Object detection device and object detection method

Abstract

To provide an object detection device and an object detection method which can prevent erroneous detection even when a road has slopes or undulations.SOLUTION: An object detection device 10 includes: an input part 11a in which point group information is input from a LiDAR 2 installed on the ground; and a setting part 11c for setting such a road surface definition sd that the point group information on a road part among the point group information input in the input part 11a is defined as a road surface. The setting part 11c changes the road surface definition sd in such a manner that a differential value between information h indicating the height of each point included in the point group information on the road part and the road surface definition sd is, for example, a first threshold or more and a second threshold or less.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to an object detection device and an object detection method for detecting an object on, for example, a road. [Background technology]

[0002] For example, as an object detection device that detects objects on a road, a method is known in which an object is detected using clusters generated by clustering a point cloud detected by LiDAR (Light Detection And Ranging).

[0003] For example, Patent Document 1 describes a method for detecting objects in the correct units with high accuracy by dividing an over-connected cluster if it is determined that the cluster is over-connected based on image targets detected from an ambient light image and clusters obtained by clustering a point cloud, and by combining two or more clusters that exist in a part of the point cloud that corresponds to the image targets if it is determined that the cluster is over-divided. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2021-131385 A Summary of the Invention [Problem to be solved by the invention]

[0005] In the above-mentioned clustering, the road surface (pavement surface) is excluded in advance as a background region, and only objects on the road are clustered. In this case, on the premise that the road surface is flat, a process is performed in which points that are less than a height of several tens of centimeters from the road surface are collectively excluded as points originating from the road.

[0006] Such a road surface definition was basically set to a flat surface based on a specific reference position. However, a road may not only have flat parts, but may have gradients or undulations. In such a case, if the reference position is set at the highest point of the gradient, an object to be detected, such as a person, located at a low gradient may be erroneously detected as a road.

[0007] SUMMARY OF THE PRESENT EMBODIMENTS Accordingly, an object of the present invention is to provide an object detection device and an object detection method that can reduce the occurrence of erroneous detection even when the road has gradients or undulations. [Means for solving the problem]

[0008] The invention described in claim 1, which has been made to solve the above problem, is an object detection device comprising: an input unit to which point cloud information is input from a sensor that is installed on the ground, intermittently irradiates light to scan a specified space, receives reflected light of the light, and acquires point cloud information based on the reflected light; and a setting unit that sets surface height information defined as a road surface position for a road portion of the specified space, wherein the setting unit sets the length from information indicating the height of a point included in the point cloud information of the road portion to the surface height information to be greater than a first threshold and less than a second threshold. Effect of the Invention

[0009] According to the present invention, the setting unit sets the length from the information indicating the height of the points included in the point cloud information of the road portion to the surface height information to be equal to or greater than the first threshold value and equal to or less than the second threshold value, so that in the case of a slope such as a gradient, the height information can be set in a stepped manner. Therefore, even if the road has a gradient or undulations, it is possible to reduce the occurrence of false detection. [Brief description of the drawings]

[0010] [Figure 1] 1 is an example of installation of an object detection device according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a block diagram of the object detection device shown in FIG. [Diagram 3]3 is a flowchart of an operation for defining a road surface in the object detection device shown in FIG. 2. [Figure 4] FIG. 13 is an explanatory diagram of the operation of defining a road surface. [Diagram 5] FIG. 13 is an explanatory diagram of the operation of defining a road surface. [Figure 6] 4 is an example of a road surface definition set in the flowchart shown in FIG. 3. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, one embodiment of the present invention will be described with reference to Fig. 1 to Fig. 6. Fig. 1 shows an example of installation of an object detection device according to one embodiment of the present invention.

[0012] The object detection device 10 shown in Fig. 1 is installed on a pillar P fixed near an intersection, for example. The object detection device 10 detects moving objects such as pedestrians and vehicles passing through the intersection. That is, the object detection device 10 is installed on the ground. Installed on the ground here means that it is fixed to the ground directly or indirectly, that is, it means that it is installed on an object that is fixedly installed on the ground, such as a building, rather than on a moving object such as a vehicle.

[0013] The object detection device 10 includes a main body 1 and a LiDAR 2. The main body 1 is provided as a control box or the like at the bottom of a support P. As shown in the block diagram of FIG. 2, the main body 1 includes a control unit 11 and a communication unit 12.

[0014] The control unit 11 is configured with, for example, a computer, etc. The control unit 11 is responsible for the overall control of the object detection device 10. In Fig. 1, the control unit 11 functionally includes an input unit 11a, a detection unit 11b, and a setting unit 11c.

[0015] The input unit 11a receives point cloud information acquired by the LiDAR 2. The detection unit 11b detects objects by excluding points whose height information is equal to or less than a road surface definition (to be described later) for each point included in the point cloud information input to the input unit 11a. The detection unit 11b performs, for example, a clustering process on the point cloud after excluding the points whose height information is equal to or less than the road surface definition, and determines the type of moving body, such as a pedestrian or a vehicle, from the shape, size, etc. of the cluster. That is, the detection unit 11b executes a detection process.

[0016] The setting unit 11c sets the road surface definition, which is defined as the road surface, for the point cloud information of the road portion among the point cloud information input to the input unit 11a. The road surface definition is set above the road surface as shown in Figs. 4 to 6, which will be described later. The setting of the road surface definition in the setting unit 11c will be described later.

[0017] The communication unit 12 outputs information about the object detected by the control unit 11 to the outside. The communication unit 12 may be configured as, for example, a wireless communication device (or a wired communication device). Information about the object may include the type, moving direction, and moving speed of the object. Specific examples of output destinations include moving bodies such as vehicles passing near the pillar P, and traffic control centers.

[0018] The LiDAR2 is a sensor that intermittently irradiates a laser beam to scan a predetermined space, receives reflected light of the laser beam, and acquires point cloud information (three-dimensional point cloud information) based on the reflected light. The LiDAR2 discretely measures the distance to an object present in the space of the irradiation range by the reflected light of the irradiated laser beam, and recognizes the position and shape of the object as a three-dimensional point cloud. In this embodiment, the LiDAR2 is installed on the top of the support P as shown in FIG. 1, and scans the road surface R (the road surface R and the space above it) as the predetermined space.

[0019] Next, the operation of defining a road surface definition in the object detection device 10 configured as described above will be described with reference to the flowchart in FIG. 3. First, the input unit 11a acquires observation points (step S1). That is, point cloud information obtained by scanning the illumination range of the LiDAR 2 is acquired. Hereinafter, each point included in the point cloud is referred to as an observation point as necessary. Note that the point cloud information acquired in step S1 may be used by accumulating observation points for multiple scans. However, since the point cloud information has a large amount of data, a period of about several seconds is appropriate.

[0020] Next, the setting unit 11c extracts observation points originating from the road (step S2). That is, from the point cloud information acquired in step S1, the setting unit 11c extracts observation points obtained by reflection from the road surface. A specific method includes, for example, a method of performing coordinate matching between the point cloud information acquired from the LiDAR 2 and a road map, and extracting only point clouds whose latitude, longitude, and altitude are considered to be on the road. In this case, the road map (map information) may be stored in a storage device or the like provided in the control unit 11 or the main body unit 1.

[0021] Alternatively, a method may be used in which point cloud information observed within a certain time period and camera images acquired at the same time are compared, and the area in which point clouds of objects passing on the road are present is determined as a road, and then point clouds below a certain height are extracted as originating from roads from among the point clouds observed within that area.

[0022] Next, the setting unit 11c extracts representative points for the road-derived observation points extracted in step S2 (step S3). Even if the point cloud obtained from the LiDAR2 is limited to those derived from roads, a very large number of points can be obtained. Therefore, if processing is performed for all observation points, there is a concern that the processing time will be enormous. Therefore, in this embodiment, processing is performed only for representative observation points among the road-derived point cloud. As a method for extracting the representative point, there is a method in which the depth direction in the irradiation direction of the LiDAR2 is divided into sections of a predetermined length, and the point with the largest height coordinate among the observation points observed within each section is set as the representative point.

[0023] In the above-mentioned method, if there are many observation points that are considered to be noise, outliers may be removed using statistics of height coordinates within a block, for example, by assuming that the distribution of height coordinates is normal and determining the difference from the average value using an unbiased standard deviation. Also, the intervals between blocks may be equal or variable according to the Euclidean distance from the LiDAR2 (sensor head) to the road surface with the intention of aligning the number of observation points within a block.

[0024] In addition, at the same time as searching for the representative point, a search can be conducted for an observation point that is not noise and originates from the road, with the longest distance in the depth direction, and the distance can be set as the maximum observation value of the road surface. The area beyond the maximum observation value of the road surface is treated as an area where no valid observation points originating from the road can be obtained.

[0025] Next, the setting unit 11c sets an initial value of the height definition (step S4). The initial value of the height definition is an initial value for the subsequent setting operation of the road surface definition. This initial value is a value added so as to be relatively high with respect to the height coordinates of the representative point whose distance in the depth direction is closest to 0 m (the position closest to the LiDAR 2 from which the point cloud is obtained). For example, the relationship between the first threshold value and the second threshold value described later can be set as first threshold value<second threshold value, so that first threshold value≦initial value≦second threshold value.

[0026] Also, the user or the like may set the initial value of the height to any value within the range that satisfies the above conditions. For example, in the case of an upward gradient from the vicinity of the LiDAR 2 to the distance, it is preferable to set a value close to the second threshold, and in the case of a downward gradient from the vicinity of the LiDAR 2 to the distance, it is preferable to set a value close to the first threshold. Also, if the user or the like does not set the initial value, for example, the intermediate value between the first threshold and the second threshold ((first threshold + second threshold) / 2) may be used.

[0027] The operations of the next steps S5 to S10 are repeated from a distance in the depth direction of 0 m to the maximum observation value of the road surface. In step S5, the setting unit 11c determines whether the observation point of interest (representative point) has reached the maximum observation value of the road surface in order to execute the processing of step S7 and subsequent steps.

[0028] If the maximum observation value of the road surface is reached (step S5: YES), a new section is created and the height is set to invalid (step S6). Here, the section in steps S5 to S10 is different from the section divided to extract the representative point. Moreover, the invalid height means that since the maximum observation value of the road surface is reached, the depth beyond this point is an area in which valid observation points originating from the road cannot be obtained, and the area is excluded from the process of excluding observation points below the road surface definition. That is, the setting unit 11c does not set the road surface definition (height information) beyond a predetermined distance in the depth direction. Note that the height indicated by the road surface definition does not indicate the height from the road surface R, but indicates the height coordinate value or altitude within a predetermined space scanned by the LiDAR2.

[0029] If the maximum observed value of the road surface has not been reached (step S5: NO), the setting unit 11c judges whether the difference between the information indicating the height of the representative point and the height of the road surface definition (height definition) is smaller than the first threshold (step S7). If the difference between the information indicating the height of the representative point and the height of the road surface definition is smaller than the first threshold (step S7: YES), the setting unit 11c creates a new section and changes the road surface definition by the second threshold from the currently focused representative point (step S8). The first threshold and the second threshold are thresholds set to judge whether or not to change the road surface definition, and as described above, the relationship of the first threshold<the second threshold is satisfied. That is, the length from the representative point (information indicating the height of the point included in the point cloud information of the road portion) to the road surface definition (surface height information) is set to be equal to or larger than the first threshold and equal to or smaller than the second threshold.

[0030] The first threshold is a value that determines the degree of response to small undulations on the road and measurement errors in the height direction of LiDAR2, and is generally several centimeters to several tens of centimeters. The second threshold is an allowable difference in height between the road surface definition and the actual road surface. This second threshold is set to a small value relative to the total height of the detection target, and is approximately several tens of centimeters when the detection target is a pedestrian or vehicle on the road. In other words, both the first threshold and the second threshold are relative values, rather than values ​​that indicate absolute heights such as height coordinates.

[0031] Here, steps S7 and S8 will be described with reference to Fig. 4. Fig. 4 shows an inclined surface and road surface definitions sd1 and sd2 provided corresponding to the inclined surface. In Fig. 4, black circles displayed on the flat surface and the inclined surface indicate representative points rp, and white circles indicate non-representative points p. In Fig. 4, as indicated by the arrow labeled "scanning," LiDAR2 is on the left side of the paper, and the right side of the paper is the far side as viewed from LiDAR2.

[0032] In FIG. 4, the difference value h between the value (height) of the road surface definition sd1 and the value indicating the height of the representative point rp and the threshold value h TH1 (first threshold value). Then, the difference value h is compared with the threshold value h TH1 If it is smaller than (h <h TH1 ), the threshold value h TH2 The road surface definition is set so that the road surface is raised by (second threshold value). The signpost surface definition after the setting is indicated by the symbol sd2. In other words, the difference value h is the length from the representative point rp to the road surface definition sd1. Also, creating a new section in step S8 means that the road surface definition is changed from the road surface definition sd1 to the road surface definition sd2. Also, the threshold value h TH1 < Threshold h TH2 , the road surface definition is stepped up.

[0033] That is, when the difference value h is less than the first threshold, the setting unit 11c changes the road surface definition (surface height information) by the second threshold.

[0034] Returning to the description of Fig. 3, if the difference in height between the representative point and the road surface definition is equal to or greater than the first threshold (step S7: NO), the setting unit 11c judges whether the difference in height between the representative point and the road surface definition is greater than the second threshold (step S9). If the difference in height between the representative point and the road surface definition is greater than the second threshold (step S9: YES), the setting unit 11c creates a new section and changes the road surface definition by the first threshold from the representative point currently being considered (step S10). Steps S9 and S10 will now be described with reference to Fig. 5.

[0035] FIG. 5 is basically the same as FIG. 4, except that the uphill slope seen from the LiDAR 2 in FIG. 4 has become a downhill slope. In FIG. 5, the difference value h between the value (height) of the road surface definition sd3 and the value indicating the height of the representative point rp, and the threshold value h TH2 Then, the difference value h is compared with the threshold value h TH2 If it exceeds (h>h TH2 ), the threshold value h TH1 The road surface definition is set so that the road surface is raised by sd4. The signpost surface definition after the setting is indicated by the reference symbol sd4. In other words, the difference value h is the length from the representative point rp to the road surface definition sd3. Also, creating a new section in step S10 means that the road surface definition is changed from the road surface definition sd3 to the road surface definition sd4. Also, the threshold value h TH1 < Threshold h TH2 , the road surface definition is stepped down.

[0036] That is, when the difference value h exceeds the second threshold, the setting unit 11c changes the road surface definition (surface height information) by the first threshold.

[0037] Returning to the explanation of FIG. 3, if the difference in height between the representative point and the road surface definition is equal to or less than the second threshold (step S9: NO), steps S5 to S10 are executed for the next representative point. If the execution is performed for all representative points and the maximum observed value of the road surface is reached, the loop is exited and the flowchart is terminated. The road surface definition generated by executing this flowchart is set in the setting unit 11c and is used for the process of excluding points below the surface height information in the detection unit 11b.

[0038] According to the operation of the above flowchart, the setting unit 11c identifies a road portion from the point cloud information input to the input unit 11a, and sets the length from the height coordinate of the representative point included in the point cloud information of the identified road portion to the road surface definition (surface height information) to be greater than or equal to the first threshold and less than or equal to the second threshold.

[0039] FIG. 6 shows an example of the road surface definition sd obtained according to the above-mentioned flow chart. FIG. 6 shows an overall image of the uphill slope shown in FIG. 4. That is, the road surface R is composed of a flat part Lo at the bottom of the slope, a flat part Hi at the top of the slope, and an inclined surface Sl connecting the flat parts Lo and Hi. The black circle indicates an observation point (representative point rp). The road surface definition sd is set at a position higher than the road surface R. An observation point with a height equal to or lower than this road surface definition sd is excluded from object detection as an observation point originating from the road (shaded part). The symbol M in FIG. 6 is a person (pedestrian) as an example of an object to be detected.

[0040] Conventionally, for example, the road surface definition sd was set planarly based on the flat area Hi, so that the road surface definition at the height of the flat area Hi was applied as is to the flat area Lo. Therefore, the point cloud of the lower body of the person M on the flat area Lo is excluded as a road. Therefore, the value of the road surface definition sd was excessive on the flat area Lo and the inclined surface Sl. In this case, if the person M is located close to the LiDAR2, a sufficient point cloud can be obtained with only the upper body, and the object can be detected as the person M. However, since the LiDAR2 emits laser light radially, the density of the point cloud decreases when the person moves away from the LiDAR2, and there are cases where a sufficient point cloud cannot be obtained with only the upper body.

[0041] Therefore, as shown in Fig. 6, when this embodiment is applied, the road surface definition sd is automatically set in a stepped shape on a slope. Also, the height of the road surface definition sd differs between the flat portion Hi and the flat portion Lo. Therefore, it is possible to reduce the area that is not subject to object detection and obtain enough observation points to identify an object as, for example, a person M.

[0042] 3, the height of each section divided along the depth direction is different. Therefore, the point cloud information input from the input unit 11a is compared with the road surface definition sd after it is determined which section the point cloud information corresponds to.

[0043] According to the above-mentioned flow chart, step S1 functions as an input step, and steps S2 to S10 function as a setting step.

[0044] According to this embodiment, the object detection device 10 includes an input unit 11a to which point cloud information is input from a LiDAR 2 installed on the ground, a setting unit 11c that sets a road surface definition sd that is defined as a road surface for the point cloud information of the road part among the point cloud information input to the input unit 11a, and a detection unit 11b that detects an object by excluding points whose height information h is equal to or less than the road surface definition sd from the point cloud information input to the input unit 11a. The setting unit 11c changes the road surface definition sd so that the difference value between the information h indicating the height of each point included in the point cloud information of the road part and the road surface definition sd is, for example, equal to or more than a first threshold and equal to or less than a second threshold.

[0045] By configuring the object detection device 10 as described above, the setting unit 11c changes the road surface definition sd so that the difference value between the information h indicating the height of the points included in the point cloud information of the road portion and the road surface definition sd is equal to or greater than the first threshold and equal to or less than the second threshold, so that in the case of a slope, the road surface definition sd can be changed in a stepped manner. Therefore, even if the road has a gradient or undulations, erroneous detection of a detection target such as a person as a road is unlikely to occur. In addition, since the road surface definition sd is automatically generated from the point cloud, it can be handled even if the road surface R has a complex shape.

[0046] In addition, when the difference value between the information h indicating the height of the points included in the point cloud information of the road portion and the road surface definition sd is less than the first threshold, the setting unit 11c changes the height information h by the second threshold. In this way, when the road is an uphill slope as seen from the LiDAR 2, it is possible to keep the difference value within the range of the first threshold or more and the second threshold or less.

[0047] In addition, when the difference value between the information h indicating the height of the point included in the point cloud information of the road portion and the road surface definition sd exceeds the second threshold, the setting unit 11c changes the height information h by the first threshold. By doing so, it is possible to keep the difference value within the range of the first threshold or more and the second threshold or less when the road is a downhill slope as seen from the LiDAR 2.

[0048] In addition, the setting unit 11c divides the depth direction of a predetermined space into sections of a predetermined length for the point cloud information of the road portion, extracts a representative point for each section, and changes the height information h so that the difference value between the representative point and the road surface definition sd is equal to or greater than the first threshold and equal to or less than the second threshold. By doing so, it is possible to shorten the processing time since it is not necessary to process all points.

[0049] In addition, the setting unit 11c does not set height information as invalid beyond a predetermined distance in the depth direction. By doing so, for example, for a range that exceeds the maximum observation value of the road surface, height information is not set, and unnecessary processing can be avoided.

[0050] The present invention is not limited to the above-described embodiment. In other words, a person skilled in the art can implement the present invention by modifying it in various ways in accordance with conventionally known knowledge without departing from the gist of the present invention. As long as the object detection device and object detection method of the present invention are still included in the scope of the present invention even after such modification, the object detection device and object detection method of the present invention are of course included in the scope of the present invention. [Explanation of symbols]

[0051] 1 Main unit 2 LiDAR (sensor) 10 Object detection device 11 Control section 11a Input section 11b Detection unit 11c Setting section

Claims

1. an input unit that receives the point cloud information from a sensor that is installed on the ground, intermittently irradiates light to scan a predetermined space, receives reflected light of the light, and acquires point cloud information based on the reflected light; a setting unit that sets surface height information defined as a road surface position for a road portion in the predetermined space; Equipped with the setting unit sets a length from information indicating heights of points included in the point cloud information of the road portion to the surface height information to be equal to or greater than a first threshold and equal to or less than a second threshold.

1. An object detection device comprising:

2. The object detection device according to claim 1 , wherein the setting unit changes the surface height information by the second threshold value when the length is less than the first threshold value.

3. The object detection device according to claim 1 , wherein the setting unit changes the surface height information by the first threshold value when the length exceeds the second threshold value.

4. The object detection device according to claim 1, characterized in that the setting unit divides the point cloud information of the road portion into sections of a predetermined length in the depth direction of the specified space, extracts a representative point for each section, and sets the height information so that the length from information indicating the height of the representative point to the surface height information is greater than or equal to the first threshold and less than or equal to the second threshold.

5. The object detection device according to claim 4 , wherein the setting unit does not set the height information beyond a predetermined distance in the depth direction.

6. 2. The object detection device according to claim 1, wherein the setting unit has map information, and extracts the point cloud information of the road portion by coordinate matching between the map information and the point cloud information.

7. The object detection device according to any one of claims 1 to 6, further comprising a detection unit that detects an object by excluding points whose information indicating the height of each point included in the point cloud information input to the input unit is equal to or lower than the surface height information.

8. An object detection method implemented by an object detection device including a sensor that is installed on the ground, intermittently irradiates light to scan a predetermined space, receives reflected light of the light, and acquires point cloud information based on the reflected light, an input step of inputting the point cloud information from the sensor; a setting step of setting surface height information defined as a road surface position for a road portion in the predetermined space; Including, the setting step sets a length from information indicating heights of points included in the point cloud information of the road portion to the surface height information to be equal to or greater than a first threshold and equal to or less than a second threshold.

13. An object detection method comprising:

Images