Method of determining a motion state of a target and electronic device

Abstract

The present disclosure relates to the technical field of automatic driving, and particularly relates to a method for determining the motion state of a target and an electronic device. According to the embodiment of the present disclosure, a consistent spatial reference is determined between different frame point clouds through a unified preset grid, and a corresponding relationship between the grid and a tracking unit is established. Since the grid position is fixed, the point cloud data of the tracking unit in the region in each frame of point cloud data can be determined, that is, the point cloud distribution features belonging to the same tracking unit can be accurately obtained in different frame point cloud data, and the motion state of the target in the tracking unit is determined based on the point cloud distribution features of the tracking unit in the current frame and the historical frames, thereby avoiding the position deviation caused by the fluctuation of the clustering algorithm and the matching error of the single target between frames, reducing the false detection rate of the motion state of the target in the tracking unit, and effectively improving the operation reliability and safety of the automatic driving system.

Description

Technical Field

[0001] This disclosure relates to the field of autonomous driving technology, and more particularly to a method and electronic device for determining the motion state of a target. Background Technology

[0002] In the field of autonomous driving, accurately distinguishing between static and dynamic targets in the environment based on information such as the position and speed of targets detected in the surrounding environment is an important prerequisite for autonomous driving systems to make path planning and obstacle avoidance decisions.

[0003] Currently, the perception system on vehicles relies on clustering algorithms to determine targets and judges the motion state of the detected targets based on the changes in their position between frames. However, due to fluctuations in clustering algorithms or errors in target frame matching, false detections of motion state may occur, causing the autonomous driving system to perform unnecessary avoidance behaviors or collisions, thus affecting the reliability of autonomous driving. Summary of the Invention

[0004] To address the aforementioned technical problems, this disclosure provides a method and electronic device for determining the motion state of a target.

[0005] A first aspect of this disclosure provides a method for determining the motion state of a target, comprising: Acquire at least one tracking unit for a historical frame corresponding to a historical frame point cloud, wherein the historical frame point cloud is mapped in a preset grid, the tracking unit consists of at least one grid, and the tracking unit includes the point cloud corresponding to the target to be observed; Map the current frame point cloud onto the grid, and determine the tracking unit of the current frame based on the grid corresponding to the current frame point cloud and the tracking unit of the previous frame; Determine the point cloud distribution characteristics within the tracking unit of the current frame; Obtain the point cloud distribution characteristics of the point cloud within the tracking unit of the historical frame; The motion state of the target to be observed is determined based on the point cloud distribution characteristics of the current frame and the point cloud distribution characteristics of historical frames.

[0006] A second aspect of this disclosure provides an electronic device, including: processor; Memory, used to store executable instructions; The processor is used to read executable instructions from memory and execute the executable instructions to implement the method for determining the motion state of a target provided in the first aspect above.

[0007] The technical solution provided in this disclosure has the following advantages compared with the prior art: The method and electronic device for determining the motion state of a target provided in this disclosure establish a consistent spatial reference between different frame point clouds using a unified preset grid and establish a correspondence between the grid and the tracking unit. Since the grid position is fixed, the point cloud data of the tracking unit in each frame of point cloud data can be determined. This allows for the accurate acquisition of point cloud distribution features belonging to the same tracking unit in different frame point cloud data. Based on the point cloud distribution features of the tracking unit in the current frame and historical frames, the motion state of the target within the tracking unit is determined. This avoids positional deviations caused by clustering algorithm fluctuations and single-target frame matching errors, reduces the false detection rate of the target motion state within the tracking unit, and effectively improves the operational reliability and safety of the autonomous driving system. Attached Figure Description

[0008] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0009] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart of a method for determining the motion state of a target according to an embodiment of this disclosure; Figure 2 This is a flowchart of a method for determining the motion state of a target according to another embodiment of this disclosure; Figure 3 This is a schematic diagram of the structure of a device for determining the motion state of a target according to an embodiment of this disclosure; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0011] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0012] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0013] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.

[0014] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0015] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0016] This disclosure provides a method for determining the motion state of a target, and the method will be described below with reference to specific embodiments.

[0017] Figure 1 This is a flowchart illustrating a method for determining the motion state of a target according to an embodiment of this disclosure. This method can be executed by a device for determining the motion state of the target. The device can be implemented in software and / or hardware and can be configured in an electronic device, such as a server or terminal. Specifically, the terminal includes in-vehicle devices, mobile phones, computers, or tablet computers. It is understood that the method for determining the motion state of a target provided in this disclosure can also be applied to other scenarios.

[0018] The following is about Figure 1 The method for determining the motion state of a target, as shown, will be introduced. Figure 1 As shown, the method for determining the motion state of a target provided in this embodiment includes the following steps.

[0019] S101. Obtain at least one tracking unit corresponding to the historical frame point cloud.

[0020] In this process, the historical frame point cloud is mapped in a preset grid, and the tracking unit consists of at least one grid, which includes the point cloud corresponding to the target to be observed.

[0021] Historical frame point cloud refers to the 3D point cloud data acquired frame by frame by the sensor before the current acquisition time, with each frame corresponding to a historical acquisition time.

[0022] The preset grid is obtained by dividing the three-dimensional space on a horizontal plane. Each frame's point cloud is mapped to the same grid to ensure that the reference of spatial position is consistent between frames.

[0023] A tracking unit is a spatial region consisting of at least one grid, containing point cloud data corresponding to the target to be observed, used to determine the spatial position of the same target to be observed in different frames.

[0024] After acquiring historical frame point cloud data, the electronic device determines at least one grid occupied by the target to be observed based on the distribution position of the point cloud, thus obtaining a tracking unit.

[0025] S102. Map the current frame point cloud onto the grid, and determine the tracking unit of the current frame based on the grid corresponding to the current frame point cloud and the tracking unit of the previous frame.

[0026] The current frame point cloud is the latest 3D point cloud data acquired by the sensor at the current moment. The current frame 3D point cloud is matched into a preset grid so that the current frame point cloud and the historical frame point clouds are in the same grid layout.

[0027] After the electronic device maps the current frame point cloud onto the grid, it determines the tracking unit of the current frame based on the position of the tracking unit in the grid of the previous frame and the current frame point cloud of that position and its surroundings.

[0028] For example, the electronic device transforms the point cloud in the local coordinate system to the global coordinate system based on the current frame point cloud data and vehicle pose information. Then, it discretizes the space into a cylindrical grid according to a preset voxel size. Each cylindrical grid contains all the point cloud data within that spatial region. The voxel size can be adjusted according to accuracy requirements and computing resources; for example, it can be set to 0.5 meters based on real-time and accuracy requirements.

[0029] For each columnar grid, it is divided into multiple height intervals based on the height distribution of its internal point cloud. Each height interval includes information such as upper and lower height limits, occupancy status, and reliability. In other words, the grid in this embodiment also contains point cloud height distribution information, which can more accurately represent the point cloud distribution state in three-dimensional space. Furthermore, the adoption of a regularized height layering strategy has clear physical meaning and interpretability, conforming to the design concept of a rule-based framework.

[0030] S103. Determine the point cloud distribution characteristics within the tracking unit of the current frame.

[0031] Point cloud distribution features are used to characterize the location and shape of the spatial distribution of point clouds within the tracking unit. The electronic device performs statistical analysis and calculations on the point cloud data within the spatial range of the tracking unit in the current frame, extracting feature quantities that can characterize the spatial distribution state of the target.

[0032] S104. Obtain the point cloud distribution characteristics of the point cloud within the tracking unit of the historical frame.

[0033] The electronic device extracts the same distribution features of the same tracking unit at historical moments from the time-series stored historical records, and uses them as a historical reference benchmark corresponding to the features of the current frame.

[0034] S105. Determine the motion state of the target to be observed based on the point cloud distribution characteristics of the current frame and the point cloud distribution characteristics of the historical frames.

[0035] Motion state refers to the motion attributes of the target being observed. Electronic devices determine the degree of spatial change of the target over time by analyzing the difference in point cloud distribution characteristics between the current frame and historical frames, thereby determining the motion state of the target.

[0036] This embodiment of the disclosure establishes a consistent spatial reference across different frame point clouds using a unified preset grid and establishes a correspondence between the grid and the tracking unit. Since the grid position is fixed, the point cloud data of the tracking unit in each frame of point cloud data can be determined. This allows for the accurate acquisition of point cloud distribution characteristics belonging to the same tracking unit in different frame point cloud data. Based on the point cloud distribution characteristics of the tracking unit in the current frame and historical frames, the motion state of the target within the tracking unit can be determined. This avoids positional deviations caused by clustering algorithm fluctuations and single-target frame matching errors, reduces the false detection rate of target motion state within the tracking unit, and effectively improves the operational reliability and safety of the autonomous driving system.

[0037] In some embodiments, determining the tracking unit of the current frame based on the grid corresponding to the current frame point cloud and the tracking unit of the previous frame includes: determining whether there is a point cloud in the target grid within the multiple grids corresponding to the current frame point cloud where the distance to the tracking unit of the previous frame is less than a preset distance threshold.

[0038] If so, update the tracking unit of the previous frame so that the point cloud in the target mesh is included in the updated tracking unit, and use the updated tracking unit as the tracking unit of the current frame.

[0039] If not, the tracking unit of the previous frame is used as the tracking unit of the current frame.

[0040] The multiple grids corresponding to the current frame point cloud refer to the set of grids containing point cloud data after the 3D point cloud data is acquired at the current moment and mapped to a preset grid according to the same rules as historical frames. The tracking unit of the previous frame is a set of grids containing the point cloud of the target to be observed, determined after acquiring the point cloud data of the previous frame, representing the spatial boundary of the target to be observed at the previous moment.

[0041] Target grids whose distance from the tracking unit in the previous frame is less than a preset distance threshold are grids that the target to be observed may occupy in the current frame. If point clouds exist in the target grids of multiple grids corresponding to the point cloud in the current frame, it means that the target to be observed may appear in the target grids, and these point clouds need to be included in the tracking unit for observation. The electronic device adjusts the spatial boundary of the tracking unit in the previous frame to include the target grids where point clouds are detected into the composition range of the tracking unit.

[0042] If the target grid is not found in any of the multiple grids corresponding to the point cloud in the current frame, it means that no new point cloud was detected in the vicinity of the tracking unit in the previous frame, and the original tracking unit can still completely contain the target to be observed. In this case, the spatial range of the tracking unit is not adjusted, and the tracking unit of the previous frame is directly used as the tracking unit of the current frame.

[0043] This embodiment limits the detection range by setting a preset distance threshold, thereby reducing the search space of the tracking unit and improving computational efficiency. Furthermore, it dynamically expands the boundary of the tracking unit when a point cloud exists within the target grid, avoiding missing point clouds of the target to be observed; when no point cloud exists in the target grid, it directly uses the tracking unit from the previous frame, reducing unnecessary computational overhead and balancing tracking efficiency and accuracy.

[0044] In some embodiments, if the current frame point cloud is the first frame point cloud, at least one tracking unit is determined based on the grid occupancy status and height distribution information in the first frame point cloud. For the second frame point cloud data, the first frame point cloud is used as the historical frame point cloud, and the second frame point cloud is used as the current frame point cloud.

[0045] In some embodiments, if, relative to the tracking units of the historical frames, a new grid is occupied in the current frame point cloud (excluding the target grid), or a new height distribution interval appears in the already occupied grid, then a new tracking unit is determined based on the newly occupied grid and / or height distribution interval for subsequent tracking.

[0046] The process of determining the distribution characteristics of point clouds is explained below.

[0047] In some embodiments, determining the point cloud distribution characteristics of the point cloud within the tracking unit of the current frame includes: determining the position of at least one type of feature point within the tracking unit of the current frame based on the point cloud within the tracking unit of the current frame, wherein the position of the feature point is used to characterize the point cloud distribution characteristics of the point cloud within the tracking unit.

[0048] Accordingly, obtaining the point cloud distribution features of the point cloud within the tracking unit of the historical frame includes: obtaining the position of at least one type of feature point within the tracking unit of the historical frame.

[0049] Within the defined tracking unit range of the current frame, the electronic device performs statistical calculations on the point cloud data to extract at least one type of spatially representative feature points. It is understood that the determination method for at least one type of feature points corresponding to historical frames is the same as that for at least one type of feature points corresponding to the current frame.

[0050] Optionally, a single type of feature point can be selected to adapt to low computing power scenarios, or multiple types of feature points can be selected to improve the accuracy of point cloud distribution feature representation.

[0051] One possible implementation is to determine the position of the centroid of the point cloud within the tracking unit as the position of the first type of feature point.

[0052] The centroid of a point cloud refers to the mean coordinate point of the point cloud within the tracking unit. It is the coordinate point obtained by taking the arithmetic mean of the three-dimensional coordinates of all points, reflecting the overall center position of the target to be observed.

[0053] Another possible implementation is to determine the position of the geometric center of the convex hull of the point cloud within the tracking unit as the position of the second type of feature point.

[0054] The convex hull is the smallest convex polygon that contains all point clouds within the tracking unit, reflecting the overall shape and extent of the target being observed. The geometric center is the centroid of the convex hull polygon, reflecting the center position of the target's outer contour.

[0055] Another possible implementation is to determine the position of the third type of feature point by the average position of the geometric center of multiple point cloud segments in the tracking unit. The multiple point cloud segments are obtained by segmenting the tracking unit in the vertical direction based on a preset height value, and the height value of each point cloud segment is the preset height value.

[0056] The preset height value is a pre-defined vertical slice height parameter used to divide the point cloud into multiple point cloud segments of equal height at equal intervals in the vertical direction. A point cloud segment refers to the set of point clouds contained within each height interval, representing the point cloud information of the target object within that height interval. The electronic device calculates the geometric center of the point cloud within each height segment, and then averages the center coordinates of all segments to obtain the position of the third type of feature points.

[0057] This embodiment extracts three different types of feature points to characterize the point cloud distribution within the tracking unit from different dimensions, providing rich information for motion state determination. Simultaneously, the third type of feature points incorporates vertical point cloud height distribution information, better aligning with the spatial characteristics of the three-dimensional point cloud and resulting in more accurate point cloud representation of the three-dimensional target.

[0058] After acquiring at least one type of feature points, for each type of feature point, based on the position of the feature point in the current frame and the positions of the feature points in historical frames, the orientation change information of the feature point is determined. Based on the orientation change information, the motion state corresponding to the feature point is determined. Furthermore, based on the motion state corresponding to at least one type of feature point, the motion state of the target to be observed is determined.

[0059] Orientation change information is at least one indicator characterizing the degree of spatial change of a feature point over time, including the amount of change in direction and / or the amount of change in position. The motion state corresponding to a feature point refers to the motion state of the observed target determined based on a single type of feature point.

[0060] Optionally, if only one type of feature point exists, the motion state corresponding to the feature point is the motion state of the target to be observed.

[0061] Specifically, when the orientation change information includes the direction change amount, the orientation change information of the feature point is determined based on the position of the feature point in the current frame and the position of the feature point in the historical frame. This includes determining the maximum value between the angle corresponding to the feature point in the current frame and the angle corresponding to the feature point in the historical frame as the direction change information of the feature point.

[0062] The angle corresponding to the feature point in the current frame is the angle between the displacement vector corresponding to the current frame and the displacement vector corresponding to the previous frame. The displacement vector corresponding to the current frame is a vector with the feature point in the current frame as the endpoint and the feature point in the previous frame as the starting point.

[0063] The displacement vector is composed of the positions of the same type of feature points in two consecutive frames, starting from the position of the feature point in the previous frame and ending at the position of the feature point in the current frame. The angle between the displacement vectors of two adjacent frames is used to characterize the stability of the motion direction of the observed target. The smaller the angle, the smaller the change in the motion direction of the observed target; the larger the angle, the more drastic the change in the motion direction of the observed target.

[0064] The maximum angle among the feature points in historical frames represents the maximum change in the direction of motion of the target during tracking. By taking the maximum value between the angles of the feature points in the current frame and those in historical frames, the electronic device obtains the direction change information of the feature points, making it easier to identify the direction changes of the target.

[0065] When the orientation change information includes the position change amount, the orientation change information of the feature point is determined based on the position of the feature point in the current frame and the position of the feature point in the historical frame. This includes determining the maximum value between the displacement of the feature point in the current frame and the displacement of the feature point in the historical frame as the position change amount of the feature point.

[0066] The displacement of a feature point in the current frame is the displacement distance between the feature point in the current frame and the feature point in the previous frame.

[0067] Displacement distance is the linear spatial distance between corresponding feature points of the same type in two consecutive frame point clouds, used to characterize the magnitude of the feature point displacement. The maximum displacement of feature points in historical frames is the maximum displacement distance of the same type of feature points between adjacent frames during the tracking of the target.

[0068] The electronic device takes the maximum value between the displacement of the feature point in the current frame and the displacement of the feature point in the historical frame to obtain the position change information of the feature point, making it easier to identify the position change of the target to be observed.

[0069] This embodiment of the disclosure evaluates the changes of feature points in time from different perspectives by calculating the directional and positional changes of each type of feature point. At the same time, by adopting the maximum value statistical method, it can effectively deal with abnormal scenarios such as partial missing points in the point cloud and temporary occlusion of the target to be observed, avoid misjudgment caused by the degradation of single-frame point cloud quality, and improve the accuracy and reliability of judging motion state.

[0070] After obtaining the orientation change information, the orientation change information is mapped to the motion state determination result by interval matching. The determination interval corresponding to each motion state is obtained, and the motion state corresponding to the feature point is determined according to the determination interval and the orientation change information of the feature point.

[0071] The determination interval is used to characterize the range of orientation change information. The numerical range of orientation change is set for different motion attributes, and different numerical intervals correspond to different motion states.

[0072] In some embodiments, the motion state includes a stationary state, a non-stationary state, and an abnormal state.

[0073] A stationary state refers to a target that does not actually change its orientation; the change in orientation is mainly caused by factors such as point cloud noise and errors. A non-stationary state refers to a target that exhibits a real change in orientation, with the change in direction and position conforming to the physical laws of a normally moving target. An abnormal state refers to a target whose orientation change exceeds the reasonable range of normal movement, usually caused by abnormal factors such as point cloud data failure.

[0074] The boundary value of the judgment interval corresponding to each motion state can be dynamically determined based on the historical state of the target to be observed and / or the tracking time of the tracking unit.

[0075] Specifically, the historical state of the target to be observed and / or the tracking time of the tracking unit are obtained; if the historical state is a non-static state, the preset lower limit value of the preset interval corresponding to the non-static state is lowered by a preset adjustment value; and / or, based on the tracking time, the preset lower limit value of the preset interval is lowered by the target compensation value.

[0076] Historical state refers to the motion state of the target being observed within the historical period prior to the current observation time. Tracking time refers to the number of observation frames or cumulative duration experienced by the tracking unit from its creation time to the current time.

[0077] The preset lower limit is a pre-set initial benchmark threshold for determining non-stationary states, while the preset adjustment value is a fixed compensation amount set for historical non-stationary states. Since target motion is generally continuous, the judgment criteria for observed targets already judged as moving are appropriately relaxed based on historical states to avoid misjudging temporary deceleration or point cloud fluctuations as stationary. Therefore, lowering the preset lower limit makes it easier for historically moving targets to meet the non-stationary state judgment conditions.

[0078] Optionally, a historical cumulative map is constructed based on the historical frame point cloud to store the occupancy status of the grid after multi-frame fusion. The occupancy probability of each height interval is updated using a Bayesian filtering method, while the historical state of the target to be observed is recorded to provide temporal information for motion state determination. Bayesian filtering theory has a solid mathematical foundation and good interpretability, making it suitable for application in rule-based frameworks.

[0079] Tracking time is used to increase the confidence level of continuously tracked targets and reduce the probability of stable targets being mistakenly dropped. The target compensation value is a compensation amount dynamically calculated based on the tracking time. The longer the tracking time, the larger the target compensation value, and the greater the reduction in the preset lower limit. The longer the tracking unit is tracked, the lower the motion judgment threshold is, and the more consistent the state of the stable tracked target is.

[0080] Furthermore, the target upper limit is determined based on the adjusted preset lower limit. The target upper limit is the upper boundary of the static state determination interval. When the preset lower limit is lowered due to compensation, the corresponding target upper limit will also decrease, thereby maintaining the relative proportional stability of the determination interval.

[0081] Based on the adjusted preset lower limit and target upper limit, the judgment intervals corresponding to the three states can be obtained. Specifically, based on the adjusted preset lower limit and target upper limit, the first judgment interval corresponding to the non-static state is determined; the adjusted preset lower limit is determined as the upper limit of the second judgment interval corresponding to the static state; and the target upper limit is determined as the lower limit of the third judgment interval corresponding to the abnormal state.

[0082] The first judgment interval corresponds to a non-static state, indicating that the target under observation is experiencing real and reasonable motion. The second judgment interval corresponds to a static state; the change in azimuth in this interval is less than the adjusted preset lower limit, indicating that the change in the point cloud distribution of the target under observation is relatively small, and is most likely caused by random factors such as noise. The third judgment interval corresponds to an abnormal state; the change in azimuth in this interval is greater than the target's upper limit, indicating that the change in the point cloud distribution of the target under observation is too large, exceeding the reasonable range, and that there is tracking or data anomaly.

[0083] The electronic device determines the corresponding judgment interval based on the orientation change information of the feature point, and then determines the motion state corresponding to the judgment interval as the motion state of the feature point.

[0084] For example, the motion score of a feature point is calculated based on its orientation change information, and the motion state corresponding to the judgment interval into which the motion score falls is determined as the motion state of the feature point. The motion score is positively correlated with the change in direction and / or the change in position; for example, it can be obtained by weighting the normalized changes in direction and position.

[0085] The embodiments disclosed herein dynamically determine the judgment intervals corresponding to the three motion states by means of the historical state of the target to be observed and / or the tracking time of the tracking unit, so that the motion state judgment results of the feature points are more in line with the actual motion laws, improve the scene adaptability and robustness of motion state judgment, and can effectively reduce the probability of static false detection and dynamic false detection.

[0086] In some embodiments, when multiple types of feature points exist, if the motion states corresponding to the multiple types of feature points are consistent, then the motion state corresponding to the feature point is determined to be the motion state of the target to be observed; or, if the motion states corresponding to the multiple types of feature points are inconsistent, then the motion state of the target to be observed is determined to be a state to be confirmed.

[0087] If multiple types of feature points exist, and the motion state corresponding to each type of feature point is determined, then if the motion states corresponding to each type of feature point are consistent, the motion state of the target to be observed can be directly determined. For example, if each type of feature point is determined to be in a non-stationary state, then the motion state of the target to be observed is determined to be a non-stationary state.

[0088] If the motion states corresponding to multiple feature points are inconsistent, it means that there is a discrepancy in the motion state determination results of the target to be observed in the current frame. This indicates that the observation data of the current frame may be subject to noise interference, the target may be in the dynamic-static critical range, or there may be a local deviation in the tracking process. It is not possible to give a definite conclusion at this time, and further verification is needed by combining the observation data of subsequent frames.

[0089] This embodiment of the disclosure avoids random errors in the determination results of a single type of feature point by performing consistency verification on the motion states corresponding to multiple types of feature points. When the motion states are consistent, the motion state of the target to be observed is determined efficiently. When the motion states are inconsistent, the state to be confirmed is used as a buffer. This ensures both the determination efficiency in normal scenarios and the reliability of the determination results in special scenarios.

[0090] In some embodiments, the method described above for determining the motion state of a target can be used to correct the detection results obtained by conventional target detection algorithms.

[0091] Conventional target detection algorithms output the detected target and its speed or motion state.

[0092] In conventional target detection algorithms, clustering algorithms have limited ability to handle targets with complex shapes, resulting in incomplete target segmentation; or deep learning detection models have detection blind spots in specific scenarios, insufficient ability to detect certain target types or poses, and limited generalization ability in scenarios with insufficient or unevenly distributed training data, resulting in missed target detection.

[0093] Furthermore, the instability of clustering algorithms in conventional target detection algorithms can lead to changes in target shape and position, affecting the accuracy of velocity estimation. Additionally, target matching algorithms are prone to mismatches, causing misalignments between the target and its historical trajectory. Alternatively, slow-moving or temporarily stationary dynamic targets are easily misclassified as static, resulting in incorrect detection information.

[0094] Figure 2 This is a flowchart illustrating a method for determining the motion state of a target according to another embodiment of this disclosure. Figure 2 As shown, the method includes the following steps.

[0095] S201. Map the current frame point cloud onto the grid, and determine the tracking unit of the current frame based on the grid corresponding to the current frame point cloud and the tracking unit of the previous frame.

[0096] S202. Determine the motion state of the target to be observed in the tracking unit.

[0097] S203. If the target to be observed cannot be associated with any of the detected targets, then the point cloud in the tracking unit will be re-clustered.

[0098] One possible scenario is that if the target to be observed cannot be associated with any detected target (e.g., the intersection-union ratio with any detected target is less than a preset value), then it is considered that the conventional target detection algorithm may have missed detections. In this case, the point cloud in the tracking unit is re-clustered, and the re-clustered targets and the detected targets are output together to the downstream for decision-making on whether to avoid them. This allows missed targets to be detected and tracked in a timely manner, especially for targets to be observed in a non-stationary state, thus improving the security of the perception system.

[0099] S204. Correct the motion state of the target to be observed based on the motion state of the target to be detected.

[0100] Optionally, the conventional target detection algorithm outputs the detected target and determines that its movement speed is greater than a threshold. If the rate of change of the occupancy state of the detected target and the grid within a certain distance in its movement direction is less than the preset threshold in the past preset time period, and the target is occupied most of the time, then the movement speed of the detected target is corrected to 0.

[0101] Optionally, conventional target detection algorithms output the detected target and determine it to be in a stationary state. If the intersection-union ratio (IU) of the detected target is greater than or equal to a preset ratio, the observed target is determined to be in a non-stationary state, and the motion state of the detected target is corrected to a motion state.

[0102] In this embodiment of the disclosure, the method for determining the motion state of a target can be integrated as an independent algorithm module with the target detection algorithm in the existing rule framework, and each step can also be an independent algorithm module, which is convenient for individual debugging and optimization.

[0103] Figure 3 This is a schematic diagram of the structure of a device for determining the motion state of a target according to an embodiment of this disclosure.

[0104] In this embodiment of the disclosure, the device for determining the motion state of the target can be housed within an electronic device, and is understood as a functional module within the aforementioned electronic device. Specifically, the electronic device can be a server or a terminal, wherein the terminal specifically includes in-vehicle devices, mobile phones, computers, or tablet computers, etc., without limitation.

[0105] like Figure 3As shown, the device 30 for determining the motion state of a target may include a first acquisition module 31, a first determination module 32, a second determination module 33, a second acquisition module 34, and a third determination module 35. The first acquisition module 31 acquires at least one tracking unit corresponding to a historical frame point cloud, wherein the historical frame point cloud is mapped onto a preset grid, and the tracking unit consists of at least one grid, including the point cloud corresponding to the target to be observed. The first determination module 32 maps the current frame point cloud onto the grid and determines the tracking unit of the current frame based on the grid corresponding to the current frame point cloud and the tracking unit of the previous frame. The second determination module 33 determines the point cloud distribution characteristics within the tracking unit of the current frame. The second acquisition module 34 acquires the point cloud distribution characteristics within the tracking unit of a historical frame. The third determination module 35 determines the motion state of the target to be observed based on the point cloud distribution characteristics of the current frame and the point cloud distribution characteristics of the historical frames.

[0106] In some embodiments, the first determining module 32 is used to determine whether there is a point cloud in the target grid in the multiple grids corresponding to the point cloud of the current frame, where the distance to the tracking unit of the previous frame is less than a preset distance threshold; if yes, then update the tracking unit of the previous frame so that the point cloud in the target grid is included in the updated tracking unit, and use the updated tracking unit as the tracking unit of the current frame; if no, then use the tracking unit of the previous frame as the tracking unit of the current frame.

[0107] In some embodiments, the second determining module 33 includes a first determining unit, configured to determine the position of at least one type of feature point within the tracking unit of the current frame based on the point cloud within the tracking unit of the current frame, wherein the position of the feature point is used to characterize the point cloud distribution features of the point cloud within the tracking unit; and a second obtaining module 34, configured to obtain the position of at least one type of feature point within the tracking unit of a historical frame.

[0108] In some embodiments, the first determining unit is used to determine the position of the centroid of the point cloud within the tracking unit as the position of a first type of feature point; and / or, to determine the position of the geometric center of the convex hull of the point cloud within the tracking unit as the position of a second type of feature point; and / or, to determine the average position of the geometric centers of multiple point cloud segments in the tracking unit as the position of a third type of feature point, wherein the multiple point cloud segments are obtained by segmenting the tracking unit in the vertical direction based on a preset height value, and the height value of each point cloud segment is a preset height value.

[0109] In some embodiments, the third determining module 35 includes a second determining unit, a third determining unit, and a fourth determining unit; the second determining unit is used to determine the motion state of the target to be observed based on the point cloud distribution features of the current frame and the point cloud distribution features of historical frames, including: the third determining unit is used to determine the orientation change information of each type of feature point based on the position of the feature point in the current frame and the position of the feature point in the historical frame, and to determine the motion state corresponding to the feature point based on the orientation change information; the fourth determining unit is used to determine the motion state of the target to be observed based on the motion state corresponding to at least one type of feature point.

[0110] In some embodiments, the orientation change information includes the change in direction and / or the change in position. The third determining unit is used to determine the maximum value of the angle corresponding to the feature point in the current frame and the angle corresponding to the feature point in the historical frame as the orientation change information of the feature point. The angle corresponding to the feature point in the current frame is the angle between the displacement vector corresponding to the current frame and the displacement vector corresponding to the previous frame. The displacement vector corresponding to the current frame is a vector with the feature point in the current frame as the endpoint and the feature point in the previous frame as the starting point. And / or, the maximum value of the displacement of the feature point in the current frame and the displacement of the feature point in the historical frame is determined as the change in position of the feature point. The displacement of the feature point in the current frame is the displacement distance between the feature point in the current frame and the feature point in the previous frame.

[0111] In some embodiments, the third determining unit is further configured to obtain a determination interval corresponding to each motion state, the determination interval being used to characterize the range of orientation change information; and to determine the motion state corresponding to the feature point based on the determination interval and the orientation change information of the feature point.

[0112] In some embodiments, the fourth determining unit is used to determine the motion state corresponding to the feature point as the motion state of the target to be observed if the motion states corresponding to the multiple types of feature points are consistent, and to determine the motion state of the target to be observed as a state to be confirmed if the motion states corresponding to the multiple types of feature points are inconsistent.

[0113] In some embodiments, the motion state includes a stationary state, a non-stationary state, and an abnormal state. The third determining unit is further configured to acquire the historical state of the target to be observed and / or the tracking time of the tracking unit, wherein the historical state is the motion state of the target to be observed at a historical observation time; when the historical state is a non-stationary state, the preset lower limit value of the preset interval corresponding to the non-stationary state is lowered by a preset adjustment value; and / or, according to the tracking time, the preset lower limit value of the preset interval is lowered by a target compensation value, wherein the tracking time is positively correlated with the target compensation value; the target upper limit value is determined according to the adjusted preset lower limit value, wherein the target upper limit value is a preset multiple of the adjusted preset lower limit value; based on the adjusted preset lower limit value and the target upper limit value, a first determination interval corresponding to the non-stationary state is determined; the adjusted preset lower limit value is determined as the upper limit value of the second determination interval corresponding to the stationary state, and the target upper limit value is determined as the lower limit value of the third determination interval corresponding to the abnormal state.

[0114] It should be noted that, Figure 3 The device 30 for determining the motion state of the target shown can execute the various steps in the above method embodiments and achieve the various processes and effects in the above method embodiments, which will not be elaborated here.

[0115] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure.

[0116] In this embodiment of the disclosure, Figure 4 The electronic devices shown can be servers or terminals. Specifically, terminals include in-vehicle devices, mobile phones, computers, or tablets, etc., without limitation.

[0117] like Figure 4 As shown, the electronic device may include a processor 410 and a memory 420 storing computer program instructions.

[0118] Specifically, the processor 410 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this disclosure.

[0119] Memory 420 may include a large-capacity storage device for information or instructions. For example, and not limitingly, memory 420 may include a hard disk drive (HDD), a floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 420 may include removable or non-removable (or fixed) media. Where appropriate, memory 420 may be internal or external to the integrated gateway device. In a particular embodiment, memory 420 is a non-volatile solid-state memory. In a particular embodiment, memory 420 includes read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (Electrically Programmable ROM, EPROM), an electrically erasable programmable PROM (EEPROM), an electrically alterable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0120] The processor 410 reads and executes computer program instructions stored in the memory 420 to perform the steps of the method for determining the motion state of a target provided in the embodiments of this disclosure.

[0121] In one example, the electronic device may also include a transceiver 430 and a bus 440. Wherein, as... Figure 4 As shown, the processor 410, memory 420 and transceiver 430 are connected via bus 440 and communicate with each other.

[0122] Bus 440 may include hardware, software, or both. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industrial Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a MicroChannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 440 may include one or more buses.

[0123] This disclosure also provides a computer-readable storage medium that can store a computer program that, when executed by a processor, enables the processor to implement the method for determining the motion state of a target provided in this disclosure.

[0124] The aforementioned storage medium may, for example, include a memory 420 containing computer program instructions, which can be executed by a processor 410 of an electronic device to complete the method for determining the motion state of a target provided in the embodiments of this disclosure. Optionally, the storage medium may be a non-transitory computer-readable storage medium, such as a ROM, random access memory (RAM), compact disc-only memory (CD-ROM), magnetic tape, floppy disk, and optical data storage device.

[0125] This disclosure also provides a vehicle that includes electronic devices that can implement the various processes and effects described in the above embodiments of this disclosure, which will not be elaborated here.

[0126] This disclosure also provides a computer program product, which includes a computer program or instructions. When the computer program or instructions are executed by a processor, they implement the method for determining the motion state of a target provided in this disclosure, and can achieve the various processes and effects in the above embodiments of this disclosure, which will not be elaborated here.

[0127] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for determining the motion state of a target, characterized in that, include: A tracking unit for a historical frame corresponding to at least one historical frame point cloud is obtained, wherein the historical frame point cloud is mapped in a preset grid, the tracking unit is composed of at least one grid, and the tracking unit includes the point cloud corresponding to the target to be observed; Map the current frame point cloud onto the grid, and determine the tracking unit of the current frame based on the grid corresponding to the current frame point cloud and the tracking unit of the previous frame; Determine the point cloud distribution characteristics of the point cloud within the tracking unit of the current frame; Obtain the point cloud distribution characteristics of the point cloud within the tracking unit of the historical frame; The motion state of the target to be observed is determined based on the point cloud distribution characteristics of the current frame and the point cloud distribution characteristics of historical frames.

2. The method according to claim 1, characterized in that, The step of determining the tracking unit for the current frame based on the grid corresponding to the point cloud in the current frame and the tracking unit in the previous frame includes: Determine whether there is a point cloud in the target grid within the multiple grids corresponding to the point cloud in the current frame, where the distance to the tracking unit in the previous frame is less than a preset distance threshold; If so, update the tracking unit of the previous frame so that the point cloud in the target mesh is included in the updated tracking unit, and use the updated tracking unit as the tracking unit of the current frame; If not, the tracking unit of the previous frame is used as the tracking unit of the current frame.

3. The method according to claim 1, characterized in that, Determining the point cloud distribution features within the tracking unit of the current frame includes: Based on the point cloud within the tracking unit of the current frame, determine the position of at least one type of feature point within the tracking unit of the current frame. The position of the feature point is used to characterize the point cloud distribution features of the point cloud within the tracking unit. The point cloud distribution features of the point cloud within the tracking unit that acquires historical frames include: The position of at least one type of feature point within the tracking unit of the historical frame is obtained.

4. The method according to claim 3, characterized in that, Determining the position of at least one type of feature point within the tracking unit of the current frame includes: The position of the centroid of the point cloud within the tracking unit is determined as the position of the first type of feature point; and / or, The position of the geometric center of the convex hull bounding box of the point cloud within the tracking unit is determined as the position of the second type of feature point; and / or, The average position of the geometric center of multiple point cloud segments in the tracking unit is determined as the position of the third type of feature point. The multiple point cloud segments are obtained by dividing the tracking unit in the vertical direction based on a preset height value, and the height value of each point cloud segment is the preset height value.

5. The method according to claim 3, characterized in that, Determining the motion state of the target to be observed based on the point cloud distribution characteristics of the current frame and the point cloud distribution characteristics of historical frames includes: For each type of feature point, the orientation change information of the feature point is determined based on the position of the feature point in the current frame and the position of the feature point in the historical frames, and the motion state corresponding to the feature point is determined based on the orientation change information. The motion state of the target to be observed is determined based on the motion state corresponding to the at least one type of feature points.

6. The method according to claim 5, characterized in that, The orientation change information includes the amount of direction change and / or position change. The step of determining the orientation change information of the feature point based on the position of the feature point in the current frame and the position of the feature point in historical frames includes: The maximum value of the angle corresponding to the feature point in the current frame and the angle corresponding to the feature point in the historical frame is determined as the direction change information of the feature point. The angle corresponding to the feature point in the current frame is the angle between the displacement vector corresponding to the current frame and the displacement vector corresponding to the previous frame. The displacement vector corresponding to the current frame is a vector with the feature point in the current frame as the endpoint and the feature point in the previous frame as the starting point. And / or, The maximum value between the displacement of the feature point in the current frame and the displacement of the feature point in the historical frame is determined as the position change of the feature point. The displacement of the feature point in the current frame is the displacement distance between the feature point in the current frame and the feature point in the previous frame.

7. The method according to claim 5, characterized in that, Determining the motion state corresponding to the feature point based on the orientation change information includes: Obtain the determination interval corresponding to each motion state, and the determination interval is used to characterize the range of the orientation change information; Based on the determination interval and the orientation change information of the feature point, the motion state corresponding to the feature point is determined.

8. The method according to claim 5, characterized in that, Determining the motion state of the target to be observed based on the motion states corresponding to the at least one type of feature points includes: In the presence of multiple types of feature points, if the motion states corresponding to the multiple types of feature points are consistent, then the motion state corresponding to the feature point is determined to be the motion state of the target to be observed. If the motion states corresponding to the various feature points are inconsistent, then the motion state of the target to be observed is determined to be a state to be confirmed.

9. The method according to claim 7, characterized in that, Motion states include stationary state, non-stationary state, and abnormal state. The step of obtaining the determination interval corresponding to each motion state includes: The historical state of the target to be observed and / or the tracking time of the tracking unit are obtained, wherein the historical state is the motion state of the target to be observed at the historical observation time; When the historical state is a non-static state, the preset lower limit value of the preset interval corresponding to the non-static state is lowered by a preset adjustment value; and / or, according to the tracking time, the preset lower limit value of the preset interval is lowered by a target compensation value, wherein the tracking time is positively correlated with the target compensation value; The target upper limit is determined based on the adjusted preset lower limit, wherein the target upper limit is a preset multiple of the adjusted preset lower limit; Based on the adjusted preset lower limit and the target upper limit, a first determination interval corresponding to the non-static state is determined; The adjusted preset lower limit value is determined as the upper limit value of the second judgment interval corresponding to the static state, and the target upper limit value is determined as the lower limit value of the third judgment interval corresponding to the abnormal state.

10. An electronic device, characterized in that, include: processor; Memory, used to store executable instructions; The processor is configured to read the executable instructions from the memory and execute the executable instructions to implement the method of any one of claims 1-9.

Images