Curb tracking method, apparatus, computer-readable storage medium, and driving device

Abstract

This invention relates to the field of target tracking technology, specifically providing a roadside tracking method, apparatus, computer-readable storage medium, and driving device. The roadside tracking method of this invention includes: constructing an observation vector based on anchor points sampled from a roadside fitting curve corresponding to the roadside; matching this observation vector with a main trajectory; if the first matching state obtained after matching is a large-error matching, then matching the observation vector with a backup trajectory to obtain a second matching state; subsequently, updating the main trajectory state based on the first or second matching state and confirming the updated main trajectory state; and determining the output roadside tracking result based on the main trajectory or by replacing the main trajectory with a backup trajectory based on the confirmation result of the main trajectory state. This achieves the beneficial effect of effectively solving detection failure and measurement jump problems, and improving the stability and reliability of the final output roadside tracking result.

Description

Technical Field

[0001] This invention relates to the field of target tracking technology, specifically providing a roadside tracking method, apparatus, computer-readable storage medium, and driving device. Background Technology

[0002] Roadside detection is a crucial perception function for autonomous vehicles operating in urban areas. It provides road boundary information, constraining the vehicle's driving area. LiDAR, a vital sensor for autonomous driving, is commonly used for roadside detection. However, due to the complexity and variability of roads, the sparsity and instability of point clouds, as well as measurement noise, can affect the accuracy and robustness of single-frame roadside detection when using LiDAR. Applying tracking algorithms can improve the stability and anti-interference capabilities of roadside detection; however, current research on roadside tracking is limited, and most existing methods rely on simple Kalman filters, failing to effectively address issues such as detection failures and measurement jumps. Summary of the Invention

[0003] The present invention aims to solve the above-mentioned technical problems, namely, to solve the problem that existing roadside tracking methods based on Kalman filtering cannot effectively solve the problems of detection failure and measurement jump.

[0004] In a first aspect, the present invention provides a curb tracking method, comprising:

[0005] Obtain the curb fitting curve and curb historical trajectory corresponding to at least one side of the curb, wherein the curb historical trajectory includes a main trajectory and a backup trajectory. The curb fitting curve, the main trajectory and the backup trajectory are constructed based on a three-dimensional coordinate system with the driving device as the origin. The three-dimensional coordinate system has the driving device's forward direction as the y-axis direction, the direction perpendicular to the ground as the z-axis direction and the direction perpendicular to the plane containing the y-axis and z-axis and satisfying the right-hand rule as the x-axis direction.

[0006] Anchor point sampling is performed on the roadside fitting curve corresponding to at least one side of the roadside; an observation vector is constructed based on the multiple anchor points obtained by sampling, and the observation vector includes at least the x-axis coordinates corresponding to multiple anchor points;

[0007] The observation vector and the main trajectory are matched to obtain the first matching state between the observation vector and the main trajectory;

[0008] When the first matching state is a small error match or no match, update the main trajectory state;

[0009] When the first matching state is a large error matching, the observation vector and the backup trajectory are matched and the main trajectory state is updated based on the second matching state of the observation vector and the backup trajectory.

[0010] Update the tracking state quantity based on the update mode corresponding to the main trajectory state update;

[0011] The updated main trajectory status is confirmed based on the updated tracking status data;

[0012] Based on the confirmation result of the main trajectory state, the roadside tracking result is output.

[0013] In some embodiments, matching the observation vector and the main trajectory to obtain a first matching state between the observation vector and the main trajectory includes:

[0014] The first matching distance is obtained based on the observation vector and the x-axis coordinates of the multiple anchor points in the main trajectory;

[0015] When the first matching distance is less than or equal to the first matching distance threshold, the first matching state is determined to be a small error matching.

[0016] When the first matching distance is greater than the first matching distance threshold and less than or equal to the second matching distance threshold, the first matching state is determined to be a large error matching.

[0017] When the first matching distance is greater than the second matching distance threshold, the first matching state is determined to be an unmatched state; wherein the first matching distance threshold is less than the second matching distance threshold;

[0018] The matching of the observation vector and the backup trajectory includes:

[0019] The second matching distance is obtained based on the observation vector and the x-axis coordinates of the multiple anchor points in the backup trajectory;

[0020] When the second matching distance is less than or equal to the first matching distance threshold, the second matching state is determined to be a small error matching.

[0021] When the second matching distance is greater than the first matching distance threshold and less than or equal to the second matching distance threshold, the second matching state is determined to be a large error matching.

[0022] When the second matching distance is greater than the second matching distance threshold, the second matching state is determined to be an unmatched state.

[0023] In some embodiments, the first matching distance and the second matching distance are calculated using the following expressions:

[0024] Where D is the first matching distance or the second matching distance, and n is the number of anchor points. The x-axis coordinate corresponding to the i-th anchor point in the main trajectory or backup trajectory. Let δ be the x-axis coordinate corresponding to the i-th anchor point in the roadside fitting curve. i Let be the matching weight value for the i-th anchor point.

[0025] In some embodiments, the output of the road edge tracking result based on the confirmation result of the main trajectory state includes:

[0026] When the confirmation result of the main trajectory status is confirmed, the main trajectory updated by Kalman filtering is used as the target trajectory, and the roadside tracking result is output based on the target trajectory;

[0027] When the confirmation result of the main trajectory status is deletion, the backup trajectory is used as the target trajectory, and the roadside tracking result is output based on the target trajectory.

[0028] In some embodiments, when the first matching state is a small-error matching or no matching, updating the main trajectory state includes:

[0029] When the first matching state is a small error matching, the main trajectory state is updated in the measurement mode; when the first matching state is a non-matching, the main trajectory state is updated in the no-measurement mode.

[0030] Updating the main trajectory state based on the second matching state of the observation vector and the backup trajectory includes:

[0031] When the second matching state is small error matching, large error matching, or no matching, the main trajectory state is updated in measurement-free mode.

[0032] In some embodiments, the tracking status quantity includes the count value of the unassociated count counter and the count value of the associated count counter, and the step of updating the tracking status quantity based on the update mode corresponding to the main trajectory status update includes:

[0033] When the main trajectory status is updated, and the corresponding update mode is the measured mode, the count value of the associated count counter is incremented by one, and the count value of the unassociated count counter is cleared to zero.

[0034] When the update mode corresponding to the main trajectory status update is the measurement-free mode, the count value of the unassociated count counter is incremented by one.

[0035] In some embodiments, confirming the updated main trajectory state based on the updated tracking state quantity includes:

[0036] When the update mode corresponding to the main trajectory status update is the measured mode, determine whether the count value of the association count counter is greater than or equal to the association count threshold; if so, the confirmation result of the main trajectory status is confirmed.

[0037] When the update mode corresponding to the main trajectory status update is the measurement-free mode, it is determined whether the count value of the unassociated count counter is less than or equal to the unassociated count threshold; if not, the confirmation result of the main trajectory status is deletion, and the count value of the associated count counter is cleared to zero.

[0038] In a second aspect, the present invention provides a curb tracking device, comprising:

[0039] The curve acquisition module is used to acquire the curb fitting curve and the historical curb trajectory corresponding to at least one side of the curb. The historical curb trajectory includes a main trajectory and a backup trajectory. The curb fitting curve, the main trajectory, and the backup trajectory are constructed based on a three-dimensional coordinate system with the driving device as the origin. The three-dimensional coordinate system has the driving device's forward direction as the y-axis, the direction perpendicular to the ground as the z-axis, and the direction perpendicular to the plane containing the y-axis and z-axis and satisfying the right-hand rule as the x-axis.

[0040] An anchor point sampling module is used to sample the curb fitting curve of the at least one side of the road edge; and to construct an observation vector based on the multiple anchor points obtained by sampling, wherein the observation vector includes at least the x-axis coordinates corresponding to the multiple anchor points.

[0041] A distance matching module is used to match the observation vector with the main trajectory to obtain a first matching state between the observation vector and the main trajectory;

[0042] The trajectory status update module is used to update the main trajectory status when the first matching status is a small error match or no match;

[0043] When the first matching state is a large error matching, the observation vector and the backup trajectory are matched and the main trajectory state is updated based on the second matching state of the observation vector and the backup trajectory.

[0044] The tracking status update module is used to update the tracking status based on the update mode corresponding to the main trajectory status update.

[0045] The trajectory status confirmation module is used to confirm the updated main trajectory status based on the updated tracking status quantity.

[0046] The tracking output module is used to output the roadside tracking result based on the confirmation result of the main trajectory state.

[0047] In a third aspect, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the curb tracking method described in any one of the preceding claims.

[0048] In a fourth aspect, the present invention provides a driving device including a memory and a processor, wherein the memory stores a computer program that, when executed by the processor, implements the curb tracking method described in any of the preceding claims.

[0049] By employing the above technical solution, this invention constructs an observation vector based on anchor points sampled from the roadside fitting curve corresponding to the roadside, and matches this observation vector with the historical roadside trajectory. Roadside tracking is performed according to different matching states corresponding to the matching results. The historical roadside trajectory includes a main trajectory and a backup trajectory. During matching, the observation vector and the main trajectory are matched first. When the first matching state obtained after matching is a large error matching, the observation vector and the backup trajectory are matched to obtain a second matching state. The main trajectory state can then be updated based on the first or second matching state, and the tracking state quantity is updated based on the update mode corresponding to the update of the main trajectory state. The updated main trajectory state is confirmed based on the tracking state quantity. Based on the confirmation result of the main trajectory state, the output roadside tracking result is determined based on the main trajectory or by replacing the main trajectory with the backup trajectory. This achieves the beneficial effect of effectively solving the problems of detection failure and measurement jump, and improving the stability and reliability of the final output roadside tracking result. Attached Figure Description

[0050] The preferred embodiments of the present invention are described below with reference to the accompanying drawings, in which:

[0051] Figure 1 This is a schematic flowchart of a roadside tracking method provided in an embodiment of the present invention;

[0052] Figure 2 This is a flowchart illustrating the method for tracking left and right road edges provided in an embodiment of the present invention.

[0053] Figure 3 This is a schematic diagram of a roadside tracking method provided in a specific example of the present invention;

[0054] Figure 4 This is a schematic diagram of the structure of a curb tracking device provided in an embodiment of the present invention;

[0055] Figure 5 This is a schematic diagram of the driving device provided in an embodiment of the present invention. Detailed Implementation

[0056] Some embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0057] Roadside detection is a crucial perception function for autonomous vehicles operating in urban areas. It provides road boundary information, constraining the vehicle's driving area. LiDAR, a vital sensor for autonomous driving, is commonly used for roadside detection. However, due to the complexity and variability of roads, the sparsity and instability of point clouds, as well as measurement noise, can affect the accuracy and robustness of single-frame roadside detection when using LiDAR. Applying tracking algorithms can improve the stability and anti-interference capabilities of roadside detection; however, current research on roadside tracking is limited, and most existing methods rely on simple Kalman filters, failing to effectively address issues such as detection failures and measurement jumps.

[0058] In view of this, the present invention provides a roadside tracking method, which constructs an observation vector based on anchor points sampled from the roadside fitting curve corresponding to the roadside, and matches the observation vector with the historical roadside trajectory. Roadside tracking is performed according to different matching states corresponding to the matching results. The historical roadside trajectory includes a main trajectory and a backup trajectory. During matching, the observation vector and the main trajectory are matched first. When the first matching state obtained after matching is a large error matching, the observation vector and the backup trajectory are matched to obtain a second matching state. The main trajectory state can then be updated based on the first or second matching state, and the tracking state quantity is updated based on the update mode corresponding to the update of the main trajectory state. The updated main trajectory state is confirmed based on the tracking state quantity. Based on the confirmation result of the main trajectory state, the roadside tracking result is output based on the main trajectory or by replacing the main trajectory with the backup trajectory. This achieves the beneficial effect of effectively solving the detection failure and measurement jump problem, and improving the stability and reliability of the final output roadside tracking result.

[0059] See Figure 1 As shown, Figure 1 This is a flowchart illustrating a curb tracking method provided in an embodiment of the present invention, which may include:

[0060] Step S11: Obtain the curb fitting curve and curb historical trajectory corresponding to at least one side of the curb, wherein the curb historical trajectory includes the main trajectory and the backup trajectory. The curb fitting curve, the main trajectory and the backup trajectory are constructed based on a three-dimensional coordinate system with the driving device as the origin. The three-dimensional coordinate system has the driving device's forward direction as the y-axis direction, the direction perpendicular to the ground as the z-axis direction and the direction perpendicular to the plane containing the y-axis and z-axis and satisfying the right-hand rule as the x-axis direction.

[0061] Step S12: Anchor point sampling is performed on the curb fitting curve corresponding to at least one side of the curb; an observation vector is constructed based on the multiple anchor points obtained by sampling, and the observation vector includes at least the x-axis coordinates corresponding to multiple anchor points;

[0062] Step S13: Match the observation vector with the main trajectory to obtain the first matching state between the observation vector and the main trajectory;

[0063] Step S14: When the first matching state is a small error match or no match, update the main trajectory state;

[0064] Step S15: When the first matching state is a large error matching, match the observation vector and the backup trajectory and update the main trajectory state based on the second matching state of the observation vector and the backup trajectory;

[0065] Step S16: Update the tracking state quantity based on the update mode corresponding to the main trajectory state update;

[0066] Step S17: Confirm the updated main trajectory status based on the updated tracking status data;

[0067] Step S18: Output the curb tracking result based on the confirmation result of the main trajectory status.

[0068] In some embodiments, step S11 may specifically involve using a three-dimensional lidar to detect the curb and obtain a curb fitting curve for at least one side of the curb. The historical curb trajectory may be a sequence of positions of at least one side of the curb within a preset time period.

[0069] In the initial time k=0 for roadside tracking based on the roadside fitting curve and the historical roadside trajectory, the target position in the main trajectory and backup trajectory can also be initialized based on the measurement value corresponding to the roadside fitting curve at the initial time. The initialization process of the main trajectory and backup trajectory is different.

[0070] In some embodiments, step S12 may specifically involve determining the number of anchor points for each side of the road edge based on the road edge fitting curve. When the road edge fitting curve is an Nth order curve, N+1 anchor points are selected on the road edge fitting curve, and an observation vector is constructed based on the N+1 anchor points.

[0071] In some embodiments, constructing an observation vector based on N+1 anchor points can specifically involve selecting N+1 different y-coordinates on the roadside fitting curve, determining the corresponding x-coordinate for each y-coordinate according to the roadside fitting curve, and constructing an observation vector based on the x-axis coordinates corresponding to the N+1 anchor points. To improve the smoothness of the roadside tracking output, N+1 uniformly distributed different y-coordinates can be selected, and the x-coordinates and observation vectors can be determined and constructed based on these coordinates.

[0072] As a specific example, when the fitted curve of the roadside on one side is a quadratic curve, anchor points can be sampled on the fitted curve of the roadside on that side to obtain 3 anchor points. Based on the x-axis coordinates corresponding to the 3 anchor points, an observation vector z(k) = (χ0, χ1, χ2) can be constructed. T , where z(k) represents the observation vector at time k.

[0073] In some embodiments, step S13 may specifically be:

[0074] The first matching distance is obtained based on the observation vector and the x-axis coordinates of multiple anchor points in the main trajectory;

[0075] When the first matching distance is less than or equal to the first matching distance threshold, the first matching state is determined to be a small error matching.

[0076] When the first matching distance is greater than the first matching distance threshold and less than or equal to the second matching distance threshold, the first matching state is determined to be a large error matching.

[0077] When the first matching distance is greater than the second matching distance threshold, the first matching state is determined to be an unmatched state; wherein the first matching distance threshold is less than the second matching distance threshold.

[0078] In some embodiments, the first matching distance is calculated using the following expression 1-1:

[0079]

[0080] Where D is the first matching distance, and n is the number of anchor points. The x-axis coordinates of the main trajectory and the i-th anchor point. Let δ be the x-axis coordinate corresponding to the i-th anchor point in the roadside fitting curve. i Let be the matching weight value for the i-th anchor point.

[0081] In some embodiments, step S14 may specifically be:

[0082] When the first matching state is a small error matching, the main trajectory state is updated in the measurement mode; when the first matching state is an unmatched state, the main trajectory state is updated in the measurement-free mode and the backup trajectory is initialized using the measurement values.

[0083] In some embodiments, to improve the smoothness of the curb trajectory, Kalman filtering can be used to track and filter the main trajectory for different first matching states. When updating the main trajectory state in a measured mode based on Kalman filtering, the predicted state vector at the current time is predicted using the Kalman filter estimation vector x(k-1) obtained at the previous time step. And based on the observation vector z(k) at the current time and the predicted state vector at the current time. The estimation vector x(k) at the current time is obtained by fusion and output. When updating the main trajectory state in the measurement-free mode based on Kalman filtering, the predicted state vector at the current time is predicted using the Kalman filter estimation vector x(k-1) obtained at the previous time. No fusion process will be performed, based on the predicted... As output.

[0084] Among them, the roadside fitting curve is used as a quadratic curve, and three anchor points are collected. The observation vector constructed based on the three anchor points is z(k)=(χ0,χ1,χ2). T For example, using Kalman filtering for tracking, the state vector of the main trajectory can be represented as: v1, v2, and v3 represent the velocities of the three anchor points along the x-axis. Assuming the measurement noise is Gaussian white noise, the measurement model can be expressed as expression 2-1:

[0085] z(k)=x(k)+v(k))(2-1), where v(k) represents the measurement noise;

[0086] Predict the state using expression 2-2 and update the error covariance matrix using expression 2-3:

[0087]

[0088]

[0089] in, χ(k) represents the predicted state vector for the next time step k+1, and χ(k) is the state vector for the current time step k. Let P(k) be the prediction error covariance at the next time step, P(k) be the error covariance at the current time step, ω(k) be the process noise, A(k) be the state transition matrix, B(k) be the control input matrix, u(k) be the system control quantity, and C be the prediction error covariance at the next time step. q Representative process excitation noise covariance;

[0090] Calculate the Kalman filter gain using expression 2-4:

[0091]

[0092] Among them, G K (k+1) represents the Kalman filter gain updated at time k+1, C r Represents the measurement noise covariance;

[0093] Update the state using expression 2-5 and update the error covariance matrix using expression 2-6:

[0094]

[0095]

[0096] Where x(k+1) is the estimated vector for the next time step k+1, which is the anchor point x-coordinate vector of the final output of the Kalman filter tracking; P(k+1) is the error covariance updated for the next time step k+1.

[0097] In some embodiments, matching the observation vector and the backup trajectory in step S15 can specifically be as follows:

[0098] The second matching distance is obtained based on the observation vector and the x-axis coordinates of multiple anchor points in the backup trajectory;

[0099] When the second matching distance is less than or equal to the first matching distance threshold, the second matching state is determined to be a small error matching.

[0100] When the second matching distance is greater than the first matching distance threshold and less than or equal to the second matching distance threshold, the second matching state is determined to be a large error matching.

[0101] When the second matching distance is greater than the second matching distance threshold, the second matching state is determined to be an unmatched state.

[0102] The second matching distance can be calculated using expression 1-1, where D is replaced with the second matching distance. Replace it with the x-axis coordinate corresponding to the i-th anchor point in the backup trajectory.

[0103] In some embodiments, updating the main trajectory state based on the second matching state of the observation vector and the backup trajectory in step S15 includes:

[0104] When the second matching state is small error matching, large error matching, or no matching, the main trajectory state is updated in measurement-free mode.

[0105] In some embodiments, when the second matching state is a large error match or no match, the backup trajectory is initialized with the measurement value.

[0106] In some embodiments, the tracking status quantity may include the count value of the unassociated count counter and the count value of the association coefficient counter, and step S16 may specifically be as follows:

[0107] When the main trajectory status is updated, if the corresponding update mode is the measurement mode, the count value of the associated count counter is incremented by one, and the count value of the unassociated count counter is cleared to zero.

[0108] When the main trajectory status is updated, and the corresponding update mode is the no-measurement mode, the count value of the unassociated count counter is incremented by one.

[0109] In some embodiments, step S17 may specifically be:

[0110] When the main trajectory status is updated and the corresponding update mode is the measurement mode, determine whether the count value of the association count counter is greater than or equal to the association count threshold; if yes, the confirmation result of the main trajectory status is confirmed; if no, the confirmation result of the main trajectory status is not confirmed.

[0111] When the update mode corresponding to the main trajectory state is the measurement-free mode, it is determined whether the count value of the unassociated count counter is less than or equal to the unassociated count threshold. If not, the confirmation result of the main trajectory state is deletion, and the count value of the associated count counter is cleared to zero. If yes, the confirmation result of the main trajectory state is to continue the main trajectory state of the previous time step, that is, to use the Kalman filter estimation vector of the previous time step as the output.

[0112] In some embodiments, step S18 may specifically be:

[0113] When the confirmation result of the main trajectory status is confirmed, the main trajectory updated by Kalman filtering will be used as the target trajectory, and the roadside tracking result will be output based on the target trajectory.

[0114] When the confirmation result of the main trajectory status is deletion, the backup trajectory is used as the target trajectory, and the road edge tracking result is output based on the target trajectory. In some other embodiments, when the second matching state is small error matching, the backup trajectory can be updated by Kalman filtering, so that when the confirmation result of the main trajectory status is deletion, the backup trajectory updated by Kalman filtering can be used as the target trajectory.

[0115] The above describes a roadside tracking method provided by an embodiment of the present invention. It constructs an observation vector based on anchor points sampled from the roadside fitting curve corresponding to the roadside, and matches this observation vector with the historical roadside trajectory. Roadside tracking is performed according to different matching states corresponding to the matching results. The historical roadside trajectory includes a main trajectory and a backup trajectory. During matching, the observation vector and the main trajectory are first matched. When the first matching state obtained after matching is a large-error matching, the observation vector and the backup trajectory are matched again to obtain a second matching state. The main trajectory state can then be updated based on either the first or second matching state, and the tracking state quantity is updated based on the update mode corresponding to the update of the main trajectory state. The updated main trajectory state is confirmed based on the tracking state quantity. Based on the confirmation result of the main trajectory state, the output roadside tracking result is determined based on the main trajectory or by replacing the main trajectory with a backup trajectory. This effectively solves the problems of detection failure and measurement jumps, and improves the stability and reliability of the final output roadside tracking result.

[0116] In other embodiments, the curb fitting curves on both sides of the curb can be acquired simultaneously, and curb tracking can be performed separately for each curb fitting curve, as described below.

[0117] See Figure 2 As shown, Figure 2 This is a flowchart illustrating a method for tracking left and right road edges according to an embodiment of the present invention, which may include:

[0118] Step S21: Obtain the detection results of the left and right curb edges;

[0119] Perform steps S22-S24 for the left and right edges respectively:

[0120] Step S22: Match the curb measurement values ​​with the historical trajectory to obtain the matching status between the two;

[0121] Step S23: Update the trajectory according to different matching states;

[0122] Step S24: Confirm trajectory status;

[0123] Step S25: Output the left and right edge tracking results.

[0124] In this embodiment of the invention, step S21 can specifically be to obtain the roadside fitting curve and roadside historical trajectory corresponding to the left and right roadside respectively. The roadside historical trajectory includes the main trajectory and the backup trajectory. The acquisition of the roadside fitting curve and roadside historical trajectory for each roadside can be implemented in the same way as step S11 in the above embodiment. For details, please refer to the description above.

[0125] In some embodiments, anchor point sampling can be performed on the fitted curve of the left or right curb, and an observation vector can be constructed based on the x-axis coordinates corresponding to multiple anchor points, where the x-axis coordinate corresponding to each anchor point is the measured value of the curb. The historical trajectory refers to the historical curb trajectory, including the main trajectory and backup trajectories. Step S22 can specifically involve first matching the observation vector with the main trajectory to obtain a first matching state; when the first matching state is a large-error match, the observation vector is then matched with the backup trajectory to obtain a second matching state. The first matching state can be obtained in the same way as in step S13, and the second matching state can be obtained in the same way as in step S15.

[0126] In some embodiments, step S23 may specifically involve updating the main trajectory state when the first matching state is a small-error matching or no matching; updating the main trajectory state based on the second matching state when the first matching state is a large-error matching; and updating the tracking state quantity based on the update mode corresponding to the main trajectory state update. The updating of the main trajectory state and the tracking state quantity can be performed in the same manner as steps S14-S16, as described above.

[0127] In some embodiments, step S24 may specifically involve confirming the updated main trajectory status based on the updated tracking status quantity, as detailed in the implementation of step S17.

[0128] In some embodiments, step S25 may specifically be to output road edge tracking results based on the confirmation result of the main trajectory state of the corresponding side road edge for the left or right road edge. When the confirmation result of the main trajectory state is confirmed, the main trajectory updated by Kalman filtering is used as the target trajectory, and the road edge tracking result is output based on the target trajectory. When the confirmation result of the main trajectory state is deleted, the backup trajectory is used as the target trajectory, and the road edge tracking result is output based on the target trajectory, thereby obtaining the left and right road edge tracking results.

[0129] See Figure 3 As shown, Figure 3 This is a schematic diagram of a roadside tracking method provided by a specific example of the present invention.

[0130] This curb tracking method can be implemented based on the curb tracking method shown in the above embodiments of the present invention. It matches the historical curb trajectory with the curb fitting curve obtained from curb measurements. The historical curb trajectory includes a main trajectory and a backup trajectory. First, the main trajectory is matched. When the matching status of the main trajectory is a small-error match, the main trajectory filter is updated in a measurement-enabled mode. When the matching status of the main trajectory is unmatched, the main trajectory filter is updated and the backup trajectory is initialized in a measurement-free mode. When the matching status of the main trajectory is a large-error match, the backup trajectory is matched. When the matching status of the backup trajectory is a small-error match, the backup trajectory filter is updated. When the matching status of the backup trajectory is a large-error match or unmatched, the backup trajectory is initialized, and simultaneously, the main trajectory filter is updated in a measurement-free mode according to the corresponding matching status of the backup trajectory. Then, the tracking status is updated based on the update mode corresponding to the main trajectory filter update, and the trajectory is confirmed based on the updated tracking status. When the confirmation result of the main trajectory is confirmed, the main trajectory after filter update is used as the target trajectory, and the road edge tracking result is output based on the target trajectory; when the confirmation result of the main trajectory is deleted, the backup trajectory after filter update is used as the target trajectory, and the road edge tracking result is output based on the target trajectory.

[0131] See Figure 4 As shown, Figure 4 This is a schematic diagram of a curb tracking device provided in an embodiment of the present invention, which includes:

[0132] The curve acquisition module 41 is used to acquire the curb fitting curve and the historical curb trajectory corresponding to at least one side of the curb. The historical curb trajectory includes the main trajectory and the backup trajectory. The curb fitting curve, the main trajectory and the backup trajectory are constructed based on a three-dimensional coordinate system with the driving device as the origin. The three-dimensional coordinate system has the driving device's forward direction as the y-axis direction, the direction perpendicular to the ground as the z-axis direction and the direction perpendicular to the plane containing the y-axis and z-axis and satisfying the right-hand rule as the x-axis direction.

[0133] Anchor point sampling module 42 is used to sample anchor points on the roadside fitting curve corresponding to at least one side of the roadside; and to construct an observation vector based on the multiple anchor points obtained by sampling, wherein the observation vector includes at least the x-axis coordinates corresponding to the multiple anchor points.

[0134] Distance matching module 43 is used to match the observation vector and the main trajectory to obtain the first matching state between the observation vector and the main trajectory;

[0135] The trajectory status update module 44 is used to update the main trajectory status when the first matching status is a small error matching or no matching;

[0136] When the first matching state is a large error matching, the observation vector and the backup trajectory are matched and the main trajectory state is updated based on the second matching state of the observation vector and the backup trajectory.

[0137] The tracking status update module 45 is used to update the tracking status based on the update mode corresponding to the main trajectory status update.

[0138] The trajectory status confirmation module 46 is used to confirm the updated main trajectory status based on the updated tracking status quantity.

[0139] Tracking output module 47 is used to output the curb tracking result based on the confirmation result of the main trajectory state.

[0140] The curb tracking device provided by this invention can be used to execute the aforementioned curb tracking method, achieving the same beneficial effects as the curb tracking method in the above embodiments. Furthermore, it should be understood that since the various modules are only provided to illustrate the functional units of the device of this invention, the physical devices corresponding to these modules can be the processor itself, or a part of the processor's software, hardware, or a combination of both. Therefore, Figure 4The number of modules shown is merely illustrative. Those skilled in the art will understand that the modules in the device can be adaptively split or combined. Such splitting or combining of specific modules will not cause the technical solution to deviate from the principles of the present invention; therefore, the technical solutions after splitting or combining will fall within the protection scope of the present invention.

[0141] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable storage medium can include any entity or device capable of carrying the computer program code, a medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory, a random access memory, an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0142] Another aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, enables the implementation of the edge-tracking method in any of the above embodiments. The computer-readable storage medium may be a storage device comprising various electronic devices; optionally, in embodiments of the present invention, the computer-readable storage medium is a non-transitory computer-readable storage medium.

[0143] In another aspect, the present invention also provides a driving device, see [link to relevant documentation]. Figure 5 As shown, Figure 5 This is a schematic diagram of the structure of the driving device provided in an embodiment of the present invention, which may include:

[0144] The system includes a memory 51 and a processor 52. The memory 51 stores a computer program, which, when executed by the processor 52, implements the curb tracking method described in the above embodiments.

[0145] The driving device can implement the curb tracking method in the above embodiments, achieving the same beneficial effects as the above embodiments.

[0146] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A curb tracking method, characterized in that, The method includes: Obtain the curb fitting curve and curb historical trajectory corresponding to at least one side of the curb, wherein the curb historical trajectory includes a main trajectory and a backup trajectory. The curb fitting curve, the main trajectory and the backup trajectory are constructed based on a three-dimensional coordinate system with the driving device as the origin. The three-dimensional coordinate system has the driving device's forward direction as the y-axis direction, the direction perpendicular to the ground as the z-axis direction and the direction perpendicular to the plane containing the y-axis and z-axis and satisfying the right-hand rule as the x-axis direction. Anchor point sampling is performed on the roadside fitting curve corresponding to at least one side of the roadside; an observation vector is constructed based on the multiple anchor points obtained by sampling, and the observation vector includes at least the x-axis coordinates corresponding to multiple anchor points; The observation vector and the main trajectory are matched to obtain the first matching state between the observation vector and the main trajectory; When the first matching state is a small error match or no match, update the main trajectory state; When the first matching state is a large error matching, the observation vector and the backup trajectory are matched and the main trajectory state is updated based on the second matching state of the observation vector and the backup trajectory. Update the tracking state quantity based on the update mode corresponding to the main trajectory state update; The updated main trajectory status is confirmed based on the updated tracking status data; Based on the confirmation result of the main trajectory state, the roadside tracking result is output.

2. The method according to claim 1, characterized in that, Matching the observation vector with the main trajectory to obtain a first matching state between the observation vector and the main trajectory includes: The first matching distance is obtained based on the observation vector and the x-axis coordinates of the multiple anchor points in the main trajectory; When the first matching distance is less than or equal to the first matching distance threshold, the first matching state is determined to be a small error matching. When the first matching distance is greater than the first matching distance threshold and less than or equal to the second matching distance threshold, the first matching state is determined to be a large error matching. When the first matching distance is greater than the second matching distance threshold, the first matching state is determined to be an unmatched state; wherein the first matching distance threshold is less than the second matching distance threshold; The matching of the observation vector and the backup trajectory includes: The second matching distance is obtained based on the observation vector and the x-axis coordinates of the multiple anchor points in the backup trajectory; When the second matching distance is less than or equal to the first matching distance threshold, the second matching state is determined to be a small error matching. When the second matching distance is greater than the first matching distance threshold and less than or equal to the second matching distance threshold, the second matching state is determined to be a large error matching. When the second matching distance is greater than the second matching distance threshold, the second matching state is determined to be an unmatched state.

3. The method according to claim 2, characterized in that, The first matching distance and the second matching distance are calculated using the following expressions: Where D is the first matching distance or the second matching distance, and n is the number of anchor points. The x-axis coordinate corresponding to the i-th anchor point in the main trajectory or backup trajectory. Let δ be the x-axis coordinate corresponding to the i-th anchor point in the roadside fitting curve. i Let be the matching weight value for the i-th anchor point.

4. The method according to claim 1, characterized in that, The output of the roadside tracking results based on the confirmation result of the main trajectory state includes: When the confirmation result of the main trajectory status is confirmed, the main trajectory updated by Kalman filtering is used as the target trajectory, and the roadside tracking result is output based on the target trajectory; When the confirmation result of the main trajectory status is deletion, the backup trajectory is used as the target trajectory, and the roadside tracking result is output based on the target trajectory.

5. The method according to claim 1, characterized in that, When the first matching state is a small-error match or no match, update the main trajectory state, including: When the first matching state is a small error matching, the main trajectory state is updated in the measurement mode; when the first matching state is a non-matching, the main trajectory state is updated in the no-measurement mode. Updating the main trajectory state based on the second matching state of the observation vector and the backup trajectory includes: When the second matching state is small error matching, large error matching, or no matching, the main trajectory state is updated in measurement-free mode.

6. The method according to claim 5, characterized in that, The tracking status quantity includes the count value of the unassociated count counter and the count value of the associated count counter. The step of updating the tracking status quantity based on the update mode corresponding to the main trajectory status update includes: When the main trajectory status is updated, and the corresponding update mode is the measured mode, the count value of the associated count counter is incremented by one, and the count value of the unassociated count counter is cleared to zero. When the update mode corresponding to the main trajectory status update is the measurement-free mode, the count value of the unassociated count counter is incremented by one.

7. The method according to claim 6, characterized in that, The step of confirming the updated main trajectory status based on the updated tracking status data includes: When the update mode corresponding to the main trajectory status update is the measured mode, determine whether the count value of the association count counter is greater than or equal to the association count threshold; if so, the confirmation result of the main trajectory status is confirmed. When the update mode corresponding to the main trajectory status update is the measurement-free mode, it is determined whether the count value of the unassociated count counter is less than or equal to the unassociated count threshold; if not, the confirmation result of the main trajectory status is deletion, and the count value of the associated count counter is cleared to zero.

8. A curb tracking device, characterized in that, include: The curve acquisition module is used to acquire the curb fitting curve and the historical curb trajectory corresponding to at least one side of the curb. The historical curb trajectory includes a main trajectory and a backup trajectory. The curb fitting curve, the main trajectory, and the backup trajectory are constructed based on a three-dimensional coordinate system with the driving device as the origin. The three-dimensional coordinate system has the driving device's forward direction as the y-axis, the direction perpendicular to the ground as the z-axis, and the direction perpendicular to the plane containing the y-axis and z-axis and satisfying the right-hand rule as the x-axis. An anchor point sampling module is used to sample the curb fitting curve of the at least one side of the road edge; and to construct an observation vector based on the multiple anchor points obtained by sampling, wherein the observation vector includes at least the x-axis coordinates corresponding to the multiple anchor points. A distance matching module is used to match the observation vector with the main trajectory to obtain a first matching state between the observation vector and the main trajectory; The trajectory status update module is used to update the main trajectory status when the first matching status is a small error match or no match; When the first matching state is a large error matching, the observation vector and the backup trajectory are matched and the main trajectory state is updated based on the second matching state of the observation vector and the backup trajectory. The tracking status update module is used to update the tracking status based on the update mode corresponding to the main trajectory status update. The trajectory status confirmation module is used to confirm the updated main trajectory status based on the updated tracking status quantity. The tracking output module is used to output the roadside tracking result based on the confirmation result of the main trajectory state.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the curb tracking method according to any one of claims 1 to 7.

10. A driving device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that, when executed by the processor, implements the curb tracking method according to any one of claims 1 to 7.

Images