Obstacle Detection Method and Electronic Device

Abstract

The present application provides an obstacle detection method and an electronic device. The obstacle detection method includes: obtaining consecutive multiple frames of a target obstacle, where the target obstacle is an obstacle for which it is necessary to track whether it truly exists or disappears; determining whether the target obstacles in every two adjacent frames among the consecutive multiple frames meet the obstacle existence and disappearance condition; the obstacle existence and disappearance condition includes an obstacle true existence condition; if the target obstacles in every two adjacent frames meet the obstacle true existence condition, it is determined that the target obstacle in the current frame truly exists. In this way, the accuracy of determining the target obstacle is relatively high.

Description

Technical Field

[0001] This application relates to the field of vehicle communication, and in particular, to an obstacle detection method and an electronic device. Background Art

[0002] Currently, millimeter-wave radar detection is widely used in the field of autonomous driving. The millimeter-wave radar emits directional millimeter waves. When the millimeter waves encounter an obstacle target and are reflected back, the two-dimensional position of the obstacle can be detected by receiving the reflected millimeter waves. However, since the millimeter-wave radar is prone to detecting many false positive points, using the original detection of the millimeter-wave radar will result in low accuracy in determining obstacles. Summary of the Invention

[0003] This application provides an obstacle detection method and an electronic device.

[0004] This application provides an obstacle detection method, including:

[0005] Obtaining consecutive multiple frames of target obstacles, where the target obstacles are obstacles that need to be tracked for whether they really exist or disappear;

[0006] Determining whether the target obstacles in every two adjacent frames among the consecutive multiple frames reach the obstacle existence and disappearance condition; the obstacle existence and disappearance condition includes the obstacle real existence condition;

[0007] If the target obstacles in every two adjacent frames reach the obstacle real existence condition, it is determined that the target obstacles in the current frame really exist.

[0008] Further, the obtaining consecutive multiple frames of target obstacles includes:

[0009] Obtaining consecutive multiple frames of obstacles; the consecutive multiple frames of obstacles include the obstacles in the current frame and the obstacles in the previous frame adjacent to the current frame;

[0010] Comparing the obstacles in the current frame with the obstacles in the previous frame, and determining the first obstacle to be tracked and the second obstacle to be tracked in the current frame, where the first obstacle to be tracked and the second obstacle to be tracked are respectively included in the target obstacles, the first obstacle to be tracked exists in the previous frame and does not exist in the current frame, and the second obstacle to be tracked does not exist in the previous frame and exists in the current frame.

[0011] Further, the obstacle real existence condition includes whether the total number of matches of the target obstacles in every two adjacent frames among the consecutive multiple frames reaches a preset existence number;

[0012] The step of if the target obstacles in every two adjacent frames reach the obstacle real existence condition, it is determined that the target obstacles in the current frame really exist, includes:

[0013] If the total number of matches of the second obstacle to be tracked in every two adjacent frames reaches a preset existence count, it is determined that the second obstacle to be tracked truly exists.

[0014] Further, after comparing the obstacles in the current frame with the obstacles in the previous frame to determine the first obstacle to be tracked and the second obstacle to be tracked in the current frame, the method further includes:

[0015] Record the second obstacle to be tracked in the list of obstacles to be tracked.

[0016] The step of determining that the second obstacle to be tracked truly exists if the total number of matches of the second obstacle to be tracked in every two adjacent frames reaches a preset existence count includes:

[0017] Obtain the next frame adjacent to the current frame as the current frame.

[0018] Match the obstacles in the current frame with the obstacles in the list of obstacles to be tracked to determine whether there is a second obstacle to be tracked in the current frame. If there is a second obstacle to be tracked in the current frame, determine whether the total number of matches of the second obstacle to be tracked in every two adjacent frames reaches a preset existence count. If the total number of matches of the second obstacle to be tracked in every two adjacent frames does not reach the preset existence count, obtain the next frame adjacent to the current frame as the current frame, and continue to return to the step of matching the current frame with the list of obstacles to be tracked to determine whether there is a second obstacle to be tracked in the current frame until there is a second obstacle to be tracked in the current frame and the total number of matches of the second obstacle to be tracked in every two adjacent frames reaches the preset existence count, then determine that the second obstacle to be tracked truly exists.

[0019] Further, the obstacle existence and disappearance condition includes the obstacle true disappearance condition. After determining whether the target obstacle in every two adjacent frames in a series of consecutive frames reaches the obstacle existence and disappearance condition, the method further includes:

[0020] If the target obstacle in every two adjacent frames reaches the obstacle disappearance condition, it is determined that the obstacle in the current frame truly disappears.

[0021] Further, the obstacle true disappearance condition includes whether the number of disappearances of the first obstacle to be tracked in every two adjacent frames in a series of consecutive frames reaches a preset number of disappearances.

[0022] The step of determining that the obstacle in the current frame truly disappears if the target obstacle in every two adjacent frames reaches the obstacle disappearance condition includes:

[0023] If the number of disappearances of the first obstacle to be tracked in every two adjacent frames reaches a preset number of disappearances, it is determined that the first obstacle to be tracked has truly disappeared.

[0024] Further, after comparing the obstacles in the current frame with the obstacles in the previous frame to determine the first obstacle to be tracked and the second obstacle to be tracked in the current frame, the method further includes:

[0025] Record the first obstacle to be tracked in the tracking obstacle list;

[0026] The step of determining that the first obstacle to be tracked has truly disappeared if the number of disappearances of the first obstacle to be tracked in every two adjacent frames reaches a preset number of disappearances includes:

[0027] Obtain the next frame adjacent to the current frame as the current frame;

[0028] Match the obstacles in the current frame with the obstacles in the tracking obstacle list to determine whether there is a first obstacle to be tracked in the current frame. If there is no first obstacle to be tracked in the current frame and the number of disappearances of the first obstacle to be tracked in every two adjacent frames reaches the preset number of disappearances, it is determined that the first obstacle to be tracked has truly disappeared.

[0029] Further, the step of comparing the obstacles in the current frame with the obstacles in the previous frame to determine the first obstacle to be tracked and the second obstacle to be tracked in the current frame includes:

[0030] Obtain the first point cloud data of the current frame and the second point cloud data of the previous frame; the first point cloud data and the second point cloud data are respectively obtained by detecting the environment around the vehicle through a millimeter-wave radar.

[0031] Cluster a plurality of first detected points in the first point cloud data to determine a new obstacle list of the current frame. The new obstacle list includes at least one new obstacle, and each new obstacle corresponds to each category obtained by clustering.

[0032] Cluster a plurality of second detected points in the second point cloud data to determine a historical obstacle list of the previous frame. The historical obstacle list includes at least one historical obstacle, and each historical obstacle corresponds to each category obtained by clustering.

[0033] Compare the obstacles in the new obstacle list of the current frame with the obstacles in the historical obstacle list of the previous frame to determine the first obstacle to be tracked in the current frame and the second obstacle to be tracked in the previous frame.

[0034] Further, after the target obstacle in the current frame actually exists, the method further includes: deleting the second obstacle to be tracked in the list of tracked obstacles.

[0035] This application provides an electronic device, including a processor and a memory;

[0036] The memory is used for storing a computer program;

[0037] The processor is configured to implement the method according to any one of the above-mentioned first aspects and the method according to any one of the above-mentioned second aspects when executing the program stored in the memory.

[0038] This application provides a readable storage medium, on which a program is stored, and when the program is executed by a processor, the method as described is implemented.

[0039] In some embodiments of this application, the target obstacles in every two adjacent frames among consecutive multiple frames meet the condition for the actual existence of the obstacle, and it is determined that the obstacle actually exists. The consecutive multiple frames used in the embodiments of this application avoid misjudgment caused by many false positive points detected by the millimeter-wave radar only appearing in one frame, and the accuracy of determining the target obstacle is relatively high. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 The figure shows a system structure diagram of an embodiment of an obstacle detection system provided by this application;

[0041] Figure 2 The figure shows a flowchart of an embodiment of an obstacle determination method provided by this application;

[0042] Figure 3 As shown in Figure 2 The figure shows a flowchart of step 110 in the obstacle determination method shown;

[0043] Figure 4 As shown in Figure 2 The figure shows a flowchart of step 112 in the obstacle determination method shown;

[0044] Figure 5 As shown in Figure 2 The figure shows a flowchart of step 130 in the obstacle determination method shown;

[0045] Figure 6 The figure shows a flowchart of another embodiment of the obstacle determination method provided by this application;

[0046] Figure 7 As shown in Figure 6 The figure shows a flowchart of an embodiment of step 140 in the obstacle determination method shown;

[0047] Figure 8 The figure shows a schematic structural diagram of an embodiment of an obstacle determination device provided by the present application;

[0048] Figure 9 The figure shows a module block diagram of an embodiment of an electronic device provided by the present application. Detailed implementation manners

[0049] Here, exemplary embodiments will be described in detail, and examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with one or more embodiments of this specification. On the contrary, they are merely examples of devices and methods consistent with some aspects of one or more embodiments of this specification as detailed in the appended claims.

[0050] It should be noted that: In other embodiments, the steps of the corresponding method are not necessarily executed in the order shown and described in this specification. In some other embodiments, the steps included in the method may be more or less than those described in this specification. In addition, a single step described in this specification may be decomposed into multiple steps for description in other embodiments; and multiple steps described in this specification may also be combined into a single step for description in other embodiments.

[0051] Currently, millimeter-wave radar detection is widely used in the field of autonomous driving. The millimeter-wave radar emits directional millimeter waves. When the millimeter waves encounter an obstacle target and are reflected back, the position of the obstacle can be detected by receiving the reflected millimeter waves. However, because the millimeter-wave radar is prone to detecting many false positive points, using the original detection of the millimeter-wave radar will result in low accuracy in obstacle determination.

[0052] In an embodiment of the present application, an obstacle determination method is provided, which determines whether the target obstacles in each two adjacent frames of consecutive multiple frames meet the obstacle existence and disappearance conditions, so as to determine whether the target obstacles in the current frame actually exist. Thus, by determining whether the target obstacles actually exist through each two adjacent frames in consecutive multiple frames, the misjudgment caused by determining whether the target obstacles actually exist based on only one frame can be avoided, and the accuracy of obstacle determination can be relatively high.

[0053] Figure 1 The figure shows a system structure diagram of an embodiment of an obstacle detection system 10 provided by the present application.

[0054] As Figure 1As shown, the obstacle detection system 10 includes multiple vehicles 11 and a vehicle control center 12. The vehicle control center 12 can be a server that uniformly manages the multiple vehicles 11, and the multiple vehicles 11 are respectively communicatively connected to the vehicle control center 12. In some embodiments, the obstacle determination method provided in the embodiments of the present application can be applied to the vehicle control center 12. After the vehicle control center 12 executes the above obstacle determination method, it can plan a detour path for the vehicle 11 according to the actually existing target obstacle, and send the vehicle detour path and the control instruction for the vehicle to detour to the multiple vehicles 11, so that the multiple vehicles 11 control the vehicle 11 to travel according to the vehicle detour path and the control instruction to avoid the target obstacle.

[0055] In other embodiments, the obstacle determination method provided in the embodiments of the present application can be applied to any one of the vehicles 11 or multiple vehicles 11. After the vehicle 11 executes the above obstacle determination method, it can plan a detour path for the vehicle 11 according to the actually existing target obstacle, generate a control instruction for the vehicle to detour, and control the vehicle 11 to travel according to the vehicle detour path and the control instruction to avoid the target obstacle.

[0056] Figure 2 The figure shows a flowchart of an embodiment of the obstacle determination method provided by the present application.

[0057] As Figure 2 shown, the obstacle determination method may include the following steps 110 to 130:

[0058] Step 110, obtain multiple consecutive frames of target obstacles, where the target obstacle is an obstacle that needs to be tracked for whether it actually exists or disappears.

[0059] A frame refers to a single image frame, which is the smallest unit in an image animation. Multiple consecutive frames refer to multiple frames that appear adjacent to each other in sequence according to the chronological order. The "multiple" in multiple frames means greater than or equal to 2. Multiple consecutive frames can be used as the basis for judging a target obstacle, reflecting whether the target obstacle really exists or really disappears. For example, if the same target obstacle appears in multiple consecutive frames, it is considered that the target obstacle really exists. If the same target obstacle disappears in multiple consecutive frames, it is considered that the target obstacle really disappears. Compared with the misjudgment of a single frame, the judgment of whether the target obstacle really exists or really disappears based on multiple consecutive frames in the embodiments of the present application is more accurate. In this way, when the target obstacle appears stably, its real existence can be determined, thus avoiding false alarms of a single frame and reducing false judgments based on only one-frame data. Or, when the target obstacle disappears stably, its real disappearance can be determined, thus avoiding false alarms of a single frame and reducing false judgments based on only one-frame data. The number of multiple consecutive frames is not limited. There are multiple embodiments for obtaining multiple consecutive frames. In one embodiment, multiple consecutive frames in the historical moments before the current moment are obtained. For example, the latest 3 consecutive adjacent frames before the current moment of 23:45:54 on November 12, 2020, or, for another example, any 3 consecutive adjacent frames before the current moment of 23:45:54 on November 12, 2020. In another embodiment, the current frame at the current moment is obtained, the next frame adjacent to the current frame is obtained, the next frame adjacent to the next frame is obtained, and so on. For example, 3 consecutive frames after the current moment of 23:45:54 on November 12, 2020. During this process, after the current frame is obtained, the next frame adjacent to the current frame can be obtained, and the next frame is used as the current frame, so that the current frame can be obtained in real time. In this way, the real-time performance is higher, which is more beneficial to real-time applications.

[0060] The target obstacle is an obstacle that changes between every two adjacent frames among multiple consecutive frames. For example, an obstacle newly added or newly reduced in the current frame compared with the previous frame, which is an obstacle to track whether it really exists or really disappears. The target obstacle includes a first obstacle to be tracked and a second obstacle to be tracked. The first obstacle to be tracked is an obstacle for which it is necessary to determine whether it really disappears after it first disappears. The second obstacle to be tracked is an obstacle for which it is necessary to determine whether it really exists after it first exists.

[0061] There are multiple embodiments for obtaining the target obstacle. Figure 3 As shown Figure 2 A schematic flowchart of an embodiment of step 110 in the obstacle determination method shown. As Figure 3As shown, this step 110 may include steps 111 to 112. In one embodiment, this step 110 may include step 111 of acquiring obstacles in consecutive multiple frames; the obstacles in consecutive multiple frames include the obstacles in the current frame and the obstacles in the previous frame adjacent to the current frame. Step 112 is to compare the obstacles in the current frame with those in the previous frame, and the obstacles that can be found as the differences between the obstacles in the current frame and those in the previous frame are the target obstacles, and determine the first obstacle to be tracked and the second obstacle to be tracked in the current frame. The first obstacle to be tracked exists in the previous frame but does not exist in the current frame, and the second obstacle to be tracked does not exist in the previous frame but exists in the current frame. In this way, through the matching between the current frame and the previous frame, the target obstacles in the previous frame that need to be tracked in the current frame can be determined in a timely manner.

[0062] Exemplarily, the obstacles in the previous frame are respectively obstacle A, obstacle B, and obstacle C, and the obstacles in the current frame are respectively obstacle B, obstacle C, and obstacle D. Thus, since obstacle A exists in the previous frame but does not exist in the current frame, obstacle A is the first obstacle to be tracked. Since obstacle D does not exist in the previous frame but exists in the current frame, obstacle D is the second obstacle to be tracked.

[0063] Step 120 is to determine whether the target obstacles in every two adjacent frames among consecutive multiple frames reach the obstacle existence and disappearance condition.

[0064] The obstacle existence and disappearance condition refers to the condition for determining whether the target obstacle truly exists or truly disappears, and is used to reflect the life cycle of the target obstacle in which frames it has existed or in which frames it has disappeared. The obstacle existence and disappearance condition may include the obstacle true existence condition and / or the obstacle true disappearance condition, and the obstacle true existence condition is used to indicate that the obstacle truly exists.

[0065] Step 130, if the target obstacles in every two adjacent frames reach the obstacle true existence condition in the obstacle existence and disappearance condition, then determine that the target obstacles in the current frame truly exist.

[0066] Among them, the obstacle true existence condition may include, but is not limited to, whether the total number of matches of the target obstacles in every two adjacent frames among consecutive multiple frames reaches a preset existence number of times. Step 130 further includes: if the total number of matches of the second obstacle to be tracked in every two adjacent frames reaches the preset existence number of times, then determine that the second obstacle to be tracked truly exists. In this way, the second obstacle to be tracked can be determined by setting the preset existence number of times and matching the total number in every two adjacent frames among consecutive multiple frames, and then accurately determine whether the second obstacle to be tracked truly exists. The preset existence number of times can be set according to user needs. For example, the preset existence number of times may include 3 times.

[0067] In the embodiments of the present application, when the target obstacles in every two adjacent frames among a series of consecutive frames meet the condition of the actual existence of the obstacles, it is determined that the obstacles actually exist. The series of consecutive frames used in the embodiments of the present application avoid misjudgment caused by many false positive points detected by the millimeter-wave radar only appearing in one frame, and the accuracy of determining the target obstacles is relatively high. Moreover, the repeated judgment of the series of consecutive frames can also avoid misjudgment caused by accidental factors and improve data reliability. Further, during the driving process of the vehicle, the positions and speeds of the obstacles in the driving environment can be detected in real time, improving the driving safety of the vehicle.

[0068] Figure 4 is shown as Figure 2 a schematic flowchart of step 112 in the obstacle determination method shown as follows. As Figure 4 shown, the above step 112 may include steps 1121 to 1124.

[0069] Step 1121: Obtain the first point cloud data of the current frame and the second point cloud data of the previous frame; the first point cloud data and the second point cloud data are respectively obtained by detecting the environment around the vehicle through a millimeter-wave radar. The first point cloud data includes one or more of radial distance, horizontal azimuth angle, radial velocity, point reliability, and position error range; the second point cloud data includes one or more of radial distance, horizontal azimuth angle, radial velocity, point reliability, and position error range. Here, the number of millimeter-wave radars may be one or more, and it can be set on the vehicle. The specific number and installation position are not limited herein. In this way, the convenience of installing the millimeter-wave radar is improved. The above step 1121 further includes: each millimeter-wave radar can accumulate the point cloud data of consecutive frames to obtain the first point cloud data and the second point cloud data respectively. Detection by the millimeter-wave radar can achieve long-distance and stable detection.

[0070] Before step 1121, the method further includes: taking a predetermined timestamp as a reference, preprocessing the detected points in the point cloud data of the current frame to obtain the preprocessed point cloud data of the current frame as the first point cloud data; preprocessing the detected points in the point cloud data of the previous frame to obtain the preprocessed point cloud data of the previous frame as the second point cloud data. Among them, assuming that the movement speed of each detected point remains unchanged in a short period of time, the position of the detected point at the reference timestamp is calculated from the time difference and the speed, and other information remains unchanged. Among them, the preprocessing may include using point reliability and a filtering threshold to remove noise from the point cloud data of the current frame. The filtering threshold can be set according to the needs of the scenario and the task. The preprocessing may include using map information to remove the point cloud data that is not concerned in the point cloud data of the current frame. The point cloud data that is not concerned includes, for example, points not in the driving area and points whose distance from the vehicle is greater than a predetermined distance. The predetermined distance can be set according to user requirements.

[0071] In step 1121, the first point cloud data and the second point cloud data are determined to facilitate subsequent clustering of the first point cloud data and the second point cloud data respectively, and the obstacles are segmented.

[0072] Step 1122: Cluster multiple first detected points in the first point cloud data to determine the new obstacle list for the current frame. The new obstacle list includes at least one new obstacle, and each new obstacle corresponds to each category obtained by clustering. The first detected points among the multiple first detected points include speed, position, and / or shape. Among them, the obstacle speed is composed of the speeds of each detected point. Specifically, the speed of the obstacle is obtained by adding the speed vectors in each radial direction of each detected point. Among them, the magnitude of the speed in each radial direction is the average or the maximum / minimum value of all the radial speed magnitudes in the obstacle in this direction. For other attributes, according to the needs of the scenario and the task, when determining the speed magnitude in each radial direction, the maximum / minimum value or the average value can be selected.

[0073] The above step 1122 further includes using a clustering algorithm to cluster multiple first detected points in the first point cloud data. In this way, multiple first point cloud data can be aggregated into one category, and the positions, speed magnitudes, and directions of the first point cloud data in the same category are all close and less than the thresholds. The clustering algorithm includes the k-means algorithm, Density-Based Spatial Clustering of Applications with Noise (DBSCAN for short), and variant algorithms of DBSCAN, etc. No detailed examples are given here. The threshold can be a fixed threshold or a variable threshold. The variable threshold can be adjusted according to the distance. The position is two-dimensional data, and the speed is two-dimensional data.

[0074] Step 1123: Cluster multiple second detected points in the second point cloud data to determine the historical obstacle list for the previous frame. The historical obstacle list includes at least one historical obstacle, and each historical obstacle corresponds to each category obtained by clustering. The second detected points among the multiple second detected points include speed, position, and / or shape.

[0075] Step 1124: Compare the obstacles in the new obstacle list of the current frame with the obstacles in the historical obstacle list of the previous frame to determine the first obstacle to be tracked in the current frame and the second obstacle to be tracked in the previous frame.

[0076] In the related art, the comparison of point cloud with point cloud directly using the first point cloud data and the second point cloud data involves a large amount of calculation. Compared with the related art, in the embodiments of the present application, after clustering multiple first detected points in the first point cloud data and clustering multiple second detected points in the second point cloud data, in the comparison process, class-to-class comparison is used, and the amount of calculation is reduced. Moreover, in the clustering process, it is easier to cluster the associated detected points together, making the determined obstacles more accurate and providing a more reliable basis for subsequent execution steps.

[0077] In some embodiments, after step 112, the method further includes: recording the second obstacle to be tracked in the tracking obstacle list. In this way, the tracking obstacle list records the second obstacle to be tracked, which is equivalent to storing the second obstacle to be tracked, facilitating the later extraction and use of the second obstacle to be tracked. Each obstacle in the tracking obstacle list can save the historical information of the obstacle. Figure 5 As shown Figure 2 A flowchart of an embodiment of step 130 in the obstacle determination method shown. In one embodiment of the above step 130, in combination with Figure 3 and Figure 5 as shown, step 130 further includes steps 131 to 136:

[0078] Step 131, obtain the next frame adjacent to the current frame as the current frame.

[0079] Step 132, match the obstacles in the current frame with the obstacles in the tracking obstacle list to determine whether there is a second obstacle to be tracked in the current frame. If so, that is, there is a second obstacle to be tracked in the current frame, then execute step 133; if not, that is, there is no second obstacle to be tracked in the current frame, then execute step 134.

[0080] The above step 132 further includes the following steps 1 and 2. Step 1: Determine the association matrix between the obstacles in the current frame and the obstacles in the tracking obstacle list, and obtain the association scores between the obstacles in the current frame and the obstacles in the tracking obstacle list. Each value of the association scores represents the association score between every two obstacles among the obstacles in the current frame and the obstacles in the tracking obstacle list. The higher the association score, the higher the matching degree. The calculation method of the association scores may include the intersection-over-union (IOU) method. The IOU method is a standard for measuring the accuracy of detecting corresponding objects in a specific dataset. Further, obtaining the association scores between the obstacles in the current frame and the obstacles in the tracking obstacle list may further include using position, area, and velocity to obtain the association scores between the obstacles in the current frame and the obstacles in the tracking obstacle list. Step 2: Use a matching method to match the obstacles in the current frame with the obstacles in the tracking obstacle list, and select the obstacles in the current frame and the obstacles in the tracking obstacle list with the highest total score and the highest association. In this way, the matching degree between the obstacles in the current frame and the obstacles in the tracking obstacle list can be determined, and thus it can be determined whether there is a target obstacle.

[0081] Among them, the matching method includes but is not limited to the bipartite graph matching method. Among them, the bipartite graph matching method may include the Hungarian matching method. The matching method is selected according to actual needs. In some embodiments, if it is determined that the association scores between the obstacles in the current frame and the obstacles in the tracking obstacle list are greater than or equal to a predetermined threshold, it is determined that the obstacles in the current frame and the obstacles in the tracking obstacle list are matched, indicating successful tracking. If it is determined that the association scores between the obstacles in the current frame and the obstacles in the tracking obstacle list are less than the predetermined threshold, it is determined that the obstacles in the current frame and the obstacles in the tracking obstacle list are not matched, indicating failed tracking. The target obstacle is marked as lost, and a new target obstacle is added to the tracking obstacle list. Among them, the predetermined threshold can be determined according to the accuracy required by the user. The higher the accuracy of the target obstacle required by the user, the larger the predetermined threshold; the lower the accuracy of the target obstacle required by the user, the smaller the predetermined threshold.

[0082] Step 133: Determine whether the total number of matches of the second obstacle to be tracked in every two adjacent frames reaches a preset number of occurrences. If not, that is, the total number of matches of the second obstacle to be tracked in every two adjacent frames does not reach the preset number of occurrences, then execute step 135. If so, that is, the total number of matches of the second obstacle to be tracked in every two adjacent frames reaches the preset number of occurrences, then execute step 136.

[0083] Step 134, stop tracking the second obstacle to be tracked. There are multiple embodiments of step 134. In one embodiment of step 134, delete the second obstacle to be tracked from the list of tracked obstacles, so that the second obstacle to be tracked that does not need to be tracked in the list of tracked obstacles can be cleared in time, and the data volume becomes smaller, which is more conducive to subsequent matching with obstacles in other frames. In another embodiment of this step 134, take the second obstacle to be tracked as the first obstacle to be tracked, record the first obstacle to be tracked in the list of tracked obstacles, obtain the next frame adjacent to the current frame as the current frame; match the obstacles in the current frame with the obstacles in the list of tracked obstacles to determine whether there is a first obstacle to be tracked in the current frame. If there is no first obstacle to be tracked in the current frame and the disappearance times of the first obstacle to be tracked in each two adjacent frames reach the preset disappearance times, then determine that the first obstacle to be tracked is an obstacle that has disappeared.

[0084] Step 135, obtain the next frame adjacent to the current frame as the current frame, and continue to return to step 132 for execution.

[0085] Step 136, determine that the second obstacle to be tracked actually exists.

[0086] In the above embodiment of this step 130, obtain the next frame adjacent to the current frame as the current frame, so as to determine that the target obstacle in each two adjacent frames of consecutive frames meets the condition that the obstacle actually exists, which is more conducive to mastering the real-time information of the target obstacle and improving the accuracy of determining the actual existence of the target obstacle.

[0087] In another embodiment of the above step 130, step 130 may further include: take the current frame as the previous frame, obtain the next frame adjacent to the current frame as the current frame; compare the obstacles in the current frame with the obstacles in the previous frame to determine whether there is a second obstacle to be tracked in the current frame. If there is no second obstacle to be tracked in the current frame, then mark the second obstacle to be tracked as the first obstacle to be tracked, and record the disappearance times of the first obstacle to be tracked, and continue to return to the step of taking the current frame as the previous frame and obtaining the next frame adjacent to the current frame as the current frame for execution until there is no first obstacle to be tracked in the current frame and whether the total number of disappearance times of the obstacles to be tracked in each two adjacent frames reaches the preset disappearance times, then determine that the first obstacle to be tracked has actually disappeared. If there is a second obstacle to be tracked in the current frame, then mark the number of times the second obstacle to be tracked exists, and continue to return to the step of taking the current frame as the previous frame and obtaining the next frame adjacent to the current frame as the current frame for execution until there is a second obstacle to be tracked in the current frame and whether the total number of matches of the second obstacle to be tracked in each two adjacent frames reaches the preset existence times, then determine that the second obstacle to be tracked actually exists.

[0088] Exemplarily, the condition for the real existence of an obstacle may include that it exists for at least two consecutive frames after the current frame. The condition for the real disappearance of an obstacle is used to indicate the real disappearance of the obstacle. Exemplarily, the condition for the real disappearance of an obstacle may include that it disappears in one consecutive frame after the current frame. For example, if the obstacles obtained in the next adjacent frame after the current frame are obstacle B, obstacle C, and obstacle D respectively, then obstacle A really disappears. If the obstacles obtained in the next adjacent frame after the next frame are obstacle D respectively, then obstacle D really exists, and obstacle B and obstacle C are the first obstacles to be tracked respectively.

[0089] Figure 6 The figure shows a schematic flowchart of another embodiment of the obstacle determination method provided by this application.

[0090] Figure 6 The embodiment of... is similar to Figures 2 to 5 the embodiment shown in Figures 2 to 5 In the embodiment shown in Figure 6 of the embodiment, the obstacle existence and disappearance condition includes the condition for the real disappearance of an obstacle; after determining whether the target obstacle in each two adjacent frames among multiple consecutive frames reaches the obstacle existence and disappearance condition in the obstacle existence and disappearance condition, the method further includes step 140. If the target obstacle in each two adjacent frames reaches the condition for the disappearance of an obstacle in the obstacle existence and disappearance condition, it is determined that the obstacle in the current frame really disappears. The serial numbers of step 140 and step 130 are not limited in terms of the order of execution. When the execution conditions of step 140 and step 130 are both met, they can also be executed simultaneously. Other realizable execution orders also fall within the protection scope of the embodiments of this application and will not be elaborated here.

[0091] The above-mentioned condition for the real disappearance of an obstacle includes whether the number of disappearances of the first obstacle to be tracked in each two adjacent frames among multiple consecutive frames reaches a preset number of disappearances; step 140 can further include: if the number of disappearances of the first obstacle to be tracked in each two adjacent frames reaches the preset number of disappearances, it is determined that the first obstacle to be tracked really disappears. In this way, the first obstacle to be tracked can be determined by setting the preset number of disappearances and matching the total number in each two adjacent frames among multiple consecutive frames, and then accurately determining whether the first obstacle to be tracked really disappears. The preset number of disappearances can be set according to user requirements. For example, the preset number of disappearances can include 2 times.

[0092] In the embodiments of this application, when the target obstacle in each two adjacent frames among multiple consecutive frames reaches the condition for the real disappearance of an obstacle, it is determined that the obstacle really disappears. In this way, the embodiments of this application use multiple consecutive frames to avoid misjudgment caused by many false positive points detected by the millimeter-wave radar only appearing in one frame, improve the accuracy of determining the target obstacle, and the repeated judgment of multiple consecutive frames can also avoid misjudgment caused by accidental factors and improve the data reliability.

[0093] In some embodiments, after step 112, the method further includes: recording the first obstacle to be tracked in a list of obstacles to be tracked. Thus, recording the first obstacle to be tracked in the list of obstacles to be tracked is equivalent to storing the first obstacle to be tracked, which can facilitate the subsequent use of the first obstacle to be tracked extracted from the list of obstacles to be tracked. Figure 7 As shown Figure 6 A schematic flowchart of an embodiment of step 140 in the obstacle determination method shown. In one of the above embodiments of step 140, in combination with Figure 6 As shown, step 140 further includes steps 141 to 146:

[0094] Step 141, obtain the next frame adjacent to the current frame as the current frame.

[0095] Step 142, match the obstacles in the current frame with the obstacles in the list of obstacles to be tracked, and determine whether there is a first obstacle to be tracked in the current frame. If not, that is, there is no first obstacle to be tracked in the current frame, then execute step 143. If so, that is, there is a first obstacle to be tracked in the current frame, then execute step 144.

[0096] Step 143, determine whether the disappearance times of the first obstacle to be tracked in every two adjacent frames reach a preset disappearance times. If not, that is, the disappearance times of the first obstacle to be tracked in every two adjacent frames do not reach the preset disappearance times, then execute step 145; if so, that is, the disappearance times of the first obstacle to be tracked in every two adjacent frames reach the preset disappearance times, then execute step 146.

[0097] Step 144, stop tracking the first obstacle to be tracked. There are multiple embodiments of this step 144. In one embodiment of step 144, delete the first obstacle to be tracked from the tracking obstacle list. In this way, the first obstacle to be tracked that does not need to be tracked in the tracking obstacle list can be cleared in time, the data volume becomes smaller, which is more conducive to subsequent matching with other frames. In another embodiment of this step 144, take the first obstacle to be tracked as the second obstacle to be tracked, record the second obstacle to be tracked in the tracking obstacle list, obtain the next frame adjacent to the current frame as the current frame; match the obstacles in the current frame with the tracking obstacle list to determine whether there is a second obstacle to be tracked in the current frame. If there is a second obstacle to be tracked in the current frame, then determine whether the total number of matches of the second obstacle to be tracked in every two adjacent frames reaches the preset existence times; if the total number of matches of the second obstacle to be tracked in every two adjacent frames does not reach the preset existence times, then obtain the next frame adjacent to the current frame as the current frame, and continue to return to the step of matching the current frame with the tracking obstacle list to determine whether there is a second obstacle to be tracked in the current frame until there is a second obstacle to be tracked in the current frame and the total number of matches of the second obstacle to be tracked in every two adjacent frames reaches the preset existence times, then determine that the second obstacle to be tracked really exists.

[0098] Step 145, obtain the next frame adjacent to the current frame as the current frame, and continue to return to step 142 for execution.

[0099] Step 146, determine that the first obstacle to be tracked really disappears.

[0100] In the above embodiment of this step 140, obtain the next frame adjacent to the current frame as the current frame, so as to determine that the target obstacle in every two adjacent frames of consecutive frames reaches the obstacle real disappearance condition, which is more conducive to mastering the real-time information of the target obstacle and is beneficial to improving the accuracy of determining the real disappearance of the target obstacle.

[0101] Figure 8 The following shows a schematic structural diagram of an embodiment of an obstacle determination device 20 provided by the present application.

[0102] The present application provides an obstacle determination device 20, including:

[0103] An acquisition unit 21, configured to acquire target obstacles in consecutive multiple frames, where the target obstacle is an obstacle that needs to be tracked for whether it really exists or disappears;

[0104] A first processing unit 22, configured to determine whether the target obstacles in every two adjacent frames of consecutive multiple frames reach the obstacle existence and disappearance conditions; the obstacle existence and disappearance conditions include an obstacle real existence condition, and the obstacle real existence condition is used to indicate that the obstacle really appears;

[0105] A second processing unit 23, configured to determine that the target obstacle in the current frame actually exists if the target obstacles in every two adjacent frames meet the obstacle existence and disappearance conditions.

[0106] The second processing unit 23 is further configured to determine that the obstacle in the current frame actually disappears if every two adjacent frames in multiple consecutive frames meet the obstacle disappearance condition.

[0107] Figure 9 The following shows a block diagram of a module according to an embodiment of an electronic device 30 provided by the present application. Figure 9 The following is a schematic hardware structure diagram of an electronic device provided by an embodiment of the present application. The electronic device may include a processor 31, a memory 33 storing machine-executable instructions, and a communication interface 32. The processor 31 and the memory 33 may communicate via a system bus 34. And by reading and executing the machine-executable instructions corresponding to the data pulling or data feedback logic in the memory 33, the processor 31 may execute the methods described above.

[0108] The memory 33 mentioned herein may be any electronic, magnetic, optical, or other physical storage device that can contain or store information, such as executable instructions, data, and so on. For example, the machine-readable storage medium may be: RAM (Random Access Memory), volatile memory, non-volatile memory, flash memory, storage drives (such as hard disk drives), solid-state drives, any type of storage disk (such as optical discs, DVDs, etc.), or similar storage media, or a combination thereof.

[0109] In some embodiments, a machine-readable storage medium is also provided, such as Figure 9 the memory 33 in the above, which stores machine-executable instructions. When the machine-executable instructions are executed by a processor, the methods described above are implemented. For example, the machine-readable storage medium may be ROM, RAM, CD-ROM, magnetic tapes, floppy disks, and optical data storage devices, etc.

[0110] The embodiments of the present application also provide a computer program, stored in a machine-readable storage medium, such as Figure 9 the memory 33 in the above, and when the processor executes the computer program, it causes the processor 31 to execute the methods described above.

[0111] The above are only the preferred embodiments of this specification and are not intended to limit this specification. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this specification shall be included within the scope of protection of this specification.

[0112] It should also be noted that the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, commodity or device comprising said element.

Claims

1. An obstacle detection method, characterized in that, Including: Obtaining target obstacles in a series of consecutive frames, where the target obstacles are obstacles for which it is necessary to track whether they actually exist or disappear; Determining whether the target obstacles in every two adjacent frames among the series of consecutive frames meet the obstacle existence / non-existence condition; The obstacle existence / non-existence condition includes an obstacle actual existence condition; The obstacle existence / non-existence condition refers to the condition for determining whether a target obstacle actually exists or actually disappears; the obstacle actual existence condition is used to indicate that an obstacle actually exists; If the target obstacles in every two adjacent frames meet the obstacle actual existence condition, then it is determined that the target obstacles in the current frame actually exist.

2. The obstacle detection method according to claim 1, wherein The obtaining of the target obstacles in a series of consecutive frames includes: Obtaining obstacles in a series of consecutive frames; the obstacles in the series of consecutive frames include the obstacles in the current frame and the obstacles in the previous frame adjacent to the current frame; Comparing the obstacles in the current frame with the obstacles in the previous frame to determine the first obstacle to be tracked and the second obstacle to be tracked in the current frame. The first obstacle to be tracked and the second obstacle to be tracked are respectively included in the target obstacles. The first obstacle to be tracked exists in the previous frame and does not exist in the current frame, and the second obstacle to be tracked does not exist in the previous frame and exists in the current frame.

3. The obstacle detection method according to claim 2, characterized in that, The obstacle actual existence condition includes whether the total number of matches of the target obstacles in every two adjacent frames among the series of consecutive frames reaches a preset number of existence times; The step of, if the target obstacles in every two adjacent frames meet the obstacle actual existence condition, then determining that the target obstacles in the current frame actually exist, includes: If the total number of matches of the second obstacle to be tracked in every two adjacent frames reaches the preset number of existence times, then it is determined that the second obstacle to be tracked actually exists.

4. The obstacle detection method according to claim 3, wherein After comparing the obstacles in the current frame with the obstacles in the previous frame to determine the first obstacle to be tracked and the second obstacle to be tracked in the current frame, the method further includes: Recording the second obstacle to be tracked in a list of obstacles to be tracked; The step of, if the total number of matches of the second obstacle to be tracked in every two adjacent frames reaches the preset number of existence times, then determining that the second obstacle to be tracked actually exists, includes: Obtaining the next frame adjacent to the current frame as the current frame; Matching the obstacles in the current frame with the obstacles in the list of obstacles to be tracked to determine whether there is a second obstacle to be tracked in the current frame. If there is a second obstacle to be tracked in the current frame, then determine whether the total number of matches of the second obstacle to be tracked in every two adjacent frames reaches the preset number of existence times; if the total number of matches of the second obstacle to be tracked in every two adjacent frames does not reach the preset number of existence times, then obtain the next frame adjacent to the current frame as the current frame, and continue to return to the step of matching the current frame with the list of obstacles to be tracked to determine whether there is a second obstacle to be tracked in the current frame until there is a second obstacle to be tracked in the current frame and the total number of matches of the second obstacle to be tracked in every two adjacent frames reaches the preset number of existence times, then it is determined that the second obstacle to be tracked actually exists.

5. The obstacle detection method according to claim 2, wherein The obstacle existence / non-existence condition includes an obstacle actual disappearance condition; After determining whether the target obstacles in each two adjacent frames of the continuous multiple frames reach the obstacle existence / loss condition, the method further includes: If the target obstacles in each two adjacent frames reach the true obstacle disappearance condition, determine that the obstacles in the current frame truly disappear.

6. The obstacle detection method according to claim 5, wherein, The true obstacle disappearance condition includes whether the disappearance times of the first obstacle to be tracked in each two adjacent frames of the continuous multiple frames reach a preset disappearance times; The step of if the target obstacles in each two adjacent frames reach the true obstacle disappearance condition, determine that the obstacles in the current frame truly disappear, includes: If the disappearance times of the first obstacle to be tracked in each two adjacent frames reach the preset disappearance times, determine that the first obstacle to be tracked truly disappears.

7. The obstacle detection method according to claim 6, characterized in that, After comparing the obstacles in the current frame with the obstacles in the previous frame to determine the first obstacle to be tracked and the second obstacle to be tracked in the current frame, the method further includes: Record the first obstacle to be tracked in the tracking obstacle list; The step of if the disappearance times of the first obstacle to be tracked in each two adjacent frames reach the preset disappearance times, determine that the first obstacle to be tracked truly disappears, includes: Obtain the next frame adjacent to the current frame as the current frame; Match the obstacles in the current frame with the obstacles in the tracking obstacle list to determine whether there is a first obstacle to be tracked in the current frame. If there is no first obstacle to be tracked in the current frame and the disappearance times of the first obstacle to be tracked in each two adjacent frames reach the preset disappearance times, determine that the first obstacle to be tracked truly disappears.

8. The obstacle detection method according to claim 2, wherein, The step of comparing the obstacles in the current frame with the obstacles in the previous frame to determine the first obstacle to be tracked and the second obstacle to be tracked in the current frame, includes: Obtain the first point cloud data of the current frame and the second point cloud data of the previous frame; the first point cloud data and the second point cloud data are respectively obtained by detecting the environment around the vehicle through a millimeter-wave radar; Cluster a plurality of first detected points in the first point cloud data to determine a new obstacle list of the current frame, the new obstacle list includes at least one new obstacle, and each new obstacle corresponds to each category obtained by clustering; Cluster a plurality of second detected points in the second point cloud data to determine a historical obstacle list of the previous frame, the historical obstacle list includes at least one historical obstacle, and each historical obstacle corresponds to each category obtained by clustering; Compare the obstacles in the new obstacle list of the current frame with the obstacles in the historical obstacle list of the previous frame to determine the first obstacle to be tracked in the current frame and the second obstacle to be tracked in the previous frame.

9. The obstacle detection method according to claim 4, wherein After the target obstacles in the current frame truly exist, the method further includes: deleting the second obstacle to be tracked in the tracking obstacle list.

10. An electronic device, characterized in that, It includes a processor and a memory; The memory is used for storing a computer program; The processor is used for implementing the obstacle detection method according to any one of claims 1-9 when executing the program stored in the memory.

Images