Target state determination method, determination system, and storage medium

Abstract

Embodiments of the present application provide a target state determination method, a determination system and a storage medium, belonging to the technical field of automatic driving. The method comprises: in a parking process, collecting the moving track information of a judgment target in real time; based on a track point caching algorithm, caching the moving track information of the judgment target; based on the cached moving track information of the judgment target, respectively executing the moving state judgment corresponding to the judgment target; and based on the current moving state of the judgment target, determining the dynamic and static state of the current judgment target. The present application overcomes the problem of incorrect determination of the dynamic and static state of the target caused by the fact that the sensor cannot accurately perceive the moving state of the low-speed target at low speed.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, specifically to a target state determination method, a target state determination system, and a storage medium. Background Technology

[0002] In existing automated parking technology, the vehicle often needs to assess the motion and stillness of surrounding targets to avoid misjudging their motion and stillness during low-speed parking, which could lead to collisions later. If a target is directly judged as stationary, subsequent parking will be based solely on previously collected parking space information. However, a target judged as stationary may still be moving at a very low speed, altering the surrounding environment and potentially causing a collision. Current target motion and stillness assessment solutions primarily rely on sensor information to determine whether a target is moving. However, when the actual target's speed is close to the sensor error and the acceleration is very small, accurately determining its motion and stillness becomes extremely difficult. Besides automated parking, for other low-speed targets, such as moving targets during deceleration parking, passing other vehicles, and exiting parking, the slow movement and minimal change in the target's posture during these actions make it difficult for current sensors to accurately determine the target's motion and stillness within acceptable error margins. Given the limited accuracy of sensors, assessing the motion and stillness of targets in these low-speed driving scenarios is extremely challenging. To address the issue that existing solutions are not accurate enough in determining the dynamic and static states of low-speed targets, a new target state determination scheme needs to be created. Summary of the Invention

[0003] The purpose of this invention is to provide a target state determination method, determination system, and storage medium to solve the problem that existing solutions have low accuracy in determining the dynamic and static states of low-speed targets.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] The first aspect of the present invention provides a target state determination method applied to a vehicle. The target state determination method includes: during driving, collecting and determining the movement trajectory information of the target in real time; caching the movement trajectory information of the target based on a trajectory point caching algorithm; performing movement state determination of the corresponding target based on the cached movement trajectory information of the target; and determining whether the current target is a stationary target or a moving target based on the current movement state of the target.

[0006] Optionally, the target's movement trajectory information includes: global path information, cropped local path information, left and right boundary information, and sensor-collected information.

[0007] Optionally, the cropped local path information and left and right boundary information are cached using multiple historical trajectory points; the sensor-collected information includes: the longitudinal distance, lateral distance, heading angle, target category, and vehicle positioning of the identified target and the vehicle.

[0008] Optionally, the trajectory point caching algorithm includes: defining three initial arrays, each with a size equal to the number of historical trajectory points collected, used to store the x-coordinate, y-coordinate, and index of each historical trajectory point; defining a field for caching previously recorded targets, denoted as the historical target field; traversing all targets and determining whether the identity identifier of each target is in the valid identifier sequence; if so, the target is valid, otherwise the target is invalid; wherein, the valid identifier sequence includes the identity identifiers of all preset valid targets; determining whether the identity identifier of each target is the same as the identity identifier of the target recorded in the historical target field of the previously recorded targets; if so, the target is a cached target, otherwise, the target is determined to be a new target; caching all new targets and assigning initial array values ​​to each cached new target.

[0009] Optionally, caching all new targets and assigning initial array values ​​to each cached new target includes: filling the horizontal and vertical coordinate arrays of historical trajectory points based on the coordinate information of the current new target; adding an index to the current new target; the method further includes: determining whether the distance between the current new target and the previous recorded target is greater than a preset distance; if it is greater than the preset distance, then updating the initial arrays of all previous targets based on the current target; otherwise, only updating the initial array of the current target.

[0010] Optionally, determining the target's movement status includes: determining the target's speed, determining the curvature of the target's route, determining whether the target is within a lane, and determining the target's movement distance; wherein, the rule for determining the target's speed is as follows: the difference between the horizontal and vertical coordinates of all historical trajectory points is calculated to obtain the Euclidean distance between every two historical trajectory points, and the obtained Euclidean distances are accumulated to obtain the Euclidean distance sum; the difference between the counting period is calculated, and the sum is multiplied by a preset sampling period to obtain the cumulative time; the Euclidean distance sum is used as the total path, and the cumulative time is used as the total time to determine the target's movement speed.

[0011] Optionally, the rule for determining the road curvature is as follows: based on the current target, the nearest lane boundary point is determined using global coordinates; based on the current nearest lane boundary point, multiple lane boundary points are searched to the left and right respectively, and three lane boundary points are selected to form a circumcircle, obtaining the center and radius of the circumcircle; the road curvature of the current target is determined based on the center and radius of the circumcircle; the rule for determining whether the target is within the lane is as follows: based on the current target, the nearest lane boundary point is determined using global coordinates; a point set is formed using the coordinates of the current target and the nearest lane boundary point, and a contour is drawn based on the point set, determining whether the coordinates of the current target are within the contour; if they are within the contour, the current target is determined to be within the lane; otherwise, the current target is determined to be outside the lane; the rule for determining the movement distance is as follows: a contour is drawn based on all historical trajectory points, and contour points are extracted; based on the pairwise comparison of the extracted contour points, the two contour points with the largest distance are found, and the distance between these two contour points is used as the movement distance.

[0012] Optionally, the determination of the current target's motion state based on its movement status includes: if the target is located within the lane, comparing the target's movement distance with a first preset distance, and comparing the target's speed with a first preset speed; if the movement distance is greater than the first preset distance and the speed is greater than the first preset speed, then the target is determined to be a moving target; if the target is located within the lane, comparing the target's movement distance with the first preset distance, and comparing the target's speed with a second preset speed; if the movement distance is less than the first preset distance and the speed is less than the second preset speed, then the target is determined to be a stationary target; wherein, the first preset speed is greater than the second preset speed; if the movement distance is greater than the first preset distance, but the speed is greater than the first preset speed and less than the second preset speed, the target is still determined to be a moving target; if the target is located outside the lane, comparing the target's movement distance with the second preset distance, and comparing the target's speed with a third preset speed; if the movement distance is greater than the second preset distance and the speed is greater than the third preset speed, then the target is determined to be a moving target; otherwise, the target is determined to be a stationary target.

[0013] A second aspect of the present invention provides a target state determination system, the target state determination system comprising: a data acquisition unit for acquiring, in real time, the movement trajectory information of the target during driving; a processing unit for: caching the movement trajectory information of the target based on a trajectory point caching algorithm; and performing movement state determination of the corresponding target based on the cached movement trajectory information of the target; and a determination unit for determining the dynamic or static state of the target based on the current movement state of the target.

[0014] In another aspect, the present invention provides a computer-readable storage medium storing instructions which, when executed on a computer, cause the computer to perform the aforementioned target state determination method.

[0015] The beneficial effects of this invention are as follows: This invention proposes a novel method for determining the state of a target. The method collects and caches the target's movement trajectory information within a short parking period. The movement characteristics of the cached trajectory points are then used to determine the target's dynamic or static state. This invention determines the target's movement state based on the target's movement trajectory information within the sampling period, specifically based on the target's absolute position and attitude changes during that period. It relies on the cached trajectory within the sampling period and does not depend on the sensor's acquisition accuracy. This differs from existing technologies that use sensors in conjunction with moving target monitoring algorithms to determine the target's movement state, thus avoiding errors caused by sensor failure, malfunction, or insufficient accuracy. This overcomes the problem of misjudging the target's dynamic or static state at low speeds because sensors cannot accurately perceive the movement of low-speed targets. Attached Figure Description

[0016] Figure 1 A flowchart illustrating the steps of a target state determination method according to one embodiment of the present invention;

[0017] Figure 2 A flowchart illustrating the steps for determining the target movement state according to one embodiment of the present invention;

[0018] Figure 3 This is a system structure diagram of a target state determination system provided in one embodiment of the present invention. Detailed Implementation

[0019] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0020] In existing automated parking technology, the vehicle often needs to assess the motion and stillness of surrounding objects to avoid potential collisions during low-speed parking, where misjudgments of these objects could occur. This is because if a target is definitively identified as stationary, subsequent parking will be based solely on the previously collected parking space information. However, a target deemed stationary may still be moving at a very low speed, altering the surrounding environment and potentially causing a collision.

[0021] Current target motion / static determination schemes primarily rely on sensor information to determine target movement. However, when the actual target's speed is close to the sensing error and the acceleration is very small, accurately determining its motion / static status becomes extremely difficult. Besides automatic parking, for other low-speed targets, such as the process of a moving target slowing down to park, passing oncoming traffic, or a vehicle exiting a parking space, the target's movement speed is slow and its attitude changes minimal during these actions, making it difficult for current sensors to accurately determine the target's motion / static status within acceptable error margins. Given the limited accuracy of sensors, determining the motion / static status of targets in such low-speed driving scenarios is quite challenging.

[0022] To address the issue of low accuracy in judging the motion and stillness of low-speed targets using existing solutions, this invention proposes a novel target state determination method, system, and storage medium. This invention collects and caches the target's movement trajectory information during a short parking period. The motion characteristics of the cached trajectory points are then used to determine the target's motion and stillness. This overcomes the problem that sensors cannot accurately perceive the movement of low-speed targets, leading to inaccurate judgments of target motion and stillness.

[0023] Figure 1 This is a flowchart of a target state determination method provided in one embodiment of the present invention. Figure 1 As shown, an embodiment of the present invention provides a target state determination method, the method comprising:

[0024] Step S10: During the parking process, collect and judge the target's movement trajectory information in real time.

[0025] Specifically, the target's movement trajectory information includes: global path information, cropped local path information, left and right boundary information, and sensor-collected information.

[0026] Preferably, the cropped local path information and left and right boundary information are cached using multiple historical trajectory points; the sensor-collected information includes: the longitudinal distance, lateral distance, heading angle, target category, and vehicle positioning of the identified target and the vehicle.

[0027] In one possible implementation, the collected movement trajectory information includes: (1) global path planning and local path pruning; (2) local paths and their corresponding left and right boundaries; and (3) the target detected by sensors. The target detection data format is as follows: target tracking ID (track_id), longitudinal distance, lateral distance, heading angle, target category, and current vehicle location. Overall, the global coordinates of the target vehicle are calculated based on the current vehicle position and the longitudinal and lateral distances.

[0028] In this embodiment of the invention, the solution completely abandons the existing method of judging target motion state through sensors. Instead, it caches the historical trajectory points of the target in real time and then judges the target motion state based on the trajectory over a period of time. Although the target's moving speed is small, as long as it is moving within a certain period of time, it will inevitably form corresponding movement trajectory information. By using the correspondence between the cached movement trajectory information and the judgment time, the target's movement state is judged, thereby realizing the determination of the target's dynamic and static state. This method only requires sensors to record the target trajectory points at each moment, without requiring sensors to judge its movement state. Therefore, it can overcome the problem of inaccurate judgment of the state of low-speed targets caused by sensor errors.

[0029] Step S20: Cache the target's movement trajectory information based on the trajectory point caching algorithm.

[0030] Specifically, the trajectory point caching algorithm includes: defining three initial arrays, each with a size equal to the number of historical trajectory points collected, used to store the x-coordinate, y-coordinate, and index of each historical trajectory point; defining a field for caching previously recorded targets, denoted as the historical target field (e.g., lastObjectsInfo); traversing all targets and determining whether the identity identifier of each target is in the valid identifier sequence; if so, the target is valid, otherwise the target is invalid; wherein, the valid identifier sequence includes the identity identifiers of all preset valid targets; determining whether the identity identifier of each target is the same as the identity identifier of the target recorded in the historical target field of the previously recorded targets; if so, the target is a cached target, otherwise, the target is determined to be a new target.

[0031] In one possible implementation, the target's identity is set as a target ID. Because it is necessary to ensure that the identified target is valid, when caching subsequent trajectory points, it is necessary to ensure that the identified trajectory points are in a predetermined trajectory point sequence. For example, all trajectory points in a predetermined area are assigned IDs with specific values. The IDs of each trajectory point are different and are all positive integers. Then, it is only necessary to determine that the target ID is greater than 0 to determine that the target is a valid target. In this embodiment, it is determined whether the ID of each target is greater than 0. If it is, the target is valid; otherwise, the target is invalid. Then, it is determined whether the ID of each target is equal to the target ID recorded in the lastObjectsInfo of each target recorded last time. If it is, it means that the target is a cached target; otherwise, it is a new target. All new targets are cached, and an initial array is assigned to each cached new target.

[0032] Furthermore, the step of caching all new targets and assigning initial array values ​​to each cached new target includes: filling the horizontal and vertical coordinate arrays of historical trajectory points based on the coordinate information of the current new target; adding an index to the current new target; the method also includes: determining whether the distance between the current new target and the previously recorded target is greater than a preset distance; if it is greater, then updating the initial array of all previous targets based on the current target; otherwise, only updating the initial array of the current target.

[0033] In one possible implementation, the sensor receives data over a period of 20ms, buffering for a few seconds, resulting in hundreds of trajectory points. Most of these points are adjacent and therefore insignificant. Therefore, 50 historical trajectory points are buffered. The trajectory point buffering algorithm is as follows:

[0034] First, define three arrays, each with a size of 50. Name them history_x

[50] , history_y

[50] , and history_indx

[50] respectively; they will store the x-coordinate, y-coordinate, and index. Initialize all values ​​to 0; define lastObjectsInfo to cache the previously recorded targets. So initially end = 50, TrackLength = 50.

[0035] Then, iterate through the targets, checking if track_id > 0, indicating the target is valid. Further, check if track_id equals the target's track_id cached in lastObjectsInfo. If they equal, it means a target was found, and therefore not a new target; if they don't equal, it means a new target was added.

[0036] For a new target, the history_count is set to 1, indicating that the previous number of historical tracks is 0. The current position coordinates x and y are stored as history_x[end] and history_y[end], respectively, and the historical index history_indx is 0. The cache index HistoryStartIndex for these 50 data points is set to TrackLength.

[0037] For previous targets, the current coordinates and the last recorded coordinates of this trajectory point must be at least 0.07m apart. Simultaneously, increment the cycle count by 1 and record the cycle count for this trajectory point, adding a new recorded cycle. Specifically, assign `history_x[2:end]` to `history_x[1:end-1]`; assign `history_y[2:end]` to `history_y[1:end-1]`; and assign `history_indx[2:end]` to `history_indx[1:end-1]`. Store the current position coordinates x and y as `history_x[end]` and `history_y[end]`. Assign `history_indx[end]` the current cycle count. If the distance is less than 0.07m, only increment the cycle count by 1. The first 49 data points of `history_x`, `history_y`, and `history_indx` remain unchanged; only the last data point is updated. Update `history_x` and `history_y` with the current coordinates, and update `history_indx` with the current count.

[0038] In this embodiment of the invention, because the solution requires determining the target's motion state based on its continuous movement trajectory information over a period of time, it needs to combine trajectory points from consecutive moments to form a corresponding cached trajectory. To form a complete trajectory, while ensuring the validity of trajectory points on a single trajectory, the spacing between selected trajectory points must also meet a preset threshold to guarantee the validity of the trajectory points. The solution of this invention is based on a trajectory point caching algorithm, continuously reviewing, filtering, and caching target points to facilitate subsequent target motion state determination and ensure the reliability of the determination results.

[0039] Step S30: Based on the cached movement trajectory information of the target, perform the movement status judgment of the corresponding target respectively.

[0040] Specifically, determining the target's movement status includes: determining the target's speed, determining the curvature of the target's path, determining whether the target is within the lane, and determining the target's movement distance; wherein, for example... Figure 2 The corresponding judgment rules are as follows:

[0041] Step S301: Determine the target's speed.

[0042] Specifically, the difference between the horizontal and vertical coordinates of all historical trajectory points is calculated to obtain the Euclidean distance between every two trajectory points, and the sum of the Euclidean distances is obtained. The difference between the counting periods is calculated, and the sum is multiplied by the preset sampling period to obtain the cumulative time. The sum of the Euclidean distances is used as the total path, and the cumulative time is used as the total time to determine the moving speed of the target.

[0043] In one possible implementation, the recalculation of speed involves first differentiating the x and y coordinates of the historical trajectory, then calculating the Euclidean distance between every two trajectory points in the difference results, and finally summing them up. This is the cumulative distance traveled. Next, the difference is calculated over the counting period `history_indx`, and the sum is multiplied by the sampling period of 20ms to obtain the corresponding time. Speed ​​is then calculated by dividing the cumulative distance traveled in the previous two steps by the cumulative time.

[0044] In this embodiment of the invention, a crucial aspect of determining the motion of a target is assessing its movement speed. If the target exhibits movement speed exceeding a preset threshold, it is deemed to be in motion. This invention's solution determines target movement speed based on cached historical trajectory distance and sampling time, resulting in more accurate target movement speed readings.

[0045] Step S302: Determine the curvature of the target's travel path.

[0046] Specifically, based on the current target, the nearest lane boundary point is determined using global coordinates; based on the current nearest lane boundary point, multiple lane boundary points are searched to the left and right respectively, and three lane boundary points are selected to form a circumcircle, obtaining the center and radius of the circumcircle; based on the center and radius of the circumcircle, the road curvature state of the current target is determined.

[0047] In one possible implementation, the goal is determined to be on a straight road or a curve. The core idea is to search for the nearest left and right lane boundary points using the target's global coordinates, and then search for 10 lane boundary points forward and backward from these nearest left and right lane boundary points. The center and radius of the circumcircle are calculated using these three lane boundary points. If the radius is less than 10m, it is considered a curve. An exception occurs when the search for the nearest left and right lane boundary points is the same as the 10 lane boundary points before or after the stored boundary. In this case, the center and radius of the circumcircle are calculated using the last or first three lane boundary points (numbers 1, 11, and 21). The calculation method for the center and radius of the circumcircle of the three lane boundary points is as follows:

[0048] A=x1(y2-y3)+y1(x2-x3)+x2y3-x3y2

[0049] B = (x1) 2 +y1 2 (y3-y2)+(x2) 2 +y2 2 (y1-y3)+(x3) 2 +y3 2 (y2-y1)

[0050] C=(x1 2 +y1 2 (x2-x3)+(x2) 2 +y2 2 (x3-x1)+(x3) 2 +y3 2 (x1-x2)

[0051] D=(x1 2 +y1 2 (x3y2-x2y3)+(x2 2 +y2 2 (x1y3-x3y1)+(x3 2 +y3 2 (x2y1-x1y2)

[0052] Then the coordinates of the center (x, y) satisfy:

[0053]

[0054]

[0055] The radius r satisfies:

[0056]

[0057] In this embodiment of the invention, if the target is only moving in a straight line, its motion posture is relatively easy to predict, and the possibility of subsequent avoidance is higher. However, if the target is on a curve or is parking, its vehicle posture will change accordingly, making it more difficult to predict subsequent actions such as passing other vehicles. Therefore, when judging the target's movement state, it is necessary to determine whether it is traveling on a straight road or a curve. The solution of this invention uses the circumcircle rule to judge the target's movement state, achieving accurate identification of the target's movement posture and ensuring the reliability of the dynamic / static judgment results.

[0058] Step S303: Determine whether the target is within the lane.

[0059] Specifically, based on the current target, the nearest lane boundary point is determined using global coordinates; a point set is formed by the coordinates of the current target and the nearest lane boundary point, and a contour is drawn based on the point set. It is then determined whether the coordinates of the current target are within the contour. If they are, the current target is determined to be outside the lane; otherwise, the current target is determined to be inside the lane.

[0060] In this embodiment of the invention, existing methods for determining whether a target is inside or outside the lane line use the target's global coordinates to search for the nearest left and right lane line boundary points. After searching multiple points forward and backward from these nearest boundary points, a straight line is fitted. The global coordinates are then substituted into the equation of the fitted line, and the sign of the two coordinates is determined: if both are positive, the target is outside the lane line; if they are opposite, the target is inside. This method has a problem: because it fits a straight line equation, it doesn't work on curves. To solve this problem, this invention uses the lane line boundary points plus the target's coordinates to form a point set, and then calculates a contour based on this point set. If the contour points include the target's coordinates, the target is outside the lane; otherwise, it's inside the lane line. This method is applicable to both straight and curved lanes because it performs a combined contour fitting of the target and the lane. Therefore, as long as the lane line boundary points are determined, the fitted result will always match the lane path state. This eliminates the problem of existing methods only being able to fit straight lines and not working on curves.

[0061] Preferably, if the application scenario is a parking scenario, there is a parking completion route, i.e., the movement route within the parking space, which still includes two types: straight roads (reversing parking spaces) and curved roads (parallel parking spaces, angled parking spaces). The solution of this invention addresses this situation: for the completion scenario, if it is a straight line, multiple fitted straight lines are taken, and it is determined whether the calculated value of the target and two straight lines has the same sign. If they have the same sign, it is outside the lane line; if they have different signs, it is inside the lane line. If it is a curved road, the straight lines passing through the last two lane line boundary points are taken, and it is determined whether they have the same sign. If they have the same sign, it is outside the lane line; if they have different signs, it is inside the lane line.

[0062] In one possible implementation, when performing contour fitting based on lane boundary points, the more lane boundary points there are, the better the contour simulation effect. Therefore, the more complex the driving situation, the higher the required accuracy of the contour fitting. Preferably, the curvature of the lane has already been determined. The curvature of the lane can reveal the complexity of the current driving situation; the higher the curvature, the higher the required accuracy of the contour fitting. If the number of lane boundary points collected is 60, then for preset extreme scenarios, the number of lane boundary points can be increased to 100, allowing the number of lane boundary points to be adjusted within a certain range. Of course, this preset range can be adjusted based on the user's actual needs. The more lane boundary points collected, the higher the corresponding recognition latency. Therefore, after balancing and adaptively adjusting recognition accuracy and recognition latency, 60-100 lane boundary points are preferred. By adjusting the number of lane boundary points collected within this preset range, combined with its applicability to straight and curved road fitting characteristics, the determination of whether the target is within the lane can be satisfied under all working conditions.

[0063] In this embodiment of the invention, determining whether a target is within the lane is helpful in judging its movement tendency. If it is driving normally, it tends to stay within the lane; if there is oncoming traffic or parking, it tends to move across lanes. Having already determined straight-line and curved-line driving conditions, this invention directly utilizes those prior results and sets corresponding rules for curved roads to determine whether a target is within the lane in both straight-line and curved-line scenarios, thus improving the universality of this application's solution.

[0064] Step S304: Determine the target's movement distance.

[0065] Specifically, the contour is drawn based on all historical trajectory points, and contour points are extracted; based on the pairwise comparison of the extracted contour points, the two contour points with the largest distance are found, and the distance between these two contour points is used as the movement distance.

[0066] In one possible implementation, calculating the distance traveled is inefficient, considering the car's constant movement in and out during parking. Simply recording the displacement from the start to the end point is pointless; the displacement could be very small or very large, offering little value. If selecting the two points with the largest distance from the collected historical trajectory points is necessary, pairwise comparisons must be performed on each point, iterating through all points to determine the maximum distance. The more historical trajectory points collected, the greater the computational load. For example, with 50 historical trajectory points, calculating the maximum distance would require 1250 pairwise comparisons. Therefore, this invention first calculates the outer contour, extracts the contour points, and then calculates the maximum distance pairwise. This method achieves batch processing of historical trajectory points, contour fitting, and finally retains only the outer contour points, filtering out a large number of invalid points. This significantly reduces the number of outer contour points requiring pairwise comparisons, thus greatly reducing the computational load.

[0067] In this embodiment of the invention, the present invention uses sampling period and cached trajectory points to accurately control the global movement state of the target, ensuring the control of its movement distance during reciprocating motion, thus making the dynamic and static state determination results of the present application more accurate.

[0068] Step S40: Based on the current movement state of the target, determine whether the current target is a stationary target or a moving target.

[0069] Specifically, if the target is determined to be within the lane, the moving distance of the target is compared with a first preset distance, and the speed of the target is compared with a first preset speed. If all are greater than the first preset speed, the target is determined to be a moving target. If the target is determined to be within the lane, the moving distance of the target is compared with a second preset distance, and the speed of the target is compared with a second preset speed. If all are less than the second preset speed, the target is determined to be a stationary target. If the target is determined to be outside the lane, the moving distance of the target is compared with a third preset distance, and the speed of the target is compared with a third preset speed. If all are greater than the third preset speed, the target is determined to be a moving target. Otherwise, the target is determined to be a stationary target.

[0070] In one possible implementation, the target is first identified as being inside or outside the lane lines. If inside the lane lines, for a non-moving target, it is considered moving if the distance traveled within the lane lines exceeds 1 meter, the calculated speed exceeds 1 m / s, and the trajectory gradually exceeds 2.5 meters. For a moving target, it is considered stationary if the calculated speed is less than 1.5 m / s and the distance traveled is less than 1 meter. For targets outside the lane lines, they are considered moving only if the distance traveled exceeds 6 meters, the calculated speed exceeds 5 m / s, the sensing speed exceeds 1 m / s, and the angle between the target's movement and the lane lines is between 30 and 150 degrees.

[0071] In this embodiment of the invention, the core purpose of determining the target's dynamic and static state is to assess whether it will collide with the vehicle, thus essentially evaluating the threat posed by the target. The present invention uses multiple preset comparison thresholds for dynamic and static determination to judge the target's dynamic and static state, integrating multiple front-end movement state judgment results to ensure the reliability of the judgment results and achieve accurate target state identification under low-speed driving conditions.

[0072] Furthermore, based on data collected from actual vehicle testing, calculations are performed according to established rules, and the calculated speed, distance traveled, and information regarding lane markings (within and outside lane lines) are written to a file. Through data analysis, a threshold for determining whether a vehicle is stationary or moving is determined.

[0073] Figure 3 This is a system structure diagram of a target state determination system provided in one embodiment of the present invention. For example... Figure 3 As shown, this embodiment of the invention provides a target state determination system, the system comprising: a data acquisition unit, used to acquire and determine the movement trajectory information of the target in real time during driving; a processing unit, used to: cache the movement trajectory information of the target based on a trajectory point caching algorithm; and perform movement state determination of the corresponding target based on the cached movement trajectory information of the target; and a determination unit, used to determine the dynamic or static state of the target based on the current movement state of the target.

[0074] The present invention also provides a computer-readable storage medium storing instructions which, when executed on a computer, cause the computer to perform the above-described target state determination method.

[0075] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a microcontroller, chip, or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0076] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details described above. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention. It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not further describe the various possible combinations.

[0077] Furthermore, various different embodiments of the present invention can be combined in any way, as long as they do not violate the spirit of the embodiments of the present invention, they should also be regarded as the content disclosed by the embodiments of the present invention.

Claims

1. A target state determination method applied to a vehicle, characterized by, The target state judgment method comprises: Real-time acquisition of movement trajectory information of a judgment target; Judgment of movement trajectory information of the judgment target based on a trajectory point caching algorithm; Based on the cached movement trajectory information of the judgment target, corresponding movement state judgment of the judgment target is respectively performed; The movement state of the judgment target comprises: judgment of the speed of the judgment target, judgment of the bending state of the travel route of the judgment target, judgment of whether the judgment target is in a lane, and judgment of the movement distance of the judgment target; wherein, the judgment rule of the speed of the judgment target is: the horizontal and vertical coordinates of all historical trajectory points are differentiated to obtain the Euclidean distance between each two historical trajectory points, and the obtained Euclidean distances are accumulated to obtain a Euclidean distance sum; the counting period is differentiated, and the sum is multiplied by a preset sampling period to obtain an accumulated time; the Euclidean distance sum is taken as a total path, and the accumulated time is taken as a total time to obtain the movement speed of the judgment target; The judgment rule of the road bending state is: based on the current judgment target, the nearest lane line boundary point is determined through the global coordinates; based on the current nearest lane line boundary point, a plurality of lane line boundary points are searched to the left and right, and an outer circle is formed by selecting three lane line boundary points therefrom to obtain the center and radius of the outer circle; the road bending state of the current judgment target is judged based on the center and radius of the outer circle; The judgment rule of whether the judgment target is in the lane is: based on the current judgment target, the nearest lane line boundary point is determined through the global coordinates; the coordinates of the current judgment target and the nearest lane line boundary point form a point set, and a contour is drawn based on the point set to judge whether the coordinates of the current judgment target are located within the contour; if yes, it is determined that the current judgment target is located outside the lane; otherwise, it is determined that the current judgment target is located within the lane; The judgment rule of the movement distance is: a contour is drawn based on all historical trajectory points, and contour points are extracted; based on the extracted contour points, two-by-two comparison is performed to find out two contour points with the maximum distance, and the distance between the two contour points is taken as the movement distance; Based on the movement state of the current judgment target, it is determined whether the current target is a stationary target or a moving target.

2. The target state determination method according to claim 1, characterized by, The movement trajectory information of the judgment target comprises: Global path information, cropped local path information, left and right boundary information, and sensor acquisition information.

3. The target state determination method according to claim 2, characterized by, The cropped local path information and the left and right boundary information are both cached by taking a plurality of historical trajectory points; The sensor acquisition information comprises: Identification of the longitudinal distance, lateral distance, heading angle, target category and self-positioning of the target from the vehicle.

4. The target state determination method according to claim 1, characterized by, The trajectory point caching algorithm comprises: Defining three initial arrays, each of which has a size equal to the number of historical trajectory points collected, for storing the horizontal coordinates, vertical coordinates and indexes of each historical trajectory point respectively; Defining a field for caching each target recorded last time, denoted as a historical target field; Iterating through all targets to determine whether the identity of each target is located in a valid identity sequence; if yes, the target is valid, otherwise the target is invalid; wherein, The valid identity sequence comprises the identity of all preset valid targets. determining whether the identity of each target is same as the identity of the target recorded in the historical target field of each target recorded last time; if yes, it indicates that the target is a cached target, otherwise, it is determined that the target is a new target; caching all new targets and performing initial array assignment for each cached new target.

5. The target state determination method according to claim 4, characterized by, The caching all new targets and performing initial array assignment for each cached new target comprises: based on the coordinate information of the current new target, performing horizontal coordinate and vertical coordinate array filling of the historical trajectory point respectively; adding an index to the current new target; The method further comprises: determining whether the distance between the current new target and the last recorded target is greater than a preset distance; if greater than the preset distance, updating the initial array of all previous targets based on the current target; otherwise, only updating the initial array of the current target.

6. The target state determination method according to claim 1, characterized by, The determining, based on the movement state of the current target, whether the current target is a stationary target or a moving target comprises: if the target is located in the lane, comparing the moving distance of the target with a first preset distance and comparing the speed of the target with a first preset speed; if the moving distance is greater than the first preset distance and the speed is greater than the first preset speed, it is determined that the current target is a moving target; if the target is located in the lane, comparing the moving distance of the target with a first preset distance and comparing the speed of the target with a second preset speed; if the moving distance is less than the first preset distance and the speed is less than the second preset speed, it is determined that the current target is a stationary target; wherein, the first preset speed is greater than the second preset speed; if the moving distance is greater than the first preset distance, but the speed is greater than the first preset speed and less than the preset second preset speed, it is still determined that the current target is a moving target; if the target is located outside the lane, comparing the moving distance of the target with a second preset distance and comparing the speed of the target with a third preset speed; if the moving distance is greater than the second preset distance and the speed is greater than the third preset speed, it is determined that the current target is a moving target; otherwise, it is determined that the current target is a stationary target.

7. A target state determination system, applied to a vehicle, characterized in that, The system is applied to the target state determination method of any one of claims 1-6, and the target state determination system comprises: a collection unit configured to collect the moving trajectory information of the target in real time during the parking process; a processing unit configured to: cache the moving trajectory information of the target based on the trajectory point caching algorithm; based on the cached moving trajectory information of the target, perform the movement state determination of the corresponding target respectively; a determination unit configured to determine, based on the movement state of the current target, whether the current target is a stationary target or a moving target.

8. A computer readable storage medium, characterized in that, The computer readable storage medium stores instructions, which when executed on a computer, cause the computer to execute the target state determination method of any one of claims 1-6. The computer readable storage medium stores instructions, which when executed on a computer, cause the computer to execute the target state determination method of any one of claims 1-6.

Images