A radar-based target detection method and apparatus

Abstract

The specification discloses a radar-based target detection method and device. First, a first point cloud collected by a radar in a first detection mode and a second point cloud collected by the radar in a second detection mode are obtained. Second, for each point cloud point, an initial confidence corresponding to the point cloud point is determined according to observation stability information generated by a detection model to which the point cloud point belongs when the point cloud point is obtained. Then, the initial confidence corresponding to the point cloud point is adjusted according to a neighborhood matching result of the point cloud point in another point cloud other than the point cloud to which the point cloud point belongs, to obtain an adjusted confidence corresponding to the point cloud point. Finally, according to the adjusted confidence corresponding to each point cloud point included in the first point cloud and the second point cloud, a target point cloud point is determined from the first point cloud and the second point cloud, and target detection is performed based on the target point cloud point. The method effectively combines the point cloud data of the first detection mode and the point cloud data of the second detection mode, and improves the accuracy of the target detection result.

Description

TECHNICAL FIELD

[0001] The present specification relates to the field of unmanned driving, and in particular, to a radar-based target detection method and device. BACKGROUND

[0002] Currently, unmanned vehicles are usually equipped with millimeter wave radars, which are generally divided into two modes, namely, a near-distance mode and a far-distance mode. The near-distance mode is used to identify vehicles in the lateral direction, has a large angle range of observation, and has high distance resolution. The far-distance mode is used to identify vehicles far away in the adjacent lane, and has a long detection distance. Specifically, as shown in Figure 2

[0003] In actual applications, the millimeter wave radar on the unmanned aerial vehicle usually follows the method of the millimeter wave radar on the vehicle, which is divided into a near-distance mode and a far-distance mode. The point cloud data collected by the millimeter wave radar on the unmanned aerial vehicle is used to detect obstacles around the unmanned aerial vehicle. However, the application scenarios of the unmanned aerial vehicle and the unmanned vehicle are not the same in practice. As can be seen from Figure 2 , if the method of the millimeter wave radar on the vehicle is applied to the unmanned aerial vehicle, the far-distance mode of the unmanned aerial vehicle cannot detect obstacles in a large range on both sides of the unmanned aerial vehicle, resulting in an unreasonable flight trajectory planned by the unmanned aerial vehicle, and the unmanned aerial vehicle may fly towards the obstacles. Even if the near-distance mode of the unmanned aerial vehicle detects the obstacles, a collision may still occur due to the flight speed, calculation delay, and the like. That is, the unmanned aerial vehicle following the method of the millimeter wave radar on the vehicle cannot timely detect the obstacles around the unmanned aerial vehicle.

[0004] Therefore, how to timely detect the obstacles around the unmanned aerial vehicle and improve the accuracy of the detection results is a problem to be solved. SUMMARY

[0005] The present specification provides a radar-based target detection method and device to partially solve the above problems existing in the prior art.

[0006] The present specification adopts the following technical solutions:

[0007] The present specification provides a radar-based target detection method, wherein the radar is provided with a first detection mode and a second detection mode, the central angles or radii of the detection ranges of the first detection mode and the second detection mode are different, and the method comprises the following steps:

[0008] acquiring point cloud data collected by the radar in the first detection mode as first point cloud and point cloud data collected by the radar in the second detection mode as second point cloud;

[0009] ​determining, for each point cloud point included in the first point cloud and the second point cloud, an initial confidence degree corresponding to the point cloud point according to observation stability information generated by a detection model to which the point cloud point belongs when the point cloud point is obtained;

[0010] adjusting the initial confidence degree corresponding to the point cloud point according to a neighborhood matching result of the point cloud point in another point cloud other than the point cloud to which the point cloud point belongs, to obtain an adjusted confidence degree corresponding to the point cloud point;

[0011] determining a target point cloud point from the first point cloud and the second point cloud according to the adjusted confidence degree corresponding to each point cloud point included in the first point cloud and the second point cloud, and performing target detection according to the target point cloud point.

[0012] Optionally, the observation stability information includes at least one of a continuous observation frame number corresponding to the point cloud point and an observation continuous loss frame number corresponding to the point cloud point.

[0013] determining, for each point cloud point included in the first point cloud and the second point cloud, an initial confidence degree corresponding to the point cloud point according to observation stability information generated by a detection model to which the point cloud point belongs when the point cloud point is obtained, specifically including:

[0014] determining, for each point cloud point included in the first point cloud and the second point cloud, an initial confidence degree corresponding to the point cloud point according to a continuous observation frame number and / or an observation continuous loss frame number corresponding to the point cloud point under the detection model to which the point cloud point belongs, wherein the higher the continuous observation frame number corresponding to the point cloud point under the detection model to which the point cloud point belongs, the higher the initial confidence degree corresponding to the point cloud point, and the higher the observation continuous loss frame number corresponding to the point cloud point under the detection model to which the point cloud point belongs, the lower the initial confidence degree corresponding to the point cloud point.

[0015] Optionally, adjusting the initial confidence degree corresponding to the point cloud point according to a neighborhood matching result of the point cloud point in another point cloud other than the point cloud to which the point cloud point belongs, to obtain an adjusted confidence degree corresponding to the point cloud point, specifically includes:

[0016] if it is determined according to the neighborhood matching result that there is a point cloud point in the other point cloud whose position is less than a set distance from the point cloud point, increasing the initial confidence degree corresponding to the point cloud point to obtain an adjusted confidence degree corresponding to the point cloud point.

[0017] Optionally, if it is determined according to the neighborhood matching result that there is a point cloud point in the other point cloud whose position is less than a set distance from the point cloud point, increasing the initial confidence degree corresponding to the point cloud point to obtain an adjusted confidence degree corresponding to the point cloud point, specifically includes:

[0018] If it is determined according to the neighborhood matching result that there is a point cloud point in the other point cloud whose position is less than a set distance from the point cloud point, it is determined whether the initial confidence corresponding to the point cloud point is greater than a first confidence threshold value;

[0019] If it is determined that the initial confidence corresponding to the point cloud point is greater than the first confidence threshold value, the initial confidence corresponding to the point cloud point is increased to obtain an adjusted confidence corresponding to the point cloud point.

[0020] Optionally, if it is determined that the initial confidence corresponding to the point cloud point is greater than the first confidence threshold value, the initial confidence corresponding to the point cloud point is increased to obtain an adjusted confidence corresponding to the point cloud point, and specifically includes:

[0021] If it is determined that the initial confidence corresponding to the point cloud point is greater than the first confidence threshold value, the initial confidence corresponding to the point cloud point is increased by a preset confidence compensation value to obtain a compensated confidence corresponding to the point cloud point;

[0022] It is determined whether the compensated confidence corresponding to the point cloud point exceeds a preset maximum confidence;

[0023] If yes, the maximum confidence is taken as the adjusted confidence corresponding to the point cloud point, and if not, the compensated confidence corresponding to the point cloud point is taken as the adjusted confidence corresponding to the point cloud point.

[0024] Optionally, the initial confidence corresponding to the point cloud point is adjusted according to the neighborhood matching result of the point cloud point in the other point cloud to which the point cloud point belongs to obtain an adjusted confidence corresponding to the point cloud point, and specifically includes:

[0025] If it is determined according to the neighborhood matching result that there is no point cloud point in the other point cloud whose position is less than a set distance from the point cloud point, the initial confidence corresponding to the point cloud point is decreased to obtain an adjusted confidence corresponding to the point cloud point.

[0026] Optionally, if it is determined according to the neighborhood matching result that there is no point cloud point in the other point cloud whose position is less than a set distance from the point cloud point, the initial confidence corresponding to the point cloud point is decreased to obtain an adjusted confidence corresponding to the point cloud point, and specifically includes:

[0027] If it is determined according to the neighborhood matching result that there is no point cloud point in the other point cloud whose position is less than a set distance from the point cloud point, it is determined whether the initial confidence corresponding to the point cloud point is less than a second confidence threshold value;

[0028] If it is determined that the initial confidence corresponding to the point cloud point is less than the second confidence threshold value, the initial confidence corresponding to the point cloud point is decreased to obtain an adjusted confidence corresponding to the point cloud point.

[0029] Optionally, a central angle of the detection range of the first detection mode is not less than a central angle of the detection range of the second detection mode, and a radius of the detection range of the first detection mode is not less than a radius of the detection range of the second detection mode; or,

[0030] a central angle of the detection range of the second detection mode is not less than a central angle of the detection range of the first detection mode, and a radius of the detection range of the second detection mode is not less than a radius of the detection range of the first detection mode.

[0031] The present specification provides a radar-based target detection device, the radar being provided with a first detection mode and a second detection mode, a central angle or a radius of a detection range of the first detection mode and the second detection mode being different, comprising:

[0032] an acquisition module, configured to acquire point cloud data collected by the radar in the first detection mode as a first point cloud, and point cloud data collected by the radar in the second detection mode as a second point cloud;

[0033] a determination module, configured to determine, for each point cloud point included in the first point cloud and the second point cloud, an initial confidence corresponding to the point cloud point according to observation stability information generated by a detection model to which the point cloud point belongs when the point cloud point is obtained;

[0034] an adjustment module, configured to adjust the initial confidence corresponding to the point cloud point according to a neighborhood matching result of the point cloud point in another point cloud other than the point cloud to which the point cloud point belongs, to obtain an adjusted confidence corresponding to the point cloud point;

[0035] a detection module, configured to determine, from the first point cloud and the second point cloud, a target point cloud point according to the adjusted confidence corresponding to each point cloud point included in the first point cloud and the second point cloud, and perform target detection according to the target point cloud point.

[0036] The present specification provides a computer-readable storage medium, the storage medium storing a computer program, the computer program being executed by a processor to implement the radar-based target detection method.

[0037] The present specification provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor implementing the radar-based target detection method when executing the program.

[0038] The above at least one technical solution adopted by the present specification can achieve the following beneficial effects:

[0039] In the radar-based target detection method provided in the present specification, first, point cloud data collected by a radar in a first detection mode is obtained as a first point cloud, and point cloud data collected by the radar in a second detection mode is obtained as a second point cloud. Second, for each point cloud point contained in the first point cloud and the second point cloud, an initial confidence corresponding to the point cloud point is determined according to observation stability information generated by a detection model to which the point cloud point belongs when the point cloud point is obtained. Then, the initial confidence corresponding to the point cloud point is adjusted according to a neighborhood matching result of the point cloud point in another point cloud other than the point cloud to which the point cloud point belongs, to obtain an adjusted confidence corresponding to the point cloud point. Finally, target point cloud points are determined from the first point cloud and the second point cloud according to the adjusted confidence corresponding to each point cloud point contained in the first point cloud and the second point cloud, and target detection is performed according to the target point cloud points.

[0040] As can be seen from the above method, the present method can detect obstacles around in a relatively timely manner through a radar designed for the application scenario of the unmanned aerial vehicle. Target point cloud points are determined from the first point cloud and the second point cloud according to the adjusted confidence corresponding to each point cloud point contained in the first point cloud and the second point cloud, and target detection is performed according to the target point cloud points. Compared with the prior art, the present method has the radar installed on the unmanned aerial vehicle, which meets the target detection requirements of the unmanned aerial vehicle, can detect obstacles around in a relatively timely manner, and effectively combines the point cloud data of the first detection mode and the point cloud data of the second detection mode, thereby improving the accuracy of the target detection result. BRIEF DESCRIPTION OF DRAWINGS

[0041] The accompanying drawings, which are included to provide a further understanding of the present specification, constitute a part of the present specification, and the illustrative embodiments of the present specification and their descriptions serve to explain the present specification and do not constitute an improper limitation on the present specification. In the drawings:

[0042] Figure 1 A flowchart of a radar-based target detection method in the present specification;

[0043] Figure 2 A schematic diagram of a detection range of a millimeter wave radar in the prior art provided by the embodiment of the present specification;

[0044] Figure 3 A schematic diagram of a detection range of a radar installed on an unmanned aerial vehicle provided by the embodiment of the present specification;

[0045] Figure 4A 、 Figure 4B A schematic diagram of a detection process of a radar provided by the embodiment of the present specification;

[0046] Figure 5 A schematic diagram of a radar-based target detection device provided by the present specification;

[0047] Figure 6 A schematic view of an electronic device corresponding to Figure 1 of the present specification. DETAILED DESCRIPTION

[0048] For the purpose of making the purpose, technical scheme and advantages of the present specification clearer, the technical scheme of the present specification will be described clearly and completely in combination with the specific embodiments of the present specification and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present specification, not all the embodiments. Based on the embodiments in the present specification, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present specification.

[0049] The technical scheme provided by each embodiment of the present specification will be described in detail below in combination with the drawings.

[0050] Figure 1 A flowchart of a radar-based target detection method in the present specification, comprising the following steps:

[0051] S100: Obtain the point cloud data collected by the radar in the first detection mode as the first point cloud, and the point cloud data collected in the second detection mode as the second point cloud.

[0052] In the prior art, unmanned vehicles are equipped with millimeter wave radars, which are generally divided into two modes, namely, near-distance mode and far-distance mode. Among them, the near-distance mode of the unmanned vehicle is used to identify the vehicles in the transverse direction, so the near-distance mode of the unmanned vehicle requires a wider beam width, a shorter detection distance, and a higher distance resolution. The far-distance mode of the unmanned vehicle is used to identify vehicles in the adjacent lane at a distance, so the far-distance mode of the unmanned vehicle requires a narrower beam width, a longer detection distance, and a lower distance resolution.

[0053] In actual application, the unmanned vehicle radar on the unmanned aerial vehicle generally follows the method of the vehicle-mounted millimeter wave radar. Since the application scenarios of the unmanned aerial vehicle and the unmanned vehicle are not the same in practice, if the method of the vehicle-mounted millimeter wave radar is applied to the unmanned aerial vehicle, the unmanned aerial vehicle cannot detect obstacles in a large range on both sides of the unmanned aerial vehicle at a long distance, resulting in that the unmanned aerial vehicle cannot detect the obstacles on both sides of the unmanned aerial vehicle in a timely manner. Moreover, in the application scenario of the unmanned aerial vehicle, static targets are mainly identified, and dynamic obstacles crossing the flight path can generally not be considered, and a narrower beam width is required at a short distance. Further, static targets can exist in various forms of distribution, and a higher distance resolution is required to obtain better detection effect.

[0054] Therefore, the near mode of the UAV is used to identify small obstacles (wires, branches, etc.) in front of the UAV, and requires a narrower beam width, a shorter detection distance, and a higher distance resolution. The far mode of the UAV is used to detect common large obstacles (trees, buildings, etc.) in a larger range, and requires a wider beam width, a longer detection distance, and a lower distance resolution.

[0055] To solve the above problems, a radar that meets the target detection requirements of the UAV can be installed on the UAV, and the radar is provided with a first detection mode and a second detection mode. The central angles or radii of the detection ranges of the first detection mode and the second detection mode are different.

[0056] Specifically, the central angle of the detection range of the first detection mode is not less than the central angle of the detection range of the second detection mode, and the radius of the detection range of the first detection mode is not less than the radius of the detection range of the second detection mode. Alternatively, the central angle of the detection range of the second detection mode is not less than the central angle of the detection range of the first detection mode, and the radius of the detection range of the second detection mode is not less than the radius of the detection range of the first detection mode.

[0057] That is, if the first detection mode is a far mode, the second detection mode is a near mode. If the first detection mode is a near mode, the second detection mode is a far mode. As shown in Figure 3 .

[0058] Figure 3 A schematic diagram of the detection range of the radar installed on the UAV according to an embodiment of the present disclosure is shown.

[0059] In the Figure 3 , if the first detection mode is a far mode, the second detection mode is a near mode. The central angle of the detection range of the first detection mode is greater than the central angle of the detection range of the second detection mode, the radius of the detection range of the first detection mode is greater than the radius of the detection range of the second detection mode, and the distance resolution of the first detection mode is less than the distance resolution of the second detection mode. In the process of target detection of the UAV, the distance resolution of the second detection mode is higher, which can better distinguish adjacent small obstacles. As can be seen from Figure 3 , the UAV can better distinguish the wires from the adjacent wire poles, and because the beam width of the second detection mode is narrower, the detection range will not cover the wire poles on the side, thereby reducing the interference of the wire poles on the wires.

[0060] In the embodiments of the present specification, the execution subject of the radar-based target detection method involved in the present specification can be a UAV, or a server, a desktop computer or other terminal device. If a terminal device such as a server or a desktop computer is taken as the execution subject, the terminal device can obtain the point cloud data collected by the radar on the UAV, and determine the target detection result according to the point cloud data, and then return the determined target detection result to the UAV. For ease of description, the target detection method provided in the present specification will be described below with the UAV as the execution subject.

[0061] In the embodiments of the present specification, the UAV can obtain the point cloud data collected by the radar in the first detection mode as the first point cloud, and the point cloud data collected by the radar in the second detection mode as the second point cloud. The point cloud data mentioned here includes the position information of each point cloud point, the velocity information of each point cloud point, and the echo intensity of each point cloud point. The position information mentioned here can refer to the spatial rectangular coordinate system defined with the UAV as the coordinate origin, and the coordinates (X, Y, Z) of each point cloud point in the spatial rectangular coordinate system are determined. The velocity information mentioned here can refer to the Doppler velocity, that is, the relative motion velocity between the UAV and the point cloud point. The echo intensity mentioned here can refer to the intensity of the received electromagnetic wave from each point cloud point.

[0062] In actual applications, since the millimeter wave radar needs to detect obstacles, the intensity of the electromagnetic wave reflected from the obstacles in the detection range of the millimeter wave radar is relatively strong. Therefore, the UAV can screen out the point cloud points with relatively strong echo intensity.

[0063] In the embodiments of the present specification, the UAV can screen out, from the point cloud points of the first detection mode, the point cloud points with echo intensity greater than a set intensity threshold based on the point cloud data collected in the first detection mode, as the first point cloud.

[0064] Similarly, the UAV can screen out, from the point cloud points of the second detection mode, the point cloud points with echo intensity greater than a set intensity threshold based on the point cloud data collected in the second detection mode, as the second point cloud.

[0065] In the present specification, the UAV applying the control method of the UAV provided in the present specification can be used to perform delivery tasks in the delivery field, such as business scenarios of using the UAV for express delivery, logistics, take-out and the like.

[0066] S102: For each point cloud point included in the first point cloud and the second point cloud, according to the observation stability information generated by the detection model to which the point cloud point belongs when obtaining the point cloud point, determine the initial confidence of the point cloud point.

[0067] In actual application, if the first detection mode and the second detection mode simultaneously detect a point cloud point at the same position or a position close to the same position, it can be considered that the probability of existence of an obstacle at the position of the point cloud point is higher, and if only one of the first detection mode and the second detection mode detects the point cloud point, it can be considered that the probability of existence of an obstacle at the position of the point cloud point is lower.

[0068] In the embodiments of the present specification, the unmanned aerial vehicle can determine, for each point cloud point included in the first point cloud and the second point cloud, an initial confidence corresponding to the point cloud point according to observation stability information generated by a detection model to which the point cloud point belongs when the point cloud point is obtained.

[0069] In actual application, the unmanned aerial vehicle is usually in a motion state to detect an obstacle, and for the same obstacle, a millimeter wave radar installed on the unmanned aerial vehicle detects different positions of the obstacle. Since the intensities of electromagnetic waves reflected back by different positions of the obstacle are not the same, in a period of time, the unmanned aerial vehicle may not detect the obstacle for part of the time in the process of detecting the obstacle. Of course, the unmanned aerial vehicle may also detect a point cloud point at a position without an obstacle in the process of detecting the obstacle, for example, the intensity of electromagnetic waves reflected by a zebra crossing on the ground received by the unmanned aerial vehicle at a certain angle can be relatively high, and the unmanned aerial vehicle can detect the zebra crossing as an obstacle. Based on this, the unmanned aerial vehicle can determine an initial confidence corresponding to a point cloud point according to position information and speed information of the point cloud point at each time, that is, the probability of existence of an obstacle at the position of the point cloud point.

[0070] In the embodiments of the present specification, the unmanned aerial vehicle can determine, for each point cloud point included in the first point cloud and the second point cloud, an initial confidence corresponding to the point cloud point according to a continuous observation frame number and / or an observation continuous loss frame number corresponding to the point cloud point under a detection model to which the point cloud point belongs, wherein the higher the continuous observation frame number corresponding to the point cloud point under the detection model to which the point cloud point belongs, the higher the initial confidence corresponding to the point cloud point, and the higher the observation continuous loss frame number corresponding to the point cloud point under the detection model to which the point cloud point belongs, the lower the initial confidence corresponding to the point cloud point.

[0071] Specifically, for each point cloud point contained in the first point cloud and the second point cloud, the unmanned aerial vehicle can first determine the position information and the speed information of the point cloud point. Secondly, from the historical point cloud data collected at a plurality of time instants before the current time instant, each point cloud point in a set range at the position of the point cloud point in a set historical time period and having a relative motion speed with the unmanned aerial vehicle less than a set speed threshold is determined, and a time stamp corresponding to each point cloud point is determined. Then, according to the time stamp corresponding to each point cloud point, a time length during which the point cloud point is detected in the set historical time period and a continuous time length during which the point cloud point is not detected in the set historical time period are determined. Finally, according to the time length during which the point cloud point is detected in the set historical time period and the continuous time length during which the point cloud point is not detected in the set historical time period, an initial confidence degree corresponding to the point cloud point is determined.

[0072] wherein the longer the time length during which the point cloud point is detected in the set historical time period, the longer the time of stable tracking of the obstacle, and the higher the initial confidence degree corresponding to the point cloud point. The longer the continuous time length during which the point cloud point is not detected in the set historical time period, the longer the time of loss detection of the obstacle, and the lower the initial confidence degree corresponding to the point cloud point.

[0073] Further, the unmanned aerial vehicle determines a specific formula of the initial confidence degree corresponding to each point cloud point as follows:

[0074]

[0075] In the above formula, C0 is used to represent a bias compensation amount, and the range of the initial confidence degree can be changed according to actual requirements, is used to represent an activation function taking tick as a variable, tick is used to represent the time length during which the point cloud point is detected in the set time period, u and v are parameters set through expert experience, and determine the curve of the activation function. Generally, the change trend of the activation function with tick is that the activation function infinitely approaches 1 as tick increases, which is consistent with the change trend of the expected initial confidence degree with tick. For example, u=3 and v=0.8. -w×age is used to represent a negative linear function taking age as a variable. Age is used to represent the continuous time length during which the point cloud point is not detected, and w is a linear coefficient. The negative linear function reduces the initial confidence degree as age increases. As can be seen from the above formula, the longer the time length during which the point cloud point is detected in the set time period, the higher the initial confidence degree corresponding to the point cloud point, and the longer the continuous time length during which the point cloud point is not detected, the lower the initial confidence degree corresponding to the point cloud point.

[0076] S104: adjusting the initial confidence degree corresponding to the point cloud point according to the neighborhood matching result of the point cloud point in another point cloud other than the point cloud to which the point cloud point belongs, to obtain an adjusted confidence degree corresponding to the point cloud point.

[0077] In practical applications, for the same target, the first detection mode and the second detection mode have great differences in radar characteristics due to different application scenarios. Obviously, for the same obstacle, the detection results of the first detection mode and the second detection mode also have great differences. For example, for the same obstacle with a large volume, the detection ranges of the first detection mode and the second detection mode are different, and different positions of the obstacle can be detected. For an obstacle with a small volume, a detection mode with a higher range resolution can obtain better detection results. Based on this, the unmanned aerial vehicle can combine the data of the first detection mode and the data of the second detection mode to increase or decrease the initial confidence corresponding to each point cloud point, and obtain the adjusted confidence corresponding to each point cloud point.

[0078] In the embodiments of the present specification, the unmanned aerial vehicle can adjust the initial confidence corresponding to the point cloud point according to the neighborhood matching result of the point cloud point in another point cloud other than the point cloud to which the point cloud point belongs, and obtain the adjusted confidence corresponding to the point cloud point. The neighborhood matching result mentioned here can refer to a matching result that the distance between the point cloud point in the position of the point cloud to which the point cloud point belongs and the point cloud point existing in the corresponding position in another point cloud is less than a set threshold. For example, the point cloud point included in the second point cloud is located within a set range of the first point cloud in position. If there is a point cloud point within the set range of the first point cloud, the point cloud point included in the second point cloud and the point cloud point within the set range of the first point cloud are neighborhood matching results.

[0079] In practical applications, if there is an obstacle in front of the unmanned aerial vehicle, the first detection mode and the second detection mode should both obtain the point cloud point reflected by the obstacle. That is, if there is a point cloud point in the same position in the first detection mode and the second detection mode, it is considered that there is a higher possibility of an obstacle in front of the unmanned aerial vehicle. The unmanned aerial vehicle can increase the initial confidence corresponding to the point cloud point to obtain the adjusted confidence corresponding to the point cloud point, so as to increase the probability that the position of the point cloud point exists a real obstacle.

[0080] In the embodiments of the present specification, the unmanned aerial vehicle can increase the initial confidence corresponding to the point cloud point to obtain the adjusted confidence corresponding to the point cloud point according to the neighborhood matching result, if it is determined that there is a point cloud point in another point cloud whose position is less than a set distance from the position of the point cloud point.

[0081] In actual application, due to the existence and fluctuation of noise, a position actually without obstacle may be determined as having obstacle. In order to reduce the probability of false alarm, the unmanned aerial vehicle can determine a point cloud point with low initial confidence, and directly take the initial confidence corresponding to the point cloud point as the adjusted confidence corresponding to the point cloud point. In addition, the unmanned aerial vehicle can determine a point cloud point with high initial confidence, and improve the initial confidence corresponding to the point cloud point to obtain the adjusted confidence corresponding to the point cloud point.

[0082] In the embodiments of the present application, if the unmanned aerial vehicle determines, according to the neighborhood matching result, that there is a point cloud point in the other point cloud whose distance to the point cloud point is less than the set distance, the unmanned aerial vehicle determines whether the initial confidence corresponding to the point cloud point is greater than the first confidence threshold. If the unmanned aerial vehicle determines that the initial confidence corresponding to the point cloud point is greater than the first confidence threshold, the unmanned aerial vehicle improves the initial confidence corresponding to the point cloud point to obtain the adjusted confidence corresponding to the point cloud point. If it is determined that the initial confidence corresponding to the point cloud point is not greater than the first confidence threshold, the initial confidence corresponding to the point cloud point is taken as the adjusted confidence corresponding to the point cloud point. The first confidence threshold mentioned here can be determined according to human experience.

[0083] In actual application, if the number of point cloud points is large, the process of calculating the confidence compensation value corresponding to the point cloud point multiple times will consume unnecessary computing resources. Therefore, the unmanned aerial vehicle can set a maximum confidence to avoid wasting computing resources.

[0084] Further, if the unmanned aerial vehicle determines that the initial confidence corresponding to the point cloud point is greater than the first confidence threshold, the unmanned aerial vehicle improves the initial confidence corresponding to the point cloud point according to the preset confidence compensation value to obtain the compensated confidence corresponding to the point cloud point. The unmanned aerial vehicle determines whether the compensated confidence corresponding to the point cloud point exceeds the preset maximum confidence. If yes, the maximum confidence is taken as the adjusted confidence corresponding to the point cloud point, otherwise, the compensated confidence corresponding to the point cloud point is taken as the adjusted confidence corresponding to the point cloud point. The confidence compensation value mentioned here can be determined according to human experience.

[0085] In actual application, for each point cloud point included in the first point cloud and the second point cloud, there may be multiple point cloud points in the matching result in which the distance between the point cloud point in the position of the point cloud to which the point cloud point belongs and the point cloud point existing in the corresponding position in the other point cloud is less than the set threshold. The more the number of point cloud points, the greater the probability that there is an obstacle at the position. Therefore, the unmanned aerial vehicle can determine a confidence compensation value for each point cloud point by the number of point cloud points existing around the position of the point cloud point in the neighborhood matching result, so as to determine the probability that there is an obstacle at the position of the point cloud point.

[0086] In the embodiments of this specification, the UAV can determine the confidence compensation value corresponding to a point cloud point based on the number of point cloud points in the neighborhood matching results for each point cloud point contained in the first and second point clouds. Specifically, as follows... Figure 4A , Figure 4B As shown.

[0087] Figure 4A , Figure 4B This is a schematic diagram of the detection process of a millimeter-wave radar provided in the embodiments of this specification.

[0088] exist Figure 4A In this process, if the first detection mode is a long-range mode, then the second detection mode is a short-range mode. The number of point cloud points corresponding to point cloud point A in the first detection mode is determined. The UAV can determine the location information of point cloud point B in the second detection mode, and then determine whether the distance between the location information of point cloud point B in the second detection mode and the location information of point cloud point A in the first detection mode is less than a set distance threshold, thus obtaining the number of point cloud points corresponding to point cloud point A in the first detection mode. Then, based on the number of point cloud points corresponding to point cloud point A, the confidence compensation value corresponding to point cloud point A is determined. Finally, based on the confidence compensation value corresponding to point cloud point A, the adjusted confidence level corresponding to point cloud point A is determined.

[0089] Similarly, in Figure 4B In this process, if the first detection mode is a long-range mode, then the second detection mode is a short-range mode. The number of point cloud points corresponding to point cloud point B in the second detection mode is determined. The UAV can determine the location information of point cloud point A in the first detection mode, and then determine whether the distance between the location information of point cloud point A in the first detection mode and the location information of point cloud point B in the second detection mode is less than a set distance threshold, thus obtaining the number of point cloud points corresponding to point cloud point B in the second detection mode. Then, based on the number of point cloud points corresponding to point cloud point B, the confidence compensation value corresponding to point cloud point B is determined. Finally, based on the confidence compensation value corresponding to point cloud point B, the adjusted confidence level corresponding to point cloud point B is determined.

[0090] In practical applications, if an obstacle exists in front of the drone, both the first and second detection modes should acquire point cloud data reflecting the obstacle. In other words, if only one detection mode's point cloud data packet contains a point cloud data point at the same location, there is a higher probability of radar detection errors. The drone can reduce the initial confidence level corresponding to that point cloud data point to obtain an adjusted confidence level, thereby reducing the probability that a real obstacle exists at the location of that point cloud data point.

[0091] In the embodiments of the present disclosure, if the UAV determines, according to the neighborhood matching result, that there is no point cloud point in the other point cloud that is within a set distance from the point cloud point, the UAV reduces the initial confidence of the point cloud point to obtain the adjusted confidence of the point cloud point.

[0092] In actual applications, due to the ubiquitous existence of noise and fluctuations, a position where an actual obstacle exists may be determined as a position where no obstacle exists. In order to reduce the probability of missing detection, the UAV can determine a point cloud point with a high initial confidence, and directly use the initial confidence of the point cloud point as the adjusted confidence of the point cloud point. In addition, the UAV can determine a point cloud point with a low initial confidence, and reduce the initial confidence of the point cloud point to obtain the adjusted confidence of the point cloud point.

[0093] In the embodiments of the present disclosure, if the UAV determines, according to the neighborhood matching result, that there is no point cloud point in the other point cloud that is within a set distance from the point cloud point, the UAV determines whether the initial confidence of the point cloud point is less than a second confidence threshold. If the UAV determines that the initial confidence of the point cloud point is less than the second confidence threshold, the UAV reduces the initial confidence of the point cloud point to obtain the adjusted confidence of the point cloud point. If the UAV determines that the initial confidence of the point cloud point is not less than the second confidence threshold, the UAV uses the initial confidence of the point cloud point as the adjusted confidence of the point cloud point. The second confidence threshold mentioned herein can be determined according to human experience.

[0094] S106: determining a target point cloud point from the first point cloud and the second point cloud according to the adjusted confidence of each point cloud point included in the first point cloud and the second point cloud, and performing target detection according to the target point cloud point.

[0095] In the embodiments of the present disclosure, the UAV can determine a target point cloud point from the first point cloud and the second point cloud according to the adjusted confidence of each point cloud point included in the first point cloud and the second point cloud, and perform target detection according to the target point cloud point.

[0096] Specifically, the UAV can filter, from the first point cloud and the second point cloud, a point cloud point with an adjusted confidence greater than a set confidence threshold as a target point cloud point according to the adjusted confidence of each point cloud point included in the first point cloud and the second point cloud.

[0097] Further, the UAV can also be installed with a camera, a laser radar, and other sensors. The UAV can combine the determined target point cloud point with image data obtained by the camera and point cloud data obtained by the laser radar to determine a target detection result.

[0098] As can be seen from the above process, the method can detect the surrounding obstacles in a timely manner through the radar designed for the application scene of the unmanned aerial vehicle. The target point cloud points are determined from the first point cloud and the second point cloud according to the adjusted confidence corresponding to each point cloud point contained in the first point cloud and the second point cloud, and target detection is performed according to the target point cloud points. The method is installed with a radar meeting the target detection requirements of the unmanned aerial vehicle on the unmanned aerial vehicle, which can detect the surrounding obstacles in a timely manner, and effectively combines the point cloud data of the first detection mode and the point cloud data of the second detection mode, thereby improving the accuracy of the target detection result.

[0099] The above is a radar-based target detection method provided by one or more embodiments of the present specification. Based on the same idea, the present specification also provides a corresponding radar-based target detection device, as shown in Figure 5 .

[0100] Figure 5 A radar-based target detection device provided by the present specification is shown in the schematic diagram. The radar is provided with a first detection mode and a second detection mode, and the central angles or radii of the detection ranges of the first detection mode and the second detection mode are different, including:

[0101] The acquisition module 500 is configured to acquire point cloud data collected by the radar in the first detection mode as a first point cloud, and point cloud data collected by the radar in the second detection mode as a second point cloud.

[0102] The determination module 502 is configured to determine, for each point cloud point contained in the first point cloud and the second point cloud, an initial confidence corresponding to the point cloud point according to observation stability information generated by the detection model to which the point cloud point belongs when the point cloud point is obtained.

[0103] The adjustment module 504 is configured to adjust the initial confidence corresponding to the point cloud point according to the neighborhood matching result of the point cloud point in another point cloud other than the point cloud to which the point cloud point belongs, to obtain an adjusted confidence corresponding to the point cloud point.

[0104] The detection module 506 is configured to determine target point cloud points from the first point cloud and the second point cloud according to the adjusted confidence corresponding to each point cloud point contained in the first point cloud and the second point cloud, and perform target detection according to the target point cloud points.

[0105] Optionally, the observation stability information includes at least one of a continuous observation frame number corresponding to the point cloud point and an observation continuous loss frame number corresponding to the point cloud point.

[0106] The determination module 502 is specifically configured to determine, for each point cloud point in the first point cloud and the second point cloud, an initial confidence corresponding to the point cloud point according to a number of continuous observation frames corresponding to the point cloud point under a detection model to which the point cloud point belongs and / or a number of continuously lost observation frames corresponding to the point cloud point under the detection model to which the point cloud point belongs, wherein the higher the number of continuous observation frames corresponding to the point cloud point under the detection model to which the point cloud point belongs, the higher the initial confidence corresponding to the point cloud point, and the higher the number of continuously lost observation frames corresponding to the point cloud point under the detection model to which the point cloud point belongs, the lower the initial confidence corresponding to the point cloud point.

[0107] Optionally, the adjustment module 504 is specifically configured to, if it is determined according to the neighborhood matching result that there is a point cloud point in the other point cloud whose distance from the point cloud point is less than a set distance, increase the initial confidence corresponding to the point cloud point to obtain an adjusted confidence corresponding to the point cloud point.

[0108] Optionally, the adjustment module 504 is specifically configured to, if it is determined according to the neighborhood matching result that there is a point cloud point in the other point cloud whose distance from the point cloud point is less than a set distance, determine whether the initial confidence corresponding to the point cloud point is greater than a first confidence threshold, if it is determined that the initial confidence corresponding to the point cloud point is greater than the first confidence threshold, increase the initial confidence corresponding to the point cloud point to obtain an adjusted confidence corresponding to the point cloud point.

[0109] Optionally, the adjustment module 504 is specifically configured to, if it is determined that the initial confidence corresponding to the point cloud point is greater than the first confidence threshold, increase the initial confidence corresponding to the point cloud point by a preset confidence compensation value to obtain a compensated confidence corresponding to the point cloud point, determine whether the compensated confidence corresponding to the point cloud point exceeds a preset maximum confidence, if yes, take the maximum confidence as the adjusted confidence corresponding to the point cloud point, and if no, take the compensated confidence corresponding to the point cloud point as the adjusted confidence corresponding to the point cloud point.

[0110] Optionally, the adjustment module 504 is specifically configured to, if it is determined according to the neighborhood matching result that there is no point cloud point in the other point cloud whose distance from the point cloud point is less than a set distance, decrease the initial confidence corresponding to the point cloud point to obtain an adjusted confidence corresponding to the point cloud point.

[0111] Optionally, the adjustment module 504 is specifically configured to, if it is determined according to the neighborhood matching result that there is no point cloud point in the other point cloud whose distance from the point cloud point is less than a set distance, determine whether the initial confidence corresponding to the point cloud point is less than a second confidence threshold, if it is determined that the initial confidence corresponding to the point cloud point is less than the second confidence threshold, decrease the initial confidence corresponding to the point cloud point to obtain an adjusted confidence corresponding to the point cloud point.

[0112] Optionally, a central angle of the detection range of the first detection mode is not less than a central angle of the detection range of the second detection mode, a radius of the detection range of the first detection mode is not less than a radius of the detection range of the second detection mode, or a central angle of the detection range of the second detection mode is not less than a central angle of the detection range of the first detection mode, a radius of the detection range of the second detection mode is not less than a radius of the detection range of the first detection mode.

[0113] The specification also provides a computer readable storage medium storing a computer program, the computer program being used to execute the above Figure 1 The specification provides a radar-based target detection method.

[0114] The specification also provides a computer readable storage medium storing a computer program, the computer program being used to execute the above Figure 6 The specification provides a radar-based target detection method. Figure 1 The specification provides a radar-based target detection method. Figure 6 As described above, at the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and of course, can also include other hardware required by the business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs to implement the above Figure 1 The specification provides a radar-based target detection method.

[0115] In the 1990s, it was relatively easy to distinguish whether an improvement in a technology was a hardware improvement (e.g., an improvement in the circuit structure of a diode, transistor, switch, etc.) or a software improvement (an improvement in a method flow). However, as technology has evolved, many improvements in method flows today can be considered as direct improvements in hardware circuit structures. Designers almost always obtain the corresponding hardware circuit structures by programming the improved method flows into hardware circuits. Therefore, it cannot be said that an improvement in a method flow cannot be implemented using hardware entity modules. For example, a programmable logic device (PLD) (e.g., a field programmable gate array (FPGA)) is an integrated circuit whose logic function is determined by user programming of the device. A digital system is "integrated" on a PLD by the designer programming the PLD, rather than by ordering a chip manufacturer to design and fabricate a custom integrated circuit chip. Moreover, instead of manually fabricating integrated circuit chips, this programming is now mostly implemented using "logic compiler" software, which is similar to software compilers used in program development, and the original code to be compiled is written in a specific programming language, which is called a hardware description language (HDL), and there are many such languages, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc., and the most commonly used are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should be aware that, as long as the method flow is logically programmed in the above-mentioned hardware description languages and programmed into an integrated circuit, a hardware circuit implementing the logical method flow can be easily obtained.

[0116] The controller can be implemented in any suitable way, for example, the controller can take the form of a microprocessor or processor and a computer readable medium storing computer readable program code, such as software or firmware, executable by the (micro)processor, logic gates, switches, an application specific integrated circuit (ASIC), a programmable logic controller and an embedded microcontroller, examples of which include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20 and Silicone Labs C8051F320, the memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also know that, in addition to being implemented in pure computer readable program code, the controller can equally well be implemented to perform the same functions using logic gates, switches, an application specific integrated circuit, a programmable logic controller and an embedded microcontroller, etc. by means of a logical programming of the method steps. The controller can thus be considered as a hardware component, and the means comprised therein for performing the various functions can be considered as structures within the hardware component. Alternatively, the means for performing the various functions can even be considered as both a software module implementing the method and a structure within the hardware component.

[0117] The systems, apparatuses, modules or units illustrated by the above embodiments can be specifically implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0118] For the sake of description, the above apparatuses are described in functional division and are described respectively as various units. Of course, the functions of the units can be implemented in the same or multiple software and / or hardware in the implementation of the specification.

[0119] Those skilled in the art will understand that the embodiments of the specification can be provided as a method, a system or a computer program product. Therefore, the specification can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the specification can take the form of a computer program product implemented on one or more computer usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0120] The specification is presented with reference to flow diagrams and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the specification. It will be understood that each block of the flow diagrams and / or block diagrams, and combinations of blocks in the flow diagrams and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processing element or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flow diagrams and / or block diagrams block or blocks. Figure 1 The flow diagrams and / or block diagrams in the specification can present a method, apparatus or computer program product according to embodiments of the specification. Flow diagrams and / or block diagrams can also present a method, apparatus or computer program product to achieve functions specified in flow diagrams and / or block diagrams block or blocks. Figure 1 The flow diagrams and / or block diagrams in the specification can present a method, apparatus or computer program product according to embodiments of the specification. Flow diagrams and / or block diagrams can also present a method, apparatus or computer program product to achieve functions specified in flow diagrams and / or block diagrams block or blocks.

[0121] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the flow diagrams and / or block diagrams block or blocks. Figure 1 The flow diagrams and / or block diagrams in the specification can present a method, apparatus or computer program product according to embodiments of the specification. Flow diagrams and / or block diagrams can also present a method, apparatus or computer program product to achieve functions specified in flow diagrams and / or block diagrams block or blocks. Figure 1 The flow diagrams and / or block diagrams in the specification can present a method, apparatus or computer program product according to embodiments of the specification. Flow diagrams and / or block diagrams can also present a method, apparatus or computer program product to achieve functions specified in flow diagrams and / or block diagrams block or blocks.

[0122] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flow diagrams and / or block diagrams block or blocks. Figure 1 The flow diagrams and / or block diagrams in the specification can present a method, apparatus or computer program product according to embodiments of the specification. Flow diagrams and / or block diagrams can also present a method, apparatus or computer program product to achieve functions specified in flow diagrams and / or block diagrams block or blocks. ​ The flow diagrams and / or block diagrams in the specification can present a method, apparatus or computer program product according to embodiments of the specification. Flow diagrams and / or block diagrams can also present a method, apparatus or computer program product to achieve functions specified in flow diagrams and / or block diagrams block or blocks.

[0123] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0124] The memory can include non-persistent memory and / or persistent memory, such as flash memory, read-only memory (ROM), and / or volatile / non-volatile random access memory (RAM), among others. The memory is an example of computer-readable media.

[0125] Computer-readable media includes permanent and non-permanent, movable and non-movable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible to a computing device. According to the definition herein, computer-readable media does not include transitory media such as modulated data signals and carriers.

[0126] It should also be noted that the terms "comprising", "containing", or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, method, article or apparatus that comprises a list of elements does not only include those elements, but can also include other elements not expressly listed or inherent to such process, method, article or apparatus. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or apparatus that includes the element.

[0127] Those skilled in the art will appreciate that embodiments of the present specification can be provided as methods, systems or computer program products. Therefore, the present specification can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Moreover, the present specification can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0128] The present specification can be described in the general context of computer-executable instructions, such as program modules, executed by computers. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The present specification can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in both local and remote computer storage media including storage devices.

[0129] The various embodiments described in this specification are described using a numbering of embodiments approach: these are each individually integrated contributions pertaining to different but related aspects of the description. Each of the various embodiments can stand on its own, and each can be combined with the subject matter of other embodiments to produce further embodiments. Where appropriate, therefore, the contents of the specification can be regarded as being incorporated by reference, including the description, drawings, claims, abstract and the like.

[0130] The above description is embodied in the form of only a description of embodiments of the present specification, and is not intended to limit the present specification. Various modifications and changes can be made by those skilled in the art based on the present specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present specification should be included in the scope of the claims of the present specification.

Claims

1. A radar-based target detection method, characterized by, The radar is provided with a first detection mode and a second detection mode, and the method comprises: acquiring point cloud data collected by the radar in the first detection mode as a first point cloud and point cloud data collected by the radar in the second detection mode as a second point cloud; for each point cloud point contained in the first point cloud and the second point cloud, determining an initial confidence degree corresponding to the point cloud point according to observation stability information generated by a detection model to which the point cloud point belongs when the point cloud point is obtained; adjusting the initial confidence degree corresponding to the point cloud point according to a neighborhood matching result of the point cloud point in another point cloud other than the point cloud to which the point cloud point belongs, to obtain an adjusted confidence degree corresponding to the point cloud point; determining a target point cloud point from the first point cloud and the second point cloud according to the adjusted confidence degree corresponding to each point cloud point contained in the first point cloud and the second point cloud, and performing target detection according to the target point cloud point; wherein a central angle of a detection range of the first detection mode is not less than a central angle of a detection range of the second detection mode, and a radius of the detection range of the first detection mode is not less than a radius of the detection range of the second detection mode; or a central angle of a detection range of the second detection mode is not less than a central angle of a detection range of the first detection mode, and a radius of the detection range of the second detection mode is not less than a radius of the detection range of the first detection mode.

2. The method of claim 1, wherein, The observation stability information comprises at least one of a continuous observation frame number corresponding to the point cloud point and an observation continuous loss frame number corresponding to the point cloud point; for each point cloud point contained in the first point cloud and the second point cloud, determining an initial confidence degree corresponding to the point cloud point according to observation stability information generated by a detection model to which the point cloud point belongs when the point cloud point is obtained, specifically comprising: for each point cloud point contained in the first point cloud and the second point cloud, determining an initial confidence degree corresponding to the point cloud point according to a continuous observation frame number and / or an observation continuous loss frame number corresponding to the point cloud point in the detection model to which the point cloud point belongs, wherein the higher the continuous observation frame number corresponding to the point cloud point in the detection model to which the point cloud point belongs, the higher the initial confidence degree corresponding to the point cloud point, and the higher the observation continuous loss frame number corresponding to the point cloud point in the detection model to which the point cloud point belongs, the lower the initial confidence degree corresponding to the point cloud point.

3. The method of claim 1, wherein, adjusting the initial confidence degree corresponding to the point cloud point according to a neighborhood matching result of the point cloud point in another point cloud other than the point cloud to which the point cloud point belongs, to obtain an adjusted confidence degree corresponding to the point cloud point, specifically comprising: if it is determined according to the neighborhood matching result that there is a point cloud point in the other point cloud whose position is less than a set distance from the position of the point cloud point, increasing the initial confidence degree corresponding to the point cloud point to obtain the adjusted confidence degree corresponding to the point cloud point.

4. The method of claim 3, wherein, if it is determined according to the neighborhood matching result that there is a point cloud point in the other point cloud whose position is less than a set distance from the position of the point cloud point, increasing the initial confidence degree corresponding to the point cloud point to obtain the adjusted confidence degree corresponding to the point cloud point, specifically comprising: if it is determined, according to the neighborhood matching result, that there is a point cloud point in the other point cloud whose position is less than a set distance from the point cloud point, determining whether the initial confidence corresponding to the point cloud point is greater than a first confidence threshold; if it is determined that the initial confidence corresponding to the point cloud point is greater than the first confidence threshold, increasing the initial confidence corresponding to the point cloud point to obtain an adjusted confidence corresponding to the point cloud point.

5. The method of claim 4, wherein, if it is determined that the initial confidence corresponding to the point cloud point is greater than the first confidence threshold, increasing the initial confidence corresponding to the point cloud point to obtain an adjusted confidence corresponding to the point cloud point, specifically including: if it is determined that the initial confidence corresponding to the point cloud point is greater than the first confidence threshold, increasing the initial confidence corresponding to the point cloud point to obtain an adjusted confidence corresponding to the point cloud point, specifically including: if it is determined that the initial confidence corresponding to the point cloud point is greater than the first confidence threshold, increasing the initial confidence corresponding to the point cloud point to obtain an adjusted confidence corresponding to the point cloud point, specifically including: if it is determined that the initial confidence corresponding to the point cloud point is greater than the first confidence threshold, increasing the initial confidence corresponding to the point cloud point to obtain an adjusted confidence corresponding to the point cloud point, specifically including:

6. The method of claim 3, wherein, if it is determined, according to the neighborhood matching result, that there is no point cloud point in the other point cloud whose position is less than a set distance from the point cloud point, decreasing the initial confidence corresponding to the point cloud point to obtain an adjusted confidence corresponding to the point cloud point. if it is determined, according to the neighborhood matching result, that there is no point cloud point in the other point cloud whose position is less than a set distance from the point cloud point, decreasing the initial confidence corresponding to the point cloud point to obtain an adjusted confidence corresponding to the point cloud point, specifically including:

7. The method of claim 6, wherein, if it is determined, according to the neighborhood matching result, that there is no point cloud point in the other point cloud whose position is less than a set distance from the point cloud point, determining whether the initial confidence corresponding to the point cloud point is less than a second confidence threshold; if it is determined that the initial confidence corresponding to the point cloud point is less than the second confidence threshold, decreasing the initial confidence corresponding to the point cloud point to obtain an adjusted confidence corresponding to the point cloud point. The storage medium stores a computer program, and the computer program is executed by the processor to implement the method in any one of claims 1-7.

8. A computer-readable storage medium, characterized in that, The processor executes the program to implement the method in any one of claims 1-7.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, ​

Images