indicates a significant change in the direction of motion of the second target. This situation has a low probability of occurrence; therefore, in this case, the prediction confidence is low. In other words, the difference between the predicted heading angle and the heading angle corresponding to the last position point in the target's existing trajectory is negatively correlated with the prediction confidence.

[0107] Related technologies propose determining the prediction confidence level of the prediction result based on the quantity and distribution pattern of the point cloud data from the surrounding sensing results of the target prediction location, thereby improving the accuracy of the chi-square test. The prediction confidence level is negatively correlated with the quantity of point cloud data at the target prediction location, and also negatively correlated with the difference in point cloud distribution. In this scheme, the confidence analysis of the motion trajectory relies on the sensing results in point cloud format reported by the sensing device. Considering that the generation and analysis of the perceived object's trajectory are two relatively independent stages of roadside sensing, the reliance on point cloud data poses a permission risk.

[0108] Trajectory confidence can be applied to various scenarios, such as intelligent driving scenarios and intelligent transportation scenarios. For ease of understanding, the following uses the roadside perception scenario in the intelligent transportation scenario as an example to introduce the applicable scenarios of the embodiments of this application.

[0109] In traditional roadside sensing scenarios, a single type of sensing device is typically used to perceive or detect target objects. This approach suffers from low detection accuracy and high false alarm rates, which may affect the accuracy of post-processing results based on trajectory information.

[0110] Therefore, related technologies utilize multiple sensing devices to perform sensing analysis on the same target object, forming multi-source heterogeneous sensing results (or multimodal sensing results). Specifically, multiple sensing devices can sense the target object separately, each obtaining a set of sensing results. For the target object, the multiple sets of sensing results from various sensing devices are referred to as multi-source heterogeneous sensing results, or multimodal sensing results. Furthermore, the multimodal sensing results can be fused based on trajectory confidence to obtain a high-precision final sensing trajectory (i.e., trajectory fusion result).

[0111] Figure 1 This is a schematic diagram of a roadside sensing scenario. Figure 1 It may include a roadside sensing device 110 and a sensing engine 120.

[0112] The roadside sensing device 110 may include cameras, radar, and sensors, etc., for detecting and sensing traffic participants on the road and detecting and sensing road conditions. In some embodiments, the roadside sensing radar can process the detection and sensing results to obtain the motion trajectory analysis results of the target object. For example, the roadside sensing radar can complete the motion trajectory analysis of the target object through point cloud processing and trajectory processing.

[0113] The roadside sensing device 110 can transmit sensing results to the sensing engine 120 in real time or periodically. Based on centralized control of multiple sensors, the sensing engine 120 can realize core roadside sensing capabilities such as radar-visual sensing data fitting, cross-pole data fitting, and cross-station trajectory stitching. It serves as the foundation of the overall smart highway solution, providing basic sensing data to support vehicle-side and cloud-based applications. Specifically, radar-visual sensing data fitting can refer to the fitting of radar sensing data and video sensing data, and / or the fitting of radar sensing data with radar sensing data. Additionally, the sensing engine 120 can also be used for blind spot fitting under poles. Since roadside sensing devices are typically deployed on poles, blind spot fitting under poles is equivalent to blind spot fitting for the sensing devices.

[0114] After acquiring the perception trajectory of the target object, the perception engine 120 can perform confidence analysis on the perception trajectory and filter out false alarms based on the confidence analysis results. At the same time, the confidence can be used to support the subsequent fusion step of multi-source heterogeneous perception trajectories.

[0115] Figure 2 This is a schematic diagram of the roadside sensing system. Figure 2 The perception system shown may include a perception engine 120, a cloud platform, and a vehicle-mounted system. The roadside perception process may involve interactions between the roadside (including the perception engine), the cloud platform, and the vehicle-mounted system.

[0116] The perception engine 120 may include a vision processing unit 121, a radar access unit 122, a time-domain confidence analysis unit 123, a radar-visual perception data fitting unit 124, a multi-station collaboration unit 125, an event recognition unit 126, a map processing unit 127, and a message routing unit 128, etc.

[0117] The external inputs of the perception engine 120 may include camera perception results and radar perception results. Camera perception results can be transmitted to the vision processing unit 121 via a standard interface, and radar perception results can be transmitted to the radar access unit 122 via a proprietary interface.

[0118] The external input to the perception engine 120 can also include map information. The map information can be transmitted to the perception engine 120 from a cloud server via a standard interface.

[0119] It should be understood that the above interface types are provided as examples only, and the above interface types may change in future standards. For example, if private interfaces are standardized, this application does not limit them.

[0120] The vision processing unit 121 can process the camera perception results to obtain the trajectory information of the target object perceived by the camera.

[0121] The radar access unit 122 can process the radar sensing results to obtain the trajectory information of the target object sensed by the radar. For example, trajectory preprocessing methods for radar sensing results include clustering algorithms for sensed point cloud data, Kalman filtering algorithms based on inertial motion laws, and time-series multi-object matching algorithms based on the Hungarian algorithm.

[0122] The vision processing unit 121 can input the trajectory information of the target object it perceives into the temporal confidence analysis unit 123, and the radar access unit 122 can input the trajectory information of the target object it perceives into the temporal confidence analysis unit 123. The temporal confidence analysis unit 123 can determine the confidence level of multiple trajectory information of the target object respectively, which is used to support post-processing operations such as trajectory fusion.

[0123] The time-domain confidence analysis unit 123 can determine the confidence level of different trajectories using the chi-square test method mentioned above. In this case, the perception engine can obtain motion state information of the target object from the vehicle, such as inertial information, or it can obtain motion state information of the target object based on the perception results of the camera or radar. Since the chi-square test method only considers the motion trajectory of the target object itself within the historical time interval, that is, it only considers the motion law of a single object in time sequence, the confidence level determined by the chi-square test method can be called the time-domain confidence level.

[0124] The radar-visual perception data fitting unit 124 can fuse the perception results of different sensing devices to obtain the final trajectory information of the target object. For example, the radar-visual perception data fitting unit 124 can perform trajectory fusion based on the trajectory confidence level output by the confidence analysis module, which helps improve the accuracy and efficiency of trajectory post-processing. For example, the radar-visual perception data fitting unit 124 may include a video + video fitting unit (i.e., C+C), a radar + radar fitting unit (i.e., R+R), and a video + radar fitting unit (i.e., C+R).

[0125] The trajectory information of the target object obtained through the above steps is only within a certain area, such as the trajectory information of a target object at a single station. To obtain the complete trajectory information of the target object, the multi-station collaboration unit 125 can acquire the trajectory information of the target object at other stations. For example, the multi-station collaboration unit 125 can acquire lane information from the map processing unit 127, which can be used, for example, for fitting the trajectory information of target objects at different stations.

[0126] In addition, the multi-station coordination unit 125 can transmit the traffic statistics results of the sensing data and the relevant information of the target object to the event recognition unit 126 for the purpose of recognizing traffic events.

[0127] The message routing unit 128 can be used to transmit the processing results in the perception engine to other units or systems.

[0128] As described above, the relevant technologies primarily rely on the chi-square test to determine trajectory confidence. While simple to implement, the chi-square test has limitations. Firstly, it only considers the regularity of the target object's motion, neglecting the influence of other factors on its behavior. Secondly, the inertial prior assumptions upon which the chi-square test is based are not applicable to all traffic and driving scenarios. For instance, the chi-square test assumes that the target object maintains uniform linear motion during its movement, but in traffic scenarios, vehicles may change lanes, turn, accelerate, or decelerate, thus failing to accurately identify the confidence level of trajectories involving lane changes, turns, or acceleration / deceleration. Thirdly, while the chi-square test assesses the confidence of newly generated points based on a known set of points, effectively analyzing anomalous points in a single frame and eliminating abnormal measurements, it is ineffective for trajectories exhibiting erroneous trends, such as sensor drift or abnormal point addition.

[0129] To address one or more of the aforementioned problems, embodiments of this application provide a method for determining trajectory confidence. Compared to the chi-square test method for determining trajectory confidence, embodiments of this application consider the spatial interaction between the target object and the environment, increasing the analytical dimension of trajectory confidence and helping to improve the accuracy of trajectory confidence.

[0130] The following is combined Figure 3 This application describes a method for determining trajectory confidence based on embodiments thereof. Exemplarily, the target object can be a traffic participant, such as a vehicle. For ease of understanding, the method provided in this application embodiment will be described below using a vehicle as the target object.

[0131] Figure 3 The method shown may include steps S310 to S330.

[0132] S310, acquire the first trajectory information of the target object. The target object can also be called the sensing object, or the target sensing object.

[0133] The aforementioned first trajectory information can be obtained by sensing the target object through sensing devices. These sensing devices may include, for example, roadside sensing devices. Roadside sensing devices are a key component of intelligent transportation systems; they utilize sensors and communication equipment deployed around roads to achieve real-time monitoring and data analysis of road traffic conditions. Sensing devices include, but are not limited to, cameras, millimeter-wave radar, and lidar, which can detect and identify traffic participants such as vehicles, pedestrians, and cyclists, as well as infrastructure such as road signs and traffic signals.

[0134] In some embodiments, by collecting the motion state of a target object through a sensing device, point traces or point cloud data of the target object can be obtained, thereby obtaining the trajectory information of the target object, such as first trajectory information. Exemplarily, preprocessing the point traces or point cloud data of the target object can obtain the trajectory information of the target object. Trajectory preprocessing methods for radar sensing results typically include clustering algorithms for sensing point cloud data, Kalman filtering algorithms based on inertial motion laws, and time-series multi-object matching algorithms based on the Hungarian algorithm, etc. This application does not limit this approach.

[0135] In some embodiments, the first trajectory information can be trajectory information generated when the target object is driving in the first environment. The first environment can be the driving environment or driving environment of the target object. When the target object is driving in the first environment, the trajectory information obtained by the sensing device from sensing the target object is the first trajectory information.

[0136] In some embodiments, the first environment can be the environment surrounding the target object, such as objects that have spatial interaction with the target object during its movement. For example, the first environment may include road-related information, information related to other objects around the target object, geographical information about the road's surroundings, weather information, etc.

[0137] As an example, road-related information can include information about road infrastructure, such as lane markings, guardrails, gantries, signs, and sensing devices. As another example, information about other objects surrounding the target object can include their distribution, location, speed, and changes in motion. As yet another example, the geographical environment surrounding the road can include one or more of the following: flat areas, mountains, plains, plateaus, or cliffs. As yet another example, weather information can include one or more of the following: rainy, snowy, windy, foggy, sunny, cloudy, or overcast.

[0138] The first environment will be described in detail later, so for the sake of brevity, it will not be repeated here.

[0139] S320: Obtain the first relationship between the target object and the first environment.

[0140] The aforementioned first relationship refers to the spatial interaction relationship between the target object and the first environment. For example, the first relationship can include vehicle-to-vehicle (V2V) interaction relationships and vehicle-to-infrastructure (V2I) interaction relationships. V2I can also refer to the interaction relationship between a vehicle (target object) and its surrounding static environment, such as the interaction relationship between the target object and road infrastructure, or the interaction relationship between the target object and the surrounding geographical environment. V2V interaction relationships can also refer to the interaction relationship between a vehicle (target object) and surrounding vehicles, or the interaction relationship between a vehicle and its dynamic environment.

[0141] The first relation will be explained in detail later, so for the sake of brevity, it will not be repeated here.

[0142] In some embodiments, the first environment may be obtained before obtaining the first relationship.

[0143] For example, a first environment, such as the position of lane lines, the position of road guardrails, and the geographical environment surrounding the road, can be obtained based on map information associated with the first trajectory information. The map information associated with the first trajectory information includes map information of the road where the first trajectory information is located. Obtaining the first environment information based on map information helps reduce the complexity and processing overhead of obtaining the first environment.

[0144] For example, a first environment can be obtained based on the perception results of a sensing device. For instance, the location of lane lines, road guardrails, and signs can be obtained based on the perception results. Alternatively, the first environment can be determined based on the perception results of other objects around the target object, such as their movement speed, trajectory information, and location. Obtaining the first environment based on the perception results of a sensing device helps improve the accuracy of the obtained first environment.

[0145] For example, the first environment can be determined based on pre-configured system information, such as at least one of the following: the location of the gantry, the location of the sign, the location of the sensing device, and the sensing range of the sensing device. Since this information is typically fixed, it can be pre-stored in the sensing system, helping to reduce the overhead of acquiring this information. When this information changes, such as when new signs or sensing devices are added to the road, the information of the first environment can be updated to ensure that the first environment information in the system is updated in a timely manner.

[0146] As mentioned earlier, by collecting the motion state of the target object through sensing devices, we can obtain the point traces or point cloud data of the target object, and then obtain the trajectory information of the target object, such as the first trajectory information.

[0147] Based on this, in some embodiments, the relationship between the target object and the first environment (referred to as the first relationship) or the interaction relationship between the target object and the first environment can be obtained based on the first trajectory information and the first environment. For example, information such as the target object's position, speed, and motion state can be obtained based on the first trajectory information, and then the first relationship can be determined based on this information and the first environment. Since the first trajectory information is a processed perception result and has high reliability, determining the first relationship based on the first trajectory information and the first environment helps improve the reliability of the obtained first relationship.

[0148] In other embodiments, the first relationship can be obtained based on point cloud data or point trace data and a first environment. In this case, the point cloud data can be obtained when the detection result of the sensing device is obtained, thereby obtaining the first relationship. At the same time, the trajectory information of the target object can be obtained based on the detection result of the sensing device. That is, the first relationship can be obtained before obtaining the first trajectory information, thereby reducing the latency of obtaining the confidence level.

[0149] S330, based on the first relationship, determines the spatial confidence level of the first trajectory information.

[0150] In some embodiments, the spatial confidence level of the first trajectory information can be determined based on the probability of the first relationship occurring. Since the spatial interaction relationship between the target object and its surrounding environment is considered in the embodiments of this application, the trajectory confidence level determined by the method provided in the embodiments of this application can be referred to as spatial confidence level or spatial confidence level.

[0151] For example, the probability of the first relationship occurring is determined, and then the spatial confidence of the first trajectory information is determined based on the characteristics of the driving scenario, such as general driving habits or prior rules, and the probability of the first relationship occurring.

[0152] Taking the first relationship as the relationship between the target object and the lane line as an example, the first relationship can include the target object driving on the lane line or the target object driving along the lane center line. In other words, the first relationship can include the distance between the target object and the lane line being 0, or the distance between the target object and the lane line being half the lane width.

[0153] Based on common driving habits, objects typically travel along the lane centerline. This means the probability of an object traveling along the lane centerline is relatively high, while the probability of it crossing the line is relatively low. In other words, the probability of the distance between the object and the lane line being zero is low, while the probability of the distance between the object and the lane line being half the lane width is relatively high.

[0154] Based on this, if the first relation includes a distance of 0 between the target object and the lane line, then the spatial confidence of the first trajectory information is low; if the first relation includes a distance of half the lane width between the target object and the lane line, then the spatial confidence of the first trajectory information is high.

[0155] Alternatively, if the first relation includes the distance X between the target object and the lane centerline, then the smaller X is, the higher the spatial confidence of the first trajectory information; conversely, the larger X is, the lower the spatial confidence of the first trajectory information. The value of X can range from 0 to half the lane width.

[0156] It should be noted that the lane lines involved in the first relationship refer to the lane lines of the lane in which the target object is located. When the target object is driving on the lane lines between the first and second lanes, the lane lines involved in the first relationship can be considered to be either the lane lines of the first lane or the lane lines of the second lane.

[0157] For example, the probability of the first relationship occurring can be obtained through actual testing. For instance, the probability of the first relationship occurring can be statistically obtained based on the interaction between the actual running trajectories of multiple objects and the driving environment. The actual running trajectories of the multiple objects can be obtained through manual inspection or high-precision inspection equipment such as drones. This application does not limit this.

[0158] In some embodiments, the relationship between the first relationship and the confidence level of the first trajectory information can be obtained through technologies such as artificial intelligence and machine learning. For example, the relationship between the first relationship and the spatial confidence level can be analyzed and modeled, such as by using neural network training and feature extraction, to obtain a second model. The input to the second model can be the first relationship, and the output of the second model can be the confidence level of the first trajectory information.

[0159] If there are multiple first relationships, or in other words, multiple spatial interaction relationships between the target object and its surrounding environment, then the initial spatial confidence level corresponding to each of the multiple first relationships can be obtained firstly. Then, the spatial confidence level of the first trajectory information can be determined based on these multiple initial spatial confidence levels. For example, when determining the spatial confidence level of the first trajectory information, multiple initial spatial confidence levels correspond one-to-one with multiple weights. For instance, the spatial confidence level of the first trajectory information can be a weighted sum or a weighted average of multiple initial spatial confidence levels. The weights corresponding to the initial spatial confidence levels can be determined based on information such as the importance and reliability of their corresponding first relationships.

[0160] It should be understood that the methods provided in this application can be implemented alone or in combination with temporal confidence (such as the chi-square test) mentioned in related technologies. For example, post-processing operations related to trajectory information, such as guiding trajectory fusion, can be performed based solely on spatial confidence, or post-processing operations related to trajectory information can be performed based on both spatial and temporal confidence. Taking guided trajectory fusion as an example, spatial and temporal confidence can correspond to different weights. The weighted sum or weighted average of the spatial and temporal confidence can be used as the final trajectory confidence to guide trajectory fusion. In this way, the weights of spatial and / or temporal confidence can be determined according to different scenario characteristics, which helps to flexibly adapt to different scenarios. As an example, the final trajectory confidence can be determined based on the average of spatial and temporal confidence, which is simple to implement and has low computational cost.

[0161] In this embodiment, the spatial confidence level of the target object's trajectory information is determined by introducing the spatial interaction relationship between the target object and its surrounding environment. This approach considers the potential influence of the surrounding environment on the target object's motion behavior, thus improving the accuracy of the trajectory confidence level. Furthermore, this approach is applicable to various traffic and driving scenarios, including those conforming to and those not conforming to the inertial prior assumption, thereby enhancing the universality and robustness of the trajectory confidence level.

[0162] This is because the environment of the target object (which can be called the target object's driving environment) may affect the target object's trajectory, and this effect may cause the target object's motion law to deviate from the inertial prior assumptions. For example, when other objects in front of the target object suddenly decelerate, the target object may also suddenly decelerate or make an emergency lane change. When the target object suddenly decelerates, its acceleration changes significantly; when the target object makes an emergency lane change, its heading angle, or steering angle, changes significantly. In this case, if the confidence level of the target object's trajectory is determined using the chi-square test, the confidence value may be low and inconsistent with the actual situation of the target object. If the interaction between the target object and its environment is considered when determining the confidence level of the target object's trajectory, such as considering the velocity changes of other objects around the target object, then the confidence value of the trajectory may be higher, which helps to improve the accuracy of the trajectory confidence level.

[0163] Furthermore, in cases of sensor drift, the perceived target object may deviate to the left or right by a certain value. When the target object is moving at a constant speed, the confidence value of the perception result determined by the chi-square test may be high, but the accuracy of the confidence level may be low. If the spatial interaction between the target object and the environment is considered, such as the interaction between the target object and the lane lines, the confidence level of the device's perception result may be even lower. This is because when the perceived target object deviates to the left or right by a certain value, the interaction between the target object and the lane lines may be that the target object's trajectory mostly coincides with the lane lines (i.e., the target object drives on the lane lines for a long time), or the target object's trajectory deviates from the center of the lane lines for a long time. However, during normal driving, the target object is more likely to drive along the center of the lane lines, and the probability of driving on the lane lines for a long time is low. Therefore, based on the above-mentioned interaction between the target object and the lane lines, it can be determined that the confidence level of the target object's trajectory is low, which helps to improve the accuracy and robustness of the trajectory confidence level, effectively identify sensor drift, and eliminate abnormal measurement values.

[0164] The first environment and the first relationship will be described in detail below.

[0165] As mentioned earlier, the first environment can include information about road infrastructure, such as lane lines, road guardrails, gantries, signs, and sensing equipment.

[0166] For example, the first environment may include the location of lane lines. Lane lines are typically used for road separation, traffic guidance, and traffic restrictions. Lane lines include various styles, such as solid lines, dashed lines, double solid lines, double dashed lines, and solid-dashed lines. Optionally, the first environment may include the location of one or more of the aforementioned lane line styles, and may also include the type of lane line (or lane line style).

[0167] For example, the first environment may include the location of a road guardrail. Road guardrails can be used to separate lanes traveling in different directions to guide traffic.

[0168] For example, the first environment may include the location of the gantry and / or the type of gantry. Gantries can be categorized according to their purpose, such as traffic monitoring gantries and road traffic sign gantries. For example, a gantry can be used to install sensing devices to monitor traffic participants, such as sensing the type of traffic participant and their speed, movement status, trajectory, etc. For example, a gantry can be used to install signs to achieve functions such as traffic guidance and traffic information indication.

[0169] For example, the first environment may include the location of signs (or road signs). Road signs can provide important traffic information to road users through elements such as graphics, symbols, colors, and text, ensuring traffic order and safety.

[0170] In some embodiments, the first environment may include information related to the sensing device.

[0171] For example, the first environment may include the deployment location of the sensing device, such as the location of the mounting pole of the sensing device.

[0172] For example, the first environment may include the sensing range of the sensing device. The sensing range (also called the sensing area) refers to the maximum distance and / or angular range that a sensing device (such as a camera, millimeter-wave radar, lidar, etc.) can detect a target object under normal operating conditions. For example, the sensing range may include a vertical sensing range and / or a horizontal sensing range. The horizontal sensing range typically refers to the lane width (or number of lanes) and / or lane length that the sensing device can sense. The vertical sensing range typically refers to the angular range that the sensing device can cover in the vertical direction to identify different traffic participants at different heights. In some cases, the sensing range may also include the sensing frequency, or the output frequency of the sensing results. Taking a camera as an example, the sensing frequency may refer to the time interval between camera shots.

[0173] In other words, the first environment may include one or more of the vertical sensing range, horizontal sensing range, or sensing frequency of the sensing device.

[0174] As mentioned earlier, the first environment may include information related to other objects surrounding the target object. These other objects may include other traffic participants in the vicinity, such as pedestrians and vehicles. For example, the first environment may include one or more of the following: the speed of movement of other objects surrounding the target object; trajectory information of other objects; direction of movement of other objects; or the location of other objects.

[0175] As mentioned earlier, the first environment can include the geographical environment, such as the geographical environment around the road the target object is traveling on. For example, the geographical environment can include one or more of flat areas, mountains, plains, plateaus, or cliffs.

[0176] As mentioned earlier, the first environment can include weather information, such as the weather information of the area where the target object's driving road is located. For example, the weather information can include one or more of the following: rainy, snowy, windy, foggy, sunny, cloudy, and overcast.

[0177] It should be understood that the first environment may include one or more of the information described above, and this application does not limit it.

[0178] It should be understood that the first environment may include objects or information that have a spatial interaction relationship with the target object. The information included in the first environment is only given as an example and is not limited in this application. For example, the first environment may include the infrastructure involved in the future road, the sensing devices deployed in the future road, the lane line patterns adopted in the future road, etc.

[0179] The first environment has been described in detail above. The first relation and the relationship between the first relation and trajectory confidence will be described in detail below.

[0180] In some embodiments, the first relationship may include the interaction relationship between the target object and the infrastructure in the road.

[0181] For example, the first relationship may include the relationship between the position of the target object and the position of the lane lines. For instance, the first relationship may include the distance between the target object and the lane lines, the distance between the target object and the lane centerline, etc. Alternatively, the first relationship may include the distribution relationship between the target object and the lane centerline, such as the target object traveling along the lane centerline and the target object not traveling along the lane centerline (or the target object traveling on the lane lines).

[0182] Considering that vehicles generally tend to travel along the center line of their lane and do not drive on the lane edges for extended periods, the confidence level of the first trajectory information is low if the first relation indicates that the target object is not traveling along the lane center line, and high if the first relation indicates that the target object is traveling along the lane center line.

[0183] Alternatively, if the first relation indicates that the distance between the target object and the lane line is small, the confidence level of the first trajectory information is low; if the first relation indicates that the distance between the target object and the lane line is large (and less than or equal to half the width of the lane line), the confidence level of the first trajectory information is high.

[0184] Alternatively, if the first relationship indicates that the distance between the target object and the center line of the lane is small, the confidence level of the first trajectory information is high; if the first relationship indicates that the distance between the target object and the center line of the lane is large, the confidence level of the first trajectory information is low.

[0185] Figure 4 An example diagram illustrating the distribution relationship between the target object and the lane centerline provided in this application embodiment. See also... Figure 4 The spatial interaction between vehicle 1 and the environment can include the trajectory information of vehicle 1 overlapping with the lane lines, or in other words, vehicle 1 not traveling along the lane centerline. The spatial interaction between vehicles 2 to 4 and the environment can include the trajectory information of vehicles 2 to 4 substantially overlapping with the lane centerline, or in other words, vehicles 2 to 4 traveling along the lane centerline.

[0186] As can be seen from the above introduction, the confidence level of the perceived trajectory of vehicle 1 is low, while the confidence level of the perceived trajectories of vehicles 2 to 4 is high.

[0187] For example, the first relation may include the type of lane the target object is in, such as an overtaking lane, a ramp, or other lanes. This is because, during vehicle operation, the speeds of vehicles traveling in different types of lanes exhibit different distribution patterns. For instance, vehicles typically travel faster in the overtaking lane, slower on ramps, and their speeds in other lanes generally fall between the ramp and overtaking lane speeds. Therefore, if the first relation indicates that the target object is in the overtaking lane, and the first trajectory information indicates that the target object's speed is slow, the confidence level of the first trajectory information is considered low; conversely, the confidence level is high. If the first relation indicates that the target object is in the ramp, and the first trajectory information indicates that the target object's speed is fast, the confidence level of the first trajectory information is considered low; conversely, the confidence level is high. If the first relation indicates that the target object is in another lane, and the first trajectory information indicates that the target object's speed is greater than the average speed of the overtaking lane or less than the average speed of the ramp, the confidence level of the first trajectory information is low; conversely, the confidence level is high.

[0188] For example, the first relationship may include the relationship between the target object and the sensing range of the sensing device. For example, the first relationship may include the relationship between the start position and / or end position of the first trajectory information within the sensing range of the sensing device and the sensing range itself.

[0189] Since the sensing range of a sensing device has boundaries, for a real vehicle, it should be quickly detected when entering the sensing range. The starting position of the sensing trajectory should be concentrated near the sensing boundary of the sensing device, or in other words, the starting position of the sensing trajectory within the sensing range should be near the boundary of the sensing range. Similarly, the ending position of the sensing trajectory should also be concentrated near the sensing boundary of the sensing device, or in other words, the ending position of the sensing trajectory within the sensing range should be near the boundary of the sensing range.

[0190] For example, if the distance between the starting and / or ending positions of the first trajectory information within the aforementioned sensing range and the boundary of the sensing range is less than or equal to a certain threshold, then the starting and / or ending positions of the first trajectory information within the sensing range are considered to be located within the boundary region of the sensing range; otherwise, the starting and / or ending positions of the first trajectory information within the sensing range of the sensing device are considered not to be located within the boundary region of the sensing range. The starting / ending positions of the sensing trajectory can be referred to as the track start / end positions.

[0191] Therefore, if the first relationship indicates that the starting position and / or ending position of the first trajectory information within the sensing range of the sensing device is located in the boundary area of ​​the sensing range, the confidence level of the first trajectory information is high; if the first relationship indicates that the starting position and / or ending position of the first trajectory information within the sensing range of the sensing device is not located in the boundary area of ​​the sensing range, the confidence level of the first trajectory information is low.

[0192] Figure 5 An example diagram illustrating the relationship between the target object and the sensing range of the sensing device, provided in an embodiment of this application. See also... Figure 5 The spatial interaction between vehicle 1 and the environment can include the fact that the starting position of vehicle 1's trajectory is not located in the boundary area of ​​the perception range, that is, the starting position of vehicle 1's trajectory is far from the boundary of the perception range. The spatial interaction between vehicles 2 to 4 and the environment can include the fact that the starting positions of vehicles 2 to 4's trajectories are located in the boundary area of ​​the perception range, that is, vehicles 2 to 4 are close to the boundary of the perception range.

[0193] As can be seen from the above introduction, the confidence level of the perceived trajectory of vehicle 1 is low, while the confidence level of the perceived trajectories of vehicles 2 to 4 is high.

[0194] For example, the first relationship may include the relationship between the position of the target object and the position of the sensing device, such as the distance between the target object and the sensing device. Generally, the accuracy of the sensing result is relatively low in areas far from the sensing device (such as the edge area of ​​the sensing area), and relatively high in areas close to the sensing device (such as the center area of ​​the sensing area). Based on this, if the target object is close to the sensing device, the confidence level of the first trajectory information is considered high; if the target object is far from the sensing device, the confidence level of the first trajectory information is considered low. Furthermore, the confidence level of the first trajectory information is low in the blind zone under the sensing device.

[0195] For example, the first relationship may include the relationship between the position of the target object and the position of the gantry, or the first relationship may include the distance between the target object and the gantry.

[0196] For example, the first relationship may include the relationship between the position of the target object and the position of the sign, or in other words, the first relationship may include the distance between the target object and the sign. The applicant discovered through actual testing that the reflections generated by the sign may interfere with the sensing device.

[0197] For example, the first relation may include the relationship between the location of the target object and the location of the road guardrail. For instance, the first relation may include the distance between the location of the target object and the location of the road guardrail. Considering that the target object typically maintains a certain distance from the road guardrail while traveling on the road, or in other words, the target object usually does not travel close to the road guardrail, therefore, if the first relation indicates a small distance between the location of the target object and the location of the road guardrail, the confidence level of the first trajectory information is low; conversely, if the first relation indicates a small distance between the location of the target object and the location of the road guardrail, the confidence level of the first trajectory information is high.

[0198] In some embodiments, the first relationship may include the interaction relationship between the target object and other surrounding objects.

[0199] For example, the first relation may include the distribution of other objects around the target object, such as dense (congested) or sparse (smooth) traffic. Generally, in congested traffic, the speed of the target object is similar to the speed of other objects around it, or the speed difference between the target object and other objects is small. Based on this, in congested traffic, if the first relation indicates that the speed difference between the target object and other objects is small, such as less than a preset threshold, then the confidence level of the first trajectory information is high; if the first relation indicates that the speed difference between the target object and other objects is large, such as greater than the preset threshold, then the confidence level of the first trajectory information is low.

[0200] For example, the first relationship may include the distance between the target object and other surrounding objects, such as front-to-back distance, left-to-right distance, etc. When vehicles are traveling on the road, they generally tend to maintain a safe distance from vehicles in front and behind them in the same lane, and a safe distance from vehicles to the left and right in adjacent lanes. Based on this, if the first relationship indicates that the distance between the target object and other surrounding objects is less than the safe distance (such as a preset safe distance value), then the confidence level of the first trajectory information is low; if the first relationship indicates that the distance between the target object and other surrounding objects is greater than the safe distance, then the confidence level of the first trajectory information is high.

[0201] Figure 6 An example diagram illustrating the distribution relationship between the target object and other surrounding objects, provided in an embodiment of this application. See also... Figure 6 The interaction between vehicle 2 and the environment may include the distance between vehicle 2 and vehicle 1 being less than the safe distance. The distance between vehicles 3 to 5 and the vehicles in front and behind in the same lane is greater than the safe distance.

[0202] As can be seen from the above introduction, the confidence levels of the perceived trajectories of vehicles 1 and 2 are low, while the confidence levels of the perceived trajectories of vehicles 3 to 5 are high.

[0203] For example, the first relation may include the difference between the speed of the target object and the speeds of other surrounding objects. During vehicle movement, the speed of the target object and the speeds of neighboring vehicles tend to converge; for example, all vehicles travel slowly during traffic jams and quickly when the road is clear. Therefore, in traffic jams, if the first relation indicates that the target object is traveling fast, or the speed difference between the target object and other surrounding objects is large, the confidence level of the first trajectory information is low; conversely, the confidence level of the first trajectory information is high. Similarly, in clear road conditions, if the first relation indicates that the target object is traveling fast, the confidence level of the first trajectory information is high; conversely, the confidence level of the first trajectory information is low.

[0204] Figure 7 An example diagram illustrating the velocity difference distribution between the target object and other surrounding objects, provided in an embodiment of this application. See also... Figure 7 The speeds of vehicles 2 to 4 are 70 km / h, 75 km / h, and 80 km / h, respectively.

[0205] As can be seen from the above introduction, if the speed of vehicle 1 is between 70km / h and 80km / h, the confidence level of the trajectory information of vehicle 1 is high; if the speed of vehicle 1 is 100km / h, the confidence level of the trajectory information of vehicle 1 is low.

[0206] For example, the first relationship may include the movement directions of other objects around the target object, or the relationship between the movement direction of the target object and the movement directions of other objects around it. If the movement direction of other objects is towards the lane where the target object is located, or if other objects have a tendency to change lanes into the lane where the target object is located, or if the movement direction of the target object conflicts with the movement direction of other objects around it, then the target object will usually slow down or change lanes. Based on this, if the first relationship indicates that the target object slows down or suddenly changes lanes when other objects have a tendency to change lanes into the lane where the target object is located, then the confidence level of the first trajectory information is high; if the first relationship indicates that the target object does not slow down or change lanes, then the confidence level of the first trajectory information is low.

[0207] For example, the first relationship may include the geographical environment in which the target object is driving, which can be one or more of the following: flat areas, mountains, plains, plateaus, or cliffs. Generally, when a vehicle is driving on a road, if the driving environment is flat (i.e., the road has a small gradient or no sharp turns or other adverse driving conditions), the vehicle speed is generally higher; if the driving environment is mountainous (i.e., the road has a large gradient or many sharp turns or other adverse driving conditions), the vehicle speed is generally lower; if the driving environment is plains (i.e., the altitude is low and the vehicle's engine performance is normal), the vehicle speed is generally higher; if the driving environment is plateaus (i.e., the altitude is high and the vehicle's engine performance may be affected by sea waves), the vehicle speed may be lower than the speed when driving on plains. If the first trajectory information indicates that the speed of the target object conforms to the driving characteristics of the above geographical environment, then the confidence level of the first trajectory information is high; otherwise, the confidence level of the first trajectory information is low.

[0208] For example, the first relation may include the target object's driving weather as one or more of the following: rain, snow, strong winds, dense fog, sunny, cloudy, and overcast. Generally, when a vehicle is driving on the road, if it encounters rain or snow, the road is slippery, so the vehicle speed is generally lower, and the distance between the vehicle and other surrounding objects may be larger to ensure a safe braking distance; if it encounters dense fog, visibility is low, so the vehicle speed is generally lower, and the distance between the vehicle and other surrounding objects may be larger to ensure a safe braking distance; if it encounters strong winds, the vehicle speed is generally lower; while in sunny, cloudy, or overcast weather, the vehicle speed is generally higher. Based on this, if the first trajectory information matches the driving characteristics of the above weather, the confidence level of the first trajectory information is high; otherwise, the confidence level of the first trajectory information is low.

[0209] There are several ways to determine the primary relationship. For example, the interaction relationship between the target object and its surrounding environment can be obtained for multiple time periods, i.e., the primary relationship. For example, the interaction relationship between the target object and its surrounding environment can be obtained for different lanes. In other words, the spatial confidence of the primary trajectory information can be analyzed for different time periods, or the spatial confidence of the primary trajectory information can be analyzed for target objects in different lanes.

[0210] As mentioned earlier, the spatial confidence level of the first trajectory information can be determined based on the probability of the first relationship occurring. In some embodiments, the probability of the first relationship occurring can be obtained based on a first model, thereby determining the spatial confidence level of the first trajectory information based on the probability of the first relationship occurring. The first model can be used to characterize the relationship between different scenarios of the first relationship and their corresponding probabilities of occurrence. For example, if the first relationship is the distance between a target object and a lane line, then the first model can be used to characterize the probability of different distance values ​​occurring between the target object and the lane line.

[0211] Before determining the spatial confidence level of the first trajectory information, the first model can be obtained first.

[0212] In some embodiments, statistical relationships between vehicles and their surroundings, and the probabilities of these relationships occurring, can be collected in a test environment to obtain a first model. For example, statistical collection of interactions between multiple objects and their environment can be performed to improve the generality of the data.

[0213] For example, second trajectory information of multiple objects can be obtained. This second trajectory information is the trajectory information generated by the multiple objects moving in the test environment. The second trajectory information is a reliable trajectory, which can be obtained through manual measurement or through highly reliable sensing devices such as drones. It should be understood that the aforementioned multiple objects may or may not include the target objects mentioned above; this application does not limit this.

[0214] The accuracy of the perceived trajectory needs to meet certain requirements to help ensure the accuracy of the trajectory confidence analysis results. Therefore, a preprocessing method is required to obtain the second trajectory information. For example, different preprocessing methods can be selected based on the principles of the sensing hardware. Common trajectory preprocessing methods for radar sensing results include clustering algorithms for sensing point cloud data, Kalman filtering algorithms based on inertial motion laws, and time-series multi-object matching algorithms based on the Hungarian algorithm. This application does not limit the scope of these methods.

[0215] For example, the test environment can be a first environment. In this way, the probability of the first relationship obtained through the first model matching the first environment is higher, and the output accuracy of the first model is higher. From the perspective of the sensing device, sensing analysis module, or sensing system, the first environment can also be called the operating environment of the sensing device, sensing analysis module, or sensing system. High consistency between the test environment and the operating environment helps improve the accuracy of the obtained trajectory spatial confidence.

[0216] Based on the second trajectory information and the test environment, a second relationship between multiple objects and the test environment can be determined.

[0217] For example, the interaction between the target object and its surrounding static environment can be calculated over various time periods. The static environment referred to here broadly encompasses all static factors that may affect the vehicle's trajectory during operation, including but not limited to those mentioned above: lane markings, the sensing range of the sensing device, road guardrails, and the geographical environment. It should be understood that static environment information requires additional file input as support (e.g., high-precision map information for highways).

[0218] For example, the interaction between the target object and its surrounding dynamic environment can be calculated within each time period. The dynamic environment mentioned here refers to all motion factors that may affect the vehicle's trajectory during its movement, including but not limited to those mentioned above: motor vehicles, non-motorized two-wheeled vehicles, pedestrians, etc.

[0219] The method for determining the second relation is similar to the method for determining the first relation mentioned earlier. For details not described in detail, please refer to the previous section on the first relation. For the sake of brevity, it will not be elaborated upon here.

[0220] In some embodiments, the second relation described above can be stored in a storage device. For example, the establishment of the first model can be triggered when the amount of data in the stored second relation reaches a certain threshold. Furthermore, after obtaining the initial first model, the update of the first model can be triggered when the amount of data in the newly added second relation reaches a threshold. This is because, when the amount of data in the second relation is small, the accuracy of the first model obtained based on the second relation may be low. Additionally, frequent modeling or model updates can incur significant overhead.

[0221] For example, the first model can be obtained based on the statistical results of the second relationship, or by fitting a probability distribution to the probability of the second relationship occurring. For instance, probability distribution models such as Gaussian, Poisson, and Gamma distributions can be used to fit the probability distribution of the second relationship occurring. The above probability distribution models are merely illustrative and are not intended to limit the scope of this application.

[0222] Figure 8 This is a flowchart illustrating the method for establishing a first model provided in an embodiment of this application. Figure 8 The method shown includes steps S810 to S850.

[0223] S810 determines the trajectory information of the target vehicle, i.e., the second trajectory information, based on the data collected by the roadside sensing device. The target vehicle mentioned here can be one of the multiple objects mentioned above.

[0224] S820 determines the interaction information between the target vehicle and the surrounding roads based on the high-precision map of the surrounding area configured inside the sensing device.

[0225] S830 extracts the interaction information between the target vehicle and surrounding vehicles based on the trajectory information of other nearby vehicles collected by the sensing device within the same time period.

[0226] It should be understood that the interaction information obtained in steps S820 and S830 is the second relationship mentioned above.

[0227] S840 stores the interaction information between the target vehicle and its surrounding environment in a designated storage space. Once the number of information entries reaches a threshold, the distribution pattern of the interaction information is statistically analyzed.

[0228] S850, based on the calculated distribution law of spatial interaction relationships, constructs a probability distribution model of spatial interaction relationships, namely the first model.

[0229] It should be understood that a single initial model can be built for one object or for multiple objects separately. For example, it can be built for all objects within the area sensed by the sensing device. As an example, it can be built for all vehicles sensed by the sensing device within a specific lane.

[0230] As mentioned earlier, determining the spatial confidence of the first trajectory information based on the first relationship may include: obtaining the probability of the first relationship occurring based on the first model; and determining the spatial confidence of the first trajectory information based on the probability of the first relationship occurring.

[0231] In some scenarios, the probability of the first relationship occurring is directly or inversely proportional to the spatial confidence level of the first trajectory information. The relationship between the probability of the first relationship occurring and the spatial confidence level of the first trajectory information can be determined based on prior experience and the driving habits mentioned earlier. Optionally, the relationship between the probability of the first relationship occurring and the spatial confidence level of the first trajectory information can be pre-stored.

[0232] In some scenarios, the relationship between the probability of interaction between a target object and its surrounding environment may differ depending on the range of values. For example, the probability distribution of the distance between a target object and other surrounding objects might be as follows: very small distances (e.g., less than a certain threshold) have a very low probability of occurrence; very large distances (e.g., greater than a certain threshold) also have a very low probability of occurrence; and moderate distances (e.g., distance M) have a higher probability of occurrence. When the distance is very small, the spatial confidence of the vehicle trajectory information is low; when the distance is moderate or greater than a certain threshold, the spatial confidence of the vehicle trajectory information is high. It can be seen that the relationship between the probability of spatial interaction and the confidence of trajectory information varies under different circumstances.

[0233] Considering the above issues, post-processing can be applied to the first model. For example, when the distance between the target object and other surrounding objects is less than or equal to M, the confidence level of the vehicle trajectory information can be determined based on the probability of spatial interaction occurring as output by the first model. For instance, the spatial confidence level of the vehicle trajectory information is proportional to the probability of spatial interaction occurring. Conversely, when the distance between the target object and other surrounding objects is greater than M, the confidence level of the vehicle trajectory information can be determined based on prior rules, such as setting the confidence level to a fixed value. As an example, this fixed value can be the same as or related to the probability of distance M occurring.

[0234] In some embodiments, artificial intelligence or machine learning techniques can be used to directly learn the relationship between spatial interaction relationships and trajectory confidence, resulting in a second model. That is, the input to the second model is the first relationship, and the output of the second model is the spatial confidence of the first trajectory information. This method is relatively simple to implement during operation.

[0235] Figure 9 This is a flowchart illustrating another method for determining trajectory confidence provided in an embodiment of this application. Figure 9 The method shown includes steps S910 to S940.

[0236] S910 determines the trajectory information of the target vehicle, i.e., the first trajectory information, based on data collected by roadside sensing devices. The target vehicle mentioned here can be the target object mentioned earlier.

[0237] The S920 determines the interaction information between the target vehicle and the surrounding roads based on the high-precision map of the surrounding area configured inside the sensing device.

[0238] S930 extracts the interaction information between the target vehicle and surrounding vehicles based on the trajectory information of other nearby vehicles collected by the sensing device within the same time period.

[0239] It should be understood that the interaction information obtained in steps S920 and S930 is the first relationship mentioned above.

[0240] S940, through the first model, determines the spatial confidence level of the first trajectory information.

[0241] In some embodiments, the first relationship can be used to update the first model. For example, the probability statistics of the occurrence of the first relationship can be updated based on the first relationship obtained in steps S920 and S930, thereby updating the first model based on the result.

[0242] For example, the first relation can be stored in a storage unit for updating the first model. As an example, when the amount of data in the stored first relation reaches a preset threshold, the update of the first model can be triggered, thereby helping to avoid the overhead caused by frequent updates to the first model.

[0243] In addition, updating the first model helps to improve its accuracy.

[0244] The uses of confidence levels in trajectory information can be varied.

[0245] In some embodiments, trajectory confidence can be used for trajectory fusion of multiple perceived trajectories. For example, the first trajectory information may include multiple trajectory information items, which are obtained by multiple sensing devices sensing the target object respectively. After obtaining the confidence levels of the multiple trajectory information items, the multiple trajectories can be fused based on the confidence levels to obtain third trajectory information. The third trajectory information is the final perceived trajectory of the target object.

[0246] It should be understood that when performing trajectory fusion, trajectory fusion can be performed using only the spatial confidence provided in this application, or it can be performed by combining the spatial confidence provided in this application with the temporal confidence in related technologies.

[0247] In some embodiments, trajectory confidence can be used to determine whether the data in the trajectory information is a true value or a false alarm, thereby eliminating false alarms in the trajectory information and improving the reliability of the trajectory information. The trajectory confidence mentioned here may include spatial confidence, or spatial confidence and time confidence.

[0248] In some embodiments, spatial confidence can be used to identify traffic events. For example, based on the spatial confidence of first trajectory information, traffic events related to a target object can be identified. These traffic events include, for example, one or more of parking, speeding, driving in the wrong direction, and lane changing. This approach expands the application scope of confidence analysis results in trajectory post-processing scenarios. In this case, the trajectory confidence analysis module can transmit the spatial confidence to the traffic event identification module for traffic event identification.

[0249] In this application embodiment, the confidence analysis dimension is expanded by analyzing the interaction between the target object and its surrounding environment. This addresses the problem that traditional trajectory confidence analysis algorithms rely on single prior assumptions and cannot fully cover the distribution patterns of vehicle trajectories in intelligent transportation scenarios. This method can be applied not only to intelligent transportation roadside perception scenarios but also to trajectory data quality analysis scenarios for similar objects and sensing carriers, such as vehicle-mounted perception systems in autonomous driving applications and UAV perception and positioning systems in base station sensing integration scenarios. This application does not limit the scope of this application.

[0250] Figure 10 This is a flowchart illustrating another method for determining trajectory confidence provided in an embodiment of this application. Figure 10 The method shown includes steps S1010 to S1090.

[0251] S1010, Input Point Trace, which is the point trace of the target object collected by the sensing device.

[0252] S1020, Initialize the first window.

[0253] S1030: Determine if the initial window initialization is complete. If complete, execute S1040; otherwise, execute S1020.

[0254] S1040: Determine whether to obtain the time-domain confidence score. If it is required, proceed to S1050; otherwise, proceed to S1060.

[0255] S1050, obtain the temporal confidence of the target object.

[0256] S1060: Determine whether to obtain spatial confidence. If spatial confidence is required, proceed to S1070; otherwise, end.

[0257] S1070, obtain the spatial confidence of the target object.

[0258] S1080, Confidence Fusion. That is, the spatial confidence and temporal confidence of the target object are fused.

[0259] S1090 outputs the confidence level of the new point trace, that is, the confidence level of the first trajectory information.

[0260] Step S1050 can be implemented using methods found in related technologies. An example of step S1050 is given below: Step S1050 includes steps S1051 to S1056.

[0261] S1051, calculate and output the confidence score of the dot pattern. The confidence score of the dot pattern can include the self-confidence score and the mutual confidence score of the dot pattern.

[0262] S1052, calculate and output the track confidence score. The track confidence score can include the track self-confidence score and the track mutual confidence score.

[0263] S1053, which integrates point confidence and track confidence.

[0264] S1054, Determine whether to consider adding points. If adding points is needed, execute step S1055 first, then execute step S1056; otherwise, execute step S1056.

[0265] S1055, the points are supplemented using the confidence decay function.

[0266] S1056 outputs the temporal confidence of the newly generated points, that is, the temporal confidence of the first trajectory information.

[0267] This application's embodiments analyze the spatial confidence of real-time perceived points by analyzing the interaction between the target object and its surrounding environment, thus expanding the dimensions of trajectory confidence analysis and improving the accuracy and robustness of the final confidence analysis results.

[0268] In some embodiments, two trajectory data points with the same time-domain confidence level may yield different spatial-domain confidence levels from the confidence analysis module. The following section combines... Figures 11 to 14 A comparison and introduction of time-domain confidence and spatial-domain confidence are presented.

[0269] Figure 11 This is a comparative diagram of time-domain confidence and spatial-domain confidence provided in an embodiment of this application. Figure 11 The vehicles 1 and 2 shown have the same motion state, such as the same speed and direction of travel. According to the time-domain confidence analysis method in related technologies, the trajectory confidence scores for both vehicles 1 and 2 are 0.9.

[0270] However, the spatial interactions between vehicle 1 and vehicle 2 and their surroundings differ. Specifically, vehicle 1's trajectory deviates from the lane centerline, while vehicle 2's trajectory follows the lane centerline. As discussed earlier, the spatial confidence level of vehicle 1 is lower than that of vehicle 2. For example, vehicle 1's trajectory confidence level might be 0.5, while vehicle 2's might be 0.9.

[0271] Figure 12 Another comparative schematic diagram of time-domain confidence and spatial-domain confidence provided for embodiments of this application. Figure 12 The vehicles 1 and 2 shown have the same motion state, such as the same speed and direction of travel. According to the time-domain confidence analysis method in related technologies, the trajectory confidence scores for both vehicles 1 and 2 are 0.9.

[0272] However, the spatial interactions between Vehicle 1 and Vehicle 2 and their surrounding environment differ. Specifically, the spatial interactions between Vehicle 1 and Vehicle 2 and their surrounding environment can include their interactions with the lane centerline and their interactions with the sensing range of the sensing device. The interactions between Vehicle 1 and Vehicle 2 and the lane centerline are essentially the same. Vehicle 1's trajectory starts closer to the sensing boundary of the sensing device, while Vehicle 2's trajectory starts farther from the sensing boundary. As mentioned earlier, the spatial confidence level of Vehicle 1 is higher than that of Vehicle 2. For example, the trajectory confidence level of Vehicle 1 might be 0.9, while that of Vehicle 2 might be 0.5.

[0273] Figure 13 This is another comparative diagram of time-domain confidence and spatial-domain confidence provided in the embodiments of this application. Figure 13 The vehicles 1 and 2 shown have the same motion state, such as the same speed and direction of travel. According to the time-domain confidence analysis method in related technologies, the trajectory confidence scores for both vehicles 1 and 2 are 0.9.

[0274] However, the spatial interactions between vehicles 1 and 2 and their surroundings differ. Specifically, the distances between vehicles 1 and 2 and other vehicles in the vicinity are different. The distance between vehicle 1 and the following vehicle (vehicle 3) is greater than the distance between vehicle 2 and the following vehicle (vehicle 4). As mentioned earlier, the spatial confidence level of vehicle 1 is greater than that of vehicle 2. For example, the trajectory confidence level of vehicle 1 might be 0.9, while that of vehicle 2 might be 0.5.

[0275] Figure 14 This is another comparative diagram of time-domain confidence and spatial-domain confidence provided in the embodiments of this application. Figure 14 The vehicles 1 and 2 shown have the same motion state, such as the same speed and direction of travel. According to the time-domain confidence analysis method in related technologies, the trajectory confidence scores for both vehicles 1 and 2 are 0.9.

[0276] However, the spatial interactions between vehicles 1 and 2 and their surrounding environment differ. Specifically, the speed differences between vehicles 1 and 2 and other surrounding vehicles are different. The speed difference between vehicle 1 and its neighboring vehicle (vehicle 3) is smaller than the speed difference between vehicle 2 and its neighboring vehicle (vehicle 4). As mentioned earlier, the spatial confidence level of vehicle 1 is greater than that of vehicle 2. For example, the trajectory confidence level of vehicle 1 might be 0.9, while that of vehicle 2 might be 0.5.

[0277] It can be seen that the method provided in the embodiments of this application helps to improve the accuracy of trajectory confidence.

[0278] It should be noted that when determining the trajectory confidence level, one can determine the confidence level of the first trajectory information, or one can determine the trajectory confidence level of different parts of the first trajectory information separately.

[0279] It should be understood that the methods provided in the embodiments of this application can also be applied to other sensing objects besides automobiles, and this application does not limit them.

[0280] It should be noted that the preset thresholds mentioned in the embodiments of this application may be the same or different, and may be pre-configured or flexibly set according to the scenario. This application does not limit this.

[0281] The method embodiments provided in this application have been described above. The apparatus embodiments provided in this application will be described below. It should be understood that the description of the apparatus embodiments corresponds to the description of the method embodiments. Therefore, any content not described in detail can be found in the method embodiments above.

[0282] Figure 15 This is a schematic diagram of the structure of a device for determining trajectory confidence, provided in an embodiment of this application. Figure 15 The device 1500 shown includes a first acquisition module 1510, a second acquisition module 1520, and a determination module 1530.

[0283] The first acquisition module 1510 can be used to acquire the first trajectory information of the target object. The first trajectory information is the trajectory information generated when the target object is driving in the first environment. The first trajectory information is obtained by sensing the target object through a sensing device.

[0284] The second acquisition module 1520 can be used to acquire the first relationship between the target object and the first environment based on the first trajectory information and the first environment.

[0285] The determination module 1530 can be used to determine the spatial confidence level of the first trajectory information based on the first relationship.

[0286] In some embodiments, the first environment includes one or more of the following: the position of lane lines; the sensing range of the sensing device; the position of road guardrails; the position of gantry; the position of signs; the geographical environment; the movement speed of other objects around the target object; the trajectory information of the other objects; or the position of the other objects.

[0287] In some embodiments, the first relationship includes one or more of the following: the relationship between the position of the target object and the position of the lane line; the relationship between the starting position of the first trajectory information within the sensing range of the sensing device and the sensing range; the relationship between the ending position of the first trajectory information within the sensing range and the sensing range; the relationship between the position of the target object and the position of the gantry; the relationship between the position of the target object and the position of the sign; the relationship between the position of the target object and the position of the road guardrail; the distance between the other object and the target object; or the difference between the movement speed of the other object and the movement speed of the target object.

[0288] In some embodiments, determining the spatial confidence level of the first trajectory information based on the first relationship includes: obtaining the probability of the first relationship occurring based on a first model; and determining the spatial confidence level of the first trajectory information based on the probability of the first relationship occurring.

[0289] In some embodiments, the device for determining trajectory confidence further includes: a third acquisition module, configured to acquire second trajectory information of a plurality of objects, wherein the second trajectory information is trajectory information generated by the plurality of objects while driving in a test environment, and the plurality of objects includes the target object, or the plurality of objects does not include the target object; a second determination module, configured to determine a second relationship between the plurality of objects and the test environment based on the second trajectory information and the test environment; and a processing module, configured to perform probability distribution fitting on the probability of the second relationship occurring to obtain the first model.

[0290] In some embodiments, the test environment is the first environment.

[0291] In some embodiments, the first relationship is used to update the first model.

[0292] In some embodiments, the first trajectory information includes multiple trajectory information, which are obtained by multiple sensing devices sensing the target object respectively. The device for determining the trajectory confidence level further includes a fusion module, which is used to obtain the third trajectory information of the target object according to the spatial confidence level corresponding to the multiple trajectory information respectively, wherein the third trajectory information is the fusion result of the multiple trajectory information.

[0293] In some embodiments, the device for determining trajectory confidence further includes: an identification module, configured to identify traffic events related to the target object based on the spatial confidence of the first trajectory information, wherein the traffic events include one or more of parking, speeding, driving in the wrong direction, and lane changing.

[0294] As one possible implementation, the vehicle-to-infrastructure (V2I) and vehicle-to-vehicle (V2V) interactions mentioned above can be handled by different modules. For example, acquiring interaction relationships, establishing probabilistic analysis models, and performing confidence inference based on interaction relationships can all be handled by different modules.

[0295] Figure 16 This is a schematic diagram of another device for determining trajectory confidence provided in an embodiment of this application. Figure 16 The device 1600 shown includes a vehicle-road interaction relationship analysis module 1610, a vehicle-road interaction relationship probability modeling module 1620, a vehicle-road relationship reasoning module 1630, a vehicle-vehicle interaction relationship analysis module 1640, a vehicle-vehicle interaction relationship probability modeling module 1650, and a vehicle-vehicle relationship reasoning module 1660.

[0296] The vehicle-road interaction analysis module 1610 can obtain vehicle-road interaction relationships based on real-time raw data points, such as the distance between vehicles and lane lines.

[0297] The vehicle-road interaction probability modeling module 1620 can perform probability modeling based on the vehicle-road interaction relationship output by the vehicle-road interaction relationship analysis module 1610, such as obtaining a vehicle-road relationship probability model through probability analysis fitting. Optionally, the vehicle-road interaction probability modeling module 1620 can also update the vehicle-road relationship probability model based on the vehicle-road interaction relationship of the target object.

[0298] The vehicle-road relationship reasoning module 1630 can perform reasoning based on the vehicle-road relationship probability model obtained through the vehicle-road interaction relationship probability modeling module 1620, such as determining the vehicle-road relationship confidence of the target object based on the vehicle-road interaction relationship of the target object.

[0299] The vehicle-to-vehicle interaction relationship analysis module 1640 can obtain vehicle-to-vehicle interaction relationships based on real-time raw data points, such as the speed difference between the vehicle and surrounding vehicles.

[0300] The vehicle-to-vehicle interaction relationship probability modeling module 1650 can perform probability modeling based on the vehicle-to-vehicle interaction relationship output by the vehicle-to-vehicle interaction relationship analysis module 1640, such as obtaining a vehicle-to-vehicle relationship probability model through probability analysis fitting. Optionally, the vehicle-to-vehicle interaction relationship probability modeling module 1650 can also update the vehicle-to-vehicle relationship probability model based on the vehicle-to-vehicle interaction relationship of the target object.

[0301] The vehicle-to-vehicle relationship reasoning module 1660 can perform reasoning based on the vehicle-to-vehicle relationship probability model obtained through the vehicle-to-vehicle interaction relationship probability modeling module 1650, such as determining the vehicle-to-vehicle relationship confidence of the target object based on the vehicle-to-vehicle interaction relationship of the target object.

[0302] The vehicle-road relationship confidence scores output by the vehicle-road relationship reasoning module 1630 and the vehicle-vehicle relationship confidence scores output by the vehicle-vehicle relationship reasoning module 1660 can be transmitted to the trajectory post-processing module for functions such as trajectory fusion or traffic event recognition.

[0303] Figure 17 A schematic diagram of another device for determining trajectory confidence provided in an embodiment of this application. Figure 17 As shown, the device 1700 for determining trajectory confidence includes at least one processor 1701 and at least one memory 1702.

[0304] At least one memory 1702 for storing computer program (or instructions) 1703, and the computer program 1703 is executable on the at least one processor 1701.

[0305] At least one processor 1701 is configured to execute part or all of a computer program (or instructions) 1703 in at least one memory 1702 to cause the method for determining trajectory confidence provided in this application to be executed.

[0306] For example, computer program 1703 may be divided into one or more modules / units, one or more of which are stored in memory 1702 and executed by processor 1701 to complete this application. One or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in device 1700 for determining trajectory confidence.

[0307] Those skilled in the art will understand that Figure 17 This is merely an example of the device 1700 for determining trajectory confidence and does not constitute a limitation on the device for determining trajectory confidence. It may include more or fewer components than illustrated, or combine certain components, or different components. For example, the device 1700 for determining trajectory confidence may also include input / output devices, network access devices, buses, etc.

[0308] It should be understood that, in one possible design, the steps in the method embodiments provided in this application can be implemented by integrated logic circuits in the processor's hardware or by instructions in software form. The steps of the method disclosed in the embodiments of this application can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules in the processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, detailed descriptions are not provided here.

[0309] It should be noted that the processor in the embodiments of this application can be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method embodiments can be completed by the integrated logic circuitry in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied as being executed by a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above methods.

[0310] It is understood that the memory in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM). It should be noted that the memory used in the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0311] The device 1700 for determining trajectory confidence provided in this embodiment can execute the above method embodiment. Its implementation principle and technical effect are similar, and will not be described again here.

[0312] Figure 18 This is a schematic diagram of the structure of a sensing system provided in an embodiment of this application. Figure 18 The sensing system 1800 shown may include multiple sensing devices 1810, a device 1820 for determining trajectory confidence, and a processing device 1830.

[0313] Multiple sensing devices 1810 can be used to acquire first trajectory information of a target object. The first trajectory information includes multiple trajectory information, which are obtained by the multiple sensing devices sensing the target object respectively. The first trajectory information is the trajectory information generated when the target object moves in the first environment.

[0314] The device 1820 for determining trajectory confidence can be used to determine the spatial confidence of the first trajectory information based on the first trajectory information and the first environment. Optionally, the device 1820 for determining trajectory confidence can be one of the devices 1500, 1600 and 1700 mentioned above.

[0315] The processing device 1830 can be used to obtain the third trajectory information of the target object based on the spatial confidence level. The third trajectory information is the fusion result of multiple trajectory information.

[0316] Figure 19 This is a schematic diagram of another perception system provided in an embodiment of this application. The perception system may include a perception device 1910 and a perception engine 1920. The perception device 1910 may be replaced by the multiple perception devices 1810 mentioned above, and the perception engine 1920 may include the device 1820 for determining trajectory confidence and the processing device 1830 mentioned above. The roadside perception process may involve the interaction between the roadside (i.e., the perception system), the cloud, and the vehicle.

[0317] The perception engine 1920 may include a vision processing unit 1921, a radar access unit 1922, a confidence analysis unit 1923, a radar vision perception data fitting unit 1924, a multi-station collaboration unit 1925, an event recognition unit 1926, a map processing unit 1927, and a message routing unit 1928, etc.

[0318] The external inputs to the perception engine 1920 may include the perception results from the perception device 1910, such as camera perception results and radar perception results. Camera perception results can be transmitted to the vision processing unit 1921 via a standard interface, and radar perception results can be transmitted to the radar access unit 1922 via a proprietary interface.

[0319] External inputs to the Perception Engine 1920 can also include map information. Map information can be transmitted to the Perception Engine 1920 from a cloud server via a standard interface.

[0320] It should be understood that the above interface types are provided as examples only, and the above interface types may change in future standards. For example, if private interfaces are standardized, this application does not limit them.

[0321] The vision processing unit 1921 can process the camera perception results to obtain the trajectory information of the target object perceived by the camera.

[0322] The radar access unit 1922 can process the radar sensing results to obtain the trajectory information of the target objects sensed by the radar. For example, trajectory preprocessing methods for radar sensing results include clustering algorithms for sensed point cloud data, Kalman filtering algorithms based on inertial motion laws, and time-series multi-object matching algorithms based on the Hungarian algorithm.

[0323] The vision processing unit 1921 can input the trajectory information of the perceived target object into the confidence analysis unit 1923, and the radar access unit 1922 can input the trajectory information of the perceived target object into the confidence analysis unit 1923. The confidence analysis unit 1923 can determine the confidence level of multiple trajectory information of the target object.

[0324] The confidence analysis unit 1923 may include a time-domain confidence unit 1923a and a spatial-domain confidence unit 1923b.

[0325] Among them, the time-domain confidence unit 1923a can determine the confidence level of different trajectories according to the chi-square test method mentioned above. In this case, the perception engine can obtain the motion state information of the target object from the vehicle, such as inertial information, or it can obtain the motion state information of the target object based on the perception results of the camera or radar.

[0326] The spatial confidence unit 1923b can determine the trajectory confidence using the methods mentioned above, such as... Figure 3 The method shown determines the spatial confidence of multiple trajectories. Alternatively, the spatial confidence unit 1923b can be a device 1820 for determining the confidence of trajectories.

[0327] In some embodiments, the confidence analysis unit 1923 may also include other units, such as a unit for obtaining the interaction relationship between the target object and the surrounding environment, and a unit for fusing the temporal confidence and the spatial confidence.

[0328] To minimize the need for modifications to existing schemes, the spatial confidence unit 1923b can be deployed after the temporal confidence unit 1923a.

[0329] The radar-visual perception data fitting unit 1924 can fuse the perception results of different sensing devices to obtain the final trajectory information of the target object. For example, the radar-visual perception data fitting unit 1924 can perform trajectory fusion based on the trajectory confidence level output by the confidence analysis module, which helps improve the accuracy and efficiency of trajectory post-processing. For example, the radar-visual perception data fitting unit 1924 may include a video + video fitting unit (i.e., C+C), a radar + radar fitting unit (i.e., R+R), or a video + radar fitting unit (i.e., C+R). Optionally, the radar-visual perception data fitting unit 1924 can be the processing device 1830 mentioned above.

[0330] The trajectory information of the target object obtained through the above steps is only within a certain area, such as the trajectory information of a target object at a single station. To obtain the complete trajectory information of the target object, the multi-station collaboration unit 1925 can acquire the trajectory information of the target object at other stations. For example, the multi-station collaboration unit 1925 can acquire lane information from the map processing unit 1927, which can be used, for example, for fitting the trajectory information of the target object at different stations.

[0331] In addition, the multi-station coordination unit 1925 can transmit the traffic statistics results of the sensing data and the relevant information of the target object to the event recognition unit 1926 for the purpose of recognizing traffic events.

[0332] The message routing unit 1928 can be used to transmit the processing results in the perception engine to other units or systems.

[0333] It can be seen that in the roadside sensing trajectory data analysis process, the trajectory data reported by the sensing device undergoes a confidence assessment first after entering the sensing engine. Subsequent operations such as false alarm rejection and trajectory fusion are then performed based on the trajectory confidence assessment results. Related technologies can utilize the inertial motion laws of moving objects to perform preliminary trajectory confidence analysis, i.e., time-domain confidence analysis, through the chi-square test.

[0334] This application embodiment can perform additional spatial confidence analysis after the temporal confidence analysis step in related technologies, and fuse the various confidence analysis results after obtaining the spatial confidence analysis results, which helps to improve the accuracy and robustness of the trajectory confidence analysis results. Using the fused confidence analysis results to guide subsequent trajectory fusion or traffic event judgment helps to improve the reliability of trajectory fusion and traffic event judgment.

[0335] Figure 20 This is a schematic diagram of another sensing system provided in an embodiment of this application. The sensing system may include a sensing device 2010 and a sensing analysis module 2020. The sensing device 2010 may be any of the sensing devices mentioned above, and the sensing analysis module 2020 may be any of the devices mentioned above for determining trajectory confidence.

[0336] See Figure 20 The dashed area represents the sensing area of ​​the sensing device. It can be seen that the sensing device's range can cover a certain road area and detect target objects passing through that area.

[0337] When the sensing device 2010 senses a target object, it can acquire the target object's trajectory information and transmit this trajectory information to the sensing analysis module 2020. For example, the sensing analysis module 2020 can preprocess the raw sensing data, such as dot data, to obtain the target object's trajectory information. Alternatively, Figure 20 The sensing system shown may also include a sensing data preprocessing module 2030 for preprocessing the raw sensing data.

[0338] The perception and analysis module 2020 may include a computing unit 2020a and a storage unit 2020b. The computing unit 2020a may also be referred to as a processing unit.

[0339] On one hand, the perception and analysis module 2020, such as the computing unit 2020a, can determine the interaction relationship between the target object and its surrounding environment based on the target object's trajectory information, and determine the confidence level of the trajectory information based on this interaction relationship, i.e., it is used to determine the trajectory confidence level in real time. In the process of determining the confidence level, the first model mentioned earlier can be used. For example, the computing unit 2020a can call the first model from the storage unit 2020b to determine the confidence level of the trajectory information based on the aforementioned interaction relationship. In other words, the first model can be stored in the storage unit 2020b.

[0340] On the other hand, the perception and analysis module 2020, such as the computing unit 2020a, can store the interaction relationship between the target object and its surrounding environment in the storage unit 2020b. When the data of the interaction relationships stored in the storage unit 2020b reaches a certain amount, the computing unit 2020a can call these stored interaction relationships to update the first model.

[0341] During the testing process, specifically the establishment of the first model, the computing unit 2020a can determine the interaction relationship between the target object and its surrounding environment based on the target object's trajectory information, and store this interaction relationship in the storage unit 2020b. When the number of interaction relationship data stored in the storage unit 2020b reaches a certain amount, the computing unit 2020a can call these stored interaction relationships, thereby triggering the establishment of the first model.

[0342] In some embodiments, Figure 20 The perception system shown may also include a perception data post-processing module 2040. The perception data post-processing module 2040 can be used to perform trajectory post-processing based on the trajectory confidence of the target object output by the perception analysis module 2020, such as trajectory fusion, false alarm elimination, and traffic event recognition.

[0343] This application also provides a computer program product, which includes computer program code. When the computer program code is run on a computer, it causes the computer to execute the various steps or processes executed by the device for determining trajectory confidence in any of the above method embodiments.

[0344] This application also provides a computer-readable storage medium storing program code that, when run on a computer, causes the computer to execute the various steps or processes performed by the device for determining trajectory confidence in any of the above method embodiments.

[0345] This application also provides a communication device, including a processor and an interface for sending and / or receiving signals, such that the processor executes the various steps or processes performed by the device for determining trajectory confidence in any of the above method embodiments.

[0346] The above-described device and method embodiments are completely corresponding, with corresponding modules or units performing corresponding steps. For example, a communication unit or communication interface performs the receiving or sending steps in the method embodiment, while other steps besides sending and receiving can be performed by a processing unit or processor.

[0347] In the embodiments of this application, the terms and English abbreviations are exemplary examples given for ease of description and should not be construed as limiting the application in any way. The embodiments of this application do not preclude the possibility of defining other terms that can achieve the same or similar functions in existing or future agreements.

[0348] The terms “component,” “module,” “system,” etc., used in this specification are used to refer to computer-related entities, hardware, firmware, combinations of hardware and software, software, or software in execution. For example, a component can be, but is not limited to, a process running on a processor, a processor, an object, an executable file, an execution thread, a program, and / or a computer. As illustrated, applications running on computing devices and computing devices can both be components. One or more components may reside in a process and / or an execution thread, and components may be located on a single computer and / or distributed among two or more computers. Furthermore, these components can be executed from various computer-readable storage media on which various data structures are stored. Components can communicate, for example, via local and / or remote processes based on signals having one or more data packets (e.g., data from two components interacting with another component between a local system, a distributed system, and / or a network, such as the Internet interacting with other systems via signals).

[0349] Those skilled in the art will recognize that the various illustrative logical blocks and steps described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0350] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be based on the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0351] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only 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 coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0352] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0353] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0354] In the above embodiments, the functions of each functional unit can be implemented entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions (programs). When the computer program instructions (programs) are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state disks, SSDs), etc.

[0355] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, essentially or in other words, the parts that contribute to the prior art, or parts of the technical solutions, 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, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0356] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for determining trajectory confidence, characterized in that, include: The first trajectory information of the target object is obtained. The first trajectory information is the trajectory information generated by the target object when it is driving in the first environment. The first trajectory information is obtained by sensing the target object through a sensing device. Based on the first trajectory information and the first environment, a first relationship between the target object and the first environment is obtained; Based on the first relationship, the spatial confidence level of the first trajectory information is determined.

2. The method according to claim 1, characterized in that, The first environment includes one or more of the following: The position of the lane lines; The sensing range of the sensing device; Location of road guardrails; The location of the gantry frame; The location of the sign; Geographical environment; The speed of movement of other objects around the target object; The trajectory information of the other objects; or The location of the other objects.

3. The method according to claim 2, characterized in that, The first relationship includes one or more of the following: The relationship between the position of the target object and the position of the lane line; The relationship between the starting position of the first trajectory information and the sensing range of the sensing device; The relationship between the termination position of the first trajectory information within the sensing range and the sensing range itself; The relationship between the position of the target object and the position of the gantry frame; The relationship between the position of the target object and the position of the sign; The relationship between the position of the target object and the position of the road guardrail; The distance between the other objects and the target object; or The difference between the speed of the other objects and the speed of the target object.

4. The method according to any one of claims 1-3, characterized in that, Determining the spatial confidence level of the first trajectory information based on the first relationship includes: Based on the first model, obtain the probability of the first relationship occurring; Based on the probability of the first relationship occurring, the spatial confidence level of the first trajectory information is determined.

5. The method according to claim 4, characterized in that, The method further includes: Acquire second trajectory information of multiple objects, wherein the second trajectory information is trajectory information generated by the multiple objects when they drive in the test environment, and the multiple objects include the target object, or the multiple objects do not include the target object; Based on the second trajectory information and the test environment, a second relationship between the plurality of objects and the test environment is determined; The probability distribution of the probability of the second relationship occurring is fitted to obtain the first model.

6. The method according to claim 5, characterized in that, The test environment is the first environment.

7. The method according to claim 5 or 6, characterized in that, The first relationship is used to update the first model.

8. The method according to any one of claims 1-7, characterized in that, The first trajectory information includes multiple trajectory information, which are obtained by multiple sensing devices sensing the target object respectively. The method further includes: Based on the spatial confidence scores corresponding to the multiple trajectory information, the third trajectory information of the target object is obtained, wherein the third trajectory information is the fusion result of the multiple trajectory information.

9. The method according to any one of claims 1-8, characterized in that, The method further includes: Based on the spatial confidence of the first trajectory information, traffic events related to the target object are identified, including one or more of parking, speeding, driving in the wrong direction, and lane changing.

10. The method according to any one of claims 1-9, characterized in that, The first environment is determined based on map information associated with the first trajectory information, and / or the first environment is determined based on the perception results of other objects around the target object.

11. A device for determining trajectory confidence, characterized in that, include: The first acquisition module is used to acquire first trajectory information of the target object. The first trajectory information is the trajectory information generated when the target object moves in the first environment. The first trajectory information is obtained by sensing the target object through a sensing device. The second acquisition module is used to acquire the first relationship between the target object and the first environment based on the first trajectory information and the first environment; The determination module is used to determine the spatial confidence level of the first trajectory information based on the first relationship.

12. A device for determining trajectory confidence, characterized in that, include: At least one memory for storing computer programs or instructions; as well as At least one processor is configured to execute some or all of the computer programs or instructions in the at least one memory such that the method as described in any one of claims 1-10 is performed.

13. A sensing system, characterized in that, include: Multiple sensing devices are used to acquire first trajectory information of a target object. The first trajectory information includes multiple trajectory information, which are obtained by the multiple sensing devices sensing the target object respectively. The first trajectory information is the trajectory information generated when the target object travels in a first environment. The device for determining trajectory confidence as described in claim 11 or 12 is used to determine the spatial confidence of the first trajectory information based on the first trajectory information and the first environment. A processing device is used to obtain third trajectory information of the target object based on the spatial confidence level, wherein the third trajectory information is the fusion result of the multiple trajectory information.

14. A computer program product, characterized in that, Includes a computer program, which, when run, causes the method as described in any one of claims 1-10 to be performed.

15. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program or instructions that, when read and executed by a computer, cause the method as described in any one of claims 1-10 to be performed.

Images