Information processing device, control method, program, and storage medium

Abstract

To provide an information processing device, a control method, and a program which can suitably detect vehicles ahead, and a storage medium in which the program is stored.SOLUTION: A control unit 7 of a lidar 100 includes: a point cloud information generation block 72 that functions as acquisition means; and a point cloud information processing block 73 that functions as reflection member point detection means, surrounding point detection means, and forward vehicle detection means. The point cloud information generation block 72 acquires point cloud information obtained by receiving the reflected light of the light emitted in a prescribed direction. The point cloud information processing block 73 detects, from a measured point cloud indicated by the point cloud information, a reflector point that represents a plurality of reflectors provided on the outside of a vehicle. The point cloud information processing block 73 also detects reflector surrounding points existing in the surrounding of the reflector point. Meanwhile, the point cloud information processing block 73 detects a forward vehicle segment formed by the reflector point and the reflector surrounding points, as a vehicle existing in the prescribed direction.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a technique for detecting a forward vehicle. [Background technology]

[0002] A distance measuring device is known that irradiates a measurement object with light, detects the reflected light from the measurement object, and calculates the distance to the measurement object based on the time difference between the timing of irradiating the measurement object with light and the timing of detecting the reflected light from the measurement object. Patent Document 1 discloses a forward vehicle recognition device that changes a lighting pattern for projecting a light projection pattern based on a detection state of the light projection pattern, and detects the distance and inclination of the forward vehicle. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2008-082750 A Summary of the Invention [Problem to be solved by the invention]

[0004] Generally, when detecting a forward vehicle based on a distance measuring device using a laser beam such as a lidar, it is difficult to detect a distant forward vehicle because the number of points measured is small. Similarly, when a forward vehicle has a color with low reflectance such as black, the number of points measured is small, making it difficult to detect.

[0005] The present invention aims to provide an information processing device, a control method, a program, and a storage medium storing the program, which are capable of suitably detecting a vehicle ahead. [Means for solving the problem]

[0006] The claimed invention is an acquisition means for acquiring point cloud information obtained by receiving reflected light of light emitted in a predetermined direction; a reflecting member point detection means for detecting reflecting member points representing a plurality of reflecting members provided outside the vehicle from the measurement point cloud indicated by the point cloud information; a surrounding point detection means for detecting surrounding points existing around the reflecting member point; a vehicle detection means for detecting a segment in the measurement point cloud formed by the reflecting member points and the surrounding points as the vehicle present in the predetermined direction; Have death, the reflecting member point detection means detects a pair of measurement points having a reflection intensity greater than a first threshold value and a predetermined number of measurement points having a reflection intensity less than a second threshold value in the measurement point cloud as a measurement point sequence that exists continuously in a horizontal direction with respect to the ground, The reflecting member point detection means determines the predetermined number based on a distance in a depth direction indicated by the sequence of measurement points. It is an information processing device. The claimed invention also includes: an acquisition means for acquiring point cloud information obtained by receiving reflected light of light emitted in a predetermined direction; a reflecting member point detection means for detecting reflecting member points representing a plurality of reflecting members provided outside the vehicle from the measurement point cloud indicated by the point cloud information; a surrounding point detection means for detecting surrounding points existing around the reflecting member point; a vehicle detection means for detecting a segment in the measurement point cloud formed by the reflecting member points and the surrounding points as the vehicle present in the predetermined direction; having the reflecting member point detection means detects a pair of measurement points as the reflecting member point when a pair of measurement points having a reflection intensity greater than other measurement points by a predetermined ratio or more and a predetermined number of measurement points between the pair having a reflection intensity smaller than the reflection intensity of the pair by a predetermined ratio or more exist in the measurement point cloud as a measurement point sequence that exists continuously in a horizontal direction with respect to the ground, The reflecting member point detection means determines the predetermined number based on a distance in a depth direction indicated by the sequence of measurement points. It is an information processing device.

[0007] The claimed invention also includes: A method executed by an information processing device, comprising: an acquisition step of acquiring point cloud information obtained by receiving reflected light of light emitted in a predetermined direction; a reflective member point detection step of detecting reflective member points representing a plurality of reflective members provided outside the vehicle from the measurement point cloud indicated by the point cloud information; a surrounding point detection step of detecting surrounding points existing around the reflecting member point; a vehicle detection step of detecting a segment in the measurement point cloud formed by the reflective member points and the surrounding points as the vehicle present in the predetermined direction; having The reflecting member point detection step detects a pair of measurement points having a reflection intensity greater than a first threshold value and a predetermined number of measurement points having a reflection intensity less than a second threshold value between the pair of measurement points as a series of measurement points that are continuously present in a horizontal direction relative to the ground in the measurement point cloud, and detects the pair as the reflecting member point; The reflecting member point detection step is a method for determining the predetermined number based on a distance in the depth direction indicated by the sequence of measurement points. The invention described in the claims is a method executed by an information processing device, an acquisition step of acquiring point cloud information obtained by receiving reflected light of light emitted in a predetermined direction; a reflective member point detection step of detecting reflective member points representing a plurality of reflective members provided outside the vehicle from the measurement point cloud indicated by the point cloud information; a surrounding point detection step of detecting surrounding points existing around the reflecting member point; a vehicle detection step of detecting a segment in the measurement point cloud formed by the reflective member points and the surrounding points as the vehicle present in the predetermined direction; having The reflecting member point detection step detects a pair of measurement points as the reflecting member point when a pair of measurement points having a reflection intensity greater than other measurement points by a predetermined ratio or more and a predetermined number of measurement points between the pair having a reflection intensity smaller than the reflection intensity of the pair by a predetermined ratio or more exist in the measurement point cloud as a measurement point sequence that exists continuously in a horizontal direction with respect to the ground, The reflecting member point detection step is a method for determining the predetermined number based on a distance in the depth direction indicated by the sequence of measurement points.

[0008] The claimed invention also includes: an acquisition means for acquiring point cloud information obtained by receiving reflected light of light emitted in a predetermined direction; a reflecting member point detection means for detecting reflecting member points representing a plurality of reflecting members provided outside the vehicle from the measurement point cloud indicated by the point cloud information; a surrounding point detection means for detecting surrounding points existing around the reflecting member point; A computer is caused to function as a vehicle detection means for detecting a segment in the measurement point cloud formed by the reflective member points and the surrounding points as the vehicle existing in the predetermined direction. 、 the reflecting member point detection means detects a pair of measurement points having a reflection intensity greater than a first threshold value and a predetermined number of measurement points having a reflection intensity less than a second threshold value in the measurement point cloud as a measurement point sequence that exists continuously in a horizontal direction with respect to the ground, The reflecting member point detection means determines the predetermined number based on a distance in a depth direction indicated by the sequence of measurement points. It is a program. The claimed invention also includes: an acquisition means for acquiring point cloud information obtained by receiving reflected light of light emitted in a predetermined direction; a reflecting member point detection means for detecting reflecting member points representing a plurality of reflecting members provided outside the vehicle from the measurement point cloud indicated by the point cloud information; a surrounding point detection means for detecting surrounding points existing around the reflecting member point; a computer that functions as a vehicle detection unit that detects a segment in the measurement point cloud formed by the reflective member points and the surrounding points as the vehicle present in the predetermined direction; the reflecting member point detection means detects a pair of measurement points as the reflecting member point when a pair of measurement points having a reflection intensity greater than other measurement points by a predetermined ratio or more and a predetermined number of measurement points between the pair having a reflection intensity smaller than the reflection intensity of the pair by a predetermined ratio or more exist in the measurement point cloud as a measurement point sequence that exists continuously in a horizontal direction with respect to the ground, The reflecting member point detection means is a program that determines the predetermined number based on the distance in the depth direction indicated by the sequence of measurement points. [Brief description of the drawings]

[0009] [Figure 1] 2 shows a schematic configuration of a lidar common to the first and second embodiments. [Diagram 2] 5 is an example of a flowchart showing a procedure of a forward vehicle detection process in the first embodiment. [Diagram 3] 13 is an example of a flowchart illustrating details of a reflector point detection process. [Figure 4] (A) to (D) show examples of measurement point sequences that satisfy the long-distance detection condition regarding reflection intensity. [Diagram 5] 4 shows a sequence of measurement points when the long-distance non-detection condition is satisfied. [Figure 6] (A) to (K) show examples of measurement point sequences that satisfy the short-distance detection condition regarding reflection intensity. [Figure 7] (A) is a diagram showing a point of interest in the current frame and its surrounding measurement points on a virtual scanning plane, and (B) is a diagram showing a point of interest in the previous frame and its surrounding measurement points on a virtual scanning plane. [Figure 8] 11 is an example of a flowchart illustrating a procedure for a surrounding point detection process. [Figure 9] 13 is an example of a flowchart illustrating a procedure of a segmentation process. [Figure 10]13 is an example of a flowchart illustrating a procedure of a segmentation process. [Figure 11] 13 is an example of a flowchart illustrating a procedure of an object identification process in a second embodiment. [Figure 12] 13 is an example of a flowchart showing a procedure of a forward vehicle detection process in a second embodiment. [Figure 13] 13 is a flowchart illustrating an example of a procedure for a segment integration process. [Figure 14] 13 is a flowchart illustrating an example of a post-processing procedure. [Figure 15] FIG. 2 shows a configuration diagram of a lidar system according to a first modified example. [Figure 16] FIG. 1 illustrates an example of a fixed lidar detecting a vehicle. [Figure 17] 13 is an overhead view of the periphery of a vehicle equipped with a lidar in a third modified example. [Figure 18] An example of a measurement point sequence that is considered to satisfy the long-distance detection condition in the fourth modified example will be shown. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] In a preferred embodiment of the present invention, the information processing device includes an acquisition means for acquiring point cloud information obtained by receiving reflected light of light emitted in a predetermined direction, a reflective member point detection means for detecting reflective member points representing a plurality of reflective members provided outside the vehicle from the measurement point cloud indicated by the point cloud information, a surrounding point detection means for detecting surrounding points existing around the reflective member points, and a vehicle detection means for detecting a segment in the measurement point cloud formed by the reflective member points and the surrounding points as the vehicle existing in the predetermined direction. In this way, the information processing device can preferably detect the vehicle by detecting the reflective members provided on the vehicle.

[0011] In one aspect of the information processing device, when a pair of measurement points whose reflection intensity is greater than a first threshold value and a predetermined number of measurement points whose reflection intensity is less than a second threshold value between the pair are present in the measurement point cloud as a measurement point sequence that exists continuously in a horizontal direction with respect to the ground, the reflective member point detection means detects the pair as the reflective member point. With this aspect, the information processing device can preferably detect the reflective member points that are the measurement points of the reflective members that are provided as pairs.

[0012] In another aspect of the information processing device, when a pair of measurement points whose reflection intensity is greater than that of other measurement points by a predetermined ratio or more and a predetermined number of measurement points between the pair whose reflection intensity is smaller than that of the pair by a predetermined ratio or more exist in the measurement point cloud as a measurement point sequence that exists continuously in a horizontal direction with respect to the ground, the reflective member point detection means detects the pair as the reflective member point. Even with this aspect, the information processing device can preferably detect the reflective member points that are the measurement points of the reflective members provided in pairs.

[0013] In another aspect of the information processing device, the reflecting member point detection means determines the predetermined number based on the depth direction distance indicated by the measurement point sequence. With this aspect, the information processing device can appropriately determine a predetermined number suitable for detecting reflecting member points. In a preferred example, the reflecting member point detection means may set a different range of the predetermined number between a case where the depth direction distance indicated by the measurement point sequence is longer than a first distance and a case where the distance indicated by the measurement point sequence is equal to or shorter than the first distance.

[0014] In another aspect of the information processing device, the reflective member point detection means determines whether the pair is the reflective member point based on at least one of the width of the pair, the distance difference in the depth direction of the pair, or the height of the pair from the ground. With this aspect, the information processing device can detect the reflective member point with higher accuracy by taking into account the characteristics of the installation position of the reflective member of the vehicle.

[0015] In another aspect of the information processing device, when a measurement point whose depth direction distance is equal to or greater than a second distance and whose reflection intensity is greater than a first threshold value is continuous horizontally with respect to the ground, the reflective material point detection means excludes the measurement point from candidates for the reflective material point. With this aspect, the information processing device can preferably prevent erroneous detection of a sign or the like with a high reflectivity as a reflective material.

[0016] In another aspect of the information processing device, the reflective member point detection means detects reflective member points at the current processing time based on the detection result of the vehicle at the previous processing time. With this aspect, the information processing device can preferably detect reflective member points based on the detection result at the previous processing time and preferably prevent reflective member points from being overlooked.

[0017] In another aspect of the information processing device, the reflective member point detection means detects, as the reflective member point, a measurement point where a difference in distance in the depth direction measured between the preprocessing time and the current processing time is within a predetermined distance difference and where the measurement point corresponds to a scanning position detected as the vehicle at the preprocessing time. With this aspect, the information processing device can accurately detect the reflective member point based on the detection result at the preprocessing time.

[0018] In another aspect of the information processing device, the acquiring means acquires, as the point cloud information, flags indicating whether each of the measurement points is a valid point that does not correspond to noise, and the reflecting member point detecting means excludes the measurement points that are flagged as not being valid points from the candidates for the reflecting member points. With this aspect, the information processing device can suitably suppress the detection of noise as a reflecting member point.

[0019] In another aspect of the information processing device, the surrounding point detection means detects, as the surrounding points, measurement points that exist around the reflecting member point and whose distance difference in the depth direction from the reflecting member point is within a predetermined distance difference. With this aspect, the information processing device can suitably detect, as the surrounding points, surrounding measurement points that are estimated to be the same object as the reflecting member point.

[0020] In another aspect of the information processing device, the vehicle detection means regards the measurement points within a rectangular parallelepiped formed based on the reflective member points and the surrounding points as part of the segment. With this aspect, the information processing device can suitably detect the measurement point cloud representing the entire vehicle as a segment.

[0021] In another aspect of the information processing device, the plurality of reflective members are provided on the front, side, or rear of the exterior of the vehicle. The information processing device can suitably detect a vehicle having a reflective member on the front, side, or rear.

[0022] In another preferred embodiment of the present invention, a control method executed by an information processing device includes an acquisition step of acquiring point cloud information obtained by receiving reflected light of light emitted in a direction including the front of the vehicle, a reflecting member point detection step of detecting reflecting member points representing a plurality of reflecting members from a measurement point cloud indicated by the point cloud information, a surrounding point detection step of detecting surrounding points existing around the reflecting member points, and a leading vehicle detection step of detecting a segment in the measurement point cloud formed by the reflecting member points and the surrounding points as a leading vehicle preceding the vehicle. By executing this control method, the information processing device can preferably detect a leading vehicle.

[0023] In another preferred embodiment of the present invention, the program causes a computer to function as an acquisition means for acquiring point cloud information obtained by receiving reflected light of light emitted in a direction including the front of the vehicle, a reflection member point detection means for detecting reflection member points representing a plurality of reflection members from a measurement point cloud indicated by the point cloud information, a surrounding point detection means for detecting surrounding points existing around the reflection member points, and a leading vehicle detection means for detecting a segment in the measurement point cloud formed by the reflection member points and the surrounding points as a leading vehicle preceding the vehicle. By executing this program, the computer can preferably detect a leading vehicle. Preferably, the program is stored in a storage medium. EXAMPLES

[0024] Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings.

[0025] <First Example> (1) Device configuration FIG. 1 shows a schematic configuration of a lidar 100 according to this embodiment. The lidar 100 is mounted on a vehicle that performs driving assistance such as autonomous driving. The lidar 100 irradiates a laser beam (also called "irradiated beam") in a predetermined angular range in the horizontal and vertical directions, and receives light (also called "reflected beam") that is reflected by an object and returns, thereby discretely measuring the distance from the lidar 100 to the object and generating point cloud information indicating the three-dimensional position of the object. The lidar 100 is installed so that the measurement range includes at least the forward direction of the vehicle on which it is mounted.

[0026] As shown in FIG. 1, the lidar 100 mainly includes a transmitter 1, a receiver 2, a beam splitter 3, a scanner 5, a piezoelectric sensor 6, a controller 7, and a memory 8.

[0027] The transmitter 1 is a light source that emits pulsed irradiation light toward the beam splitter 3. The transmitter 1 includes, for example, an infrared laser light emitting element. The transmitter 1 is driven based on a drive signal “Sg1” supplied from the controller 7.

[0028] The receiver 2 is, for example, an avalanche photodiode, generates a detection signal “Sg2” corresponding to the amount of received light, and supplies the generated detection signal Sg2 to the controller 7.

[0029] The beam splitter 3 transmits the pulsed irradiation light emitted from the transmission unit 1. In addition, the beam splitter 3 reflects the light reflected by the scanner 5 towards the reception unit 2.

[0030] The scanner 5 is, for example, an electrostatically driven mirror (MEMS mirror), and the inclination (i.e., the angle of optical scanning) changes within a predetermined range based on the drive signal "Sg3" supplied from the control unit 7. The scanner 5 reflects the illumination light that has passed through the beam splitter 3 toward the outside of the LIDAR 100, and also reflects the reflected light that is incident from the outside of the LIDAR 100 toward the beam splitter 3. In addition, a point within the measurement range of the LIDAR 100 at which the illumination light is irradiated is also called a "measurement point."

[0031] The scanner 5 is also provided with a piezoelectric sensor 6. The piezoelectric sensor 6 detects distortion caused by stress of a torsion bar that supports the mirror portion of the scanner 5. The piezoelectric sensor 6 supplies the generated detection signal “Sg4” to the control unit 7. The detection signal Sg4 is used to detect the orientation of the scanner 5.

[0032] The memory 8 is composed of various volatile and non-volatile memories such as a RAM (Random Access Memory), a ROM (Read Only Memory), and a flash memory. The memory 8 stores a program required for the control unit 7 to execute a predetermined process. The memory 8 also stores various parameters referenced by the control unit 7. The memory 8 also stores point cloud information for the latest predetermined number of frames generated by the control unit 7.

[0033] The control unit 7 includes various processors such as a central processing unit (CPU) and a graphics processing unit (GPU). The control unit 7 executes a program stored in the memory 8 to perform a predetermined process. The control unit 7 is an example of a computer that executes a program. The control unit 7 is not limited to being realized by software according to a program, but may be realized by any combination of hardware, firmware, and software. The control unit 7 may be a user-programmable integrated circuit such as a field-programmable gate array (FPGA) or a microcontroller, or may be an application specific standard produce (ASSP), application specific integrated circuit (ASIC), or the like.

[0034] The control unit 7 functionally includes a transmission driving block 70, a scanner driving block 71, a point cloud information generation block 72, and a point cloud information processing block 73.

[0035] The transmission drive block 70 outputs a drive signal Sg1 that drives the transmission unit 1. The drive signal Sg1 includes information for controlling the emission time of a laser light-emitting element included in the transmission unit 1 and the emission intensity of the laser light-emitting element. The transmission drive block 70 controls the emission intensity of the laser light-emitting element included in the transmission unit 1 based on the drive signal Sg1.

[0036] The scanner driving block 71 outputs a driving signal Sg3 for driving the scanner 5. This driving signal Sg3 includes a horizontal driving signal corresponding to the resonance frequency of the scanner 5 and a vertical driving signal for vertical scanning. In addition, the scanner driving block 71 detects the scanning angle of the scanner 5 (i.e., the emission direction of the irradiation light) by monitoring the detection signal Sg4 output from the piezo sensor 6.

[0037] Based on the detection signal Sg2 supplied from the receiving unit 2, the point cloud information generating block 72 generates point cloud information indicating the distance and direction to the object irradiated with the irradiation light for each measurement point, with the lidar 100 as the reference point. In this case, the point cloud information generating block 72 calculates the time from when the irradiation light is emitted until the receiving unit 2 detects the reflected light as the time of flight of light. Then, the point cloud information generating block 72 generates point cloud information indicating the distance according to the calculated time of flight and the irradiation direction of the irradiation light corresponding to the reflected light received by the receiving unit 2 for each measurement point, and supplies the generated point cloud information to the point cloud information processing block 73. Hereinafter, the point cloud information obtained by one scan of all the measurement points is referred to as one frame of point cloud information. The point cloud information generating block 72 is an example of an "acquisition means".

[0038] Here, the point cloud information can be regarded as an image in which the measurement points are pixels and the distance indicated by each measurement point is the pixel value. In this case, the irradiation direction of the irradiation light at the elevation and depression angles of the vertically arranged measurement points is different, and the irradiation direction of the irradiation light at the horizontal angle of the horizontally arranged measurement points is different. Hereinafter, an XYZ coordinate system is defined with the horizontal direction (i.e., horizontal direction) of the image as the "Y coordinate axis," the vertical direction (i.e., height direction) of the image as the "Z coordinate axis," and the distance indicated by the pixel value (i.e., the distance from the LIDAR 100 in the depth direction) as the "X coordinate axis." Hereinafter, the XYZ coordinate system is considered to be a three-dimensional coordinate system with the position of the LIDAR 100 as the origin. In addition, the point cloud information includes information on the received light intensity (i.e., the received light intensity at each measurement point) for each pixel.

[0039] The point cloud information generation block 72 may generate point cloud information from which data generated by erroneously detecting an object in the point cloud information (so-called false alarm data) is suitably removed. Hereinafter, measurement points corresponding to data generated by detecting an actual object are also called "valid points". In this case, the point cloud information further includes flag information indicating whether each measurement point is a valid point or not.

[0040] The point cloud information processing block 73 detects a group of measurement points (also called "forward vehicle segments") constituting a forward vehicle preceding the vehicle on which the lidar 100 is mounted, based on the point cloud information supplied from the point cloud information generation block 72. In this embodiment, the point cloud information processing block 73 detects forward vehicle segments based on the detection results of measurement points (also called "reflector points") representing reflectors provided in pairs on the left and right rear of the vehicle. In this case, the point cloud information processing block 73 adds flag information indicating whether or not each measurement point is a forward vehicle segment to the point cloud information. In addition, the point cloud information processing block 73 stores the point cloud information for each frame in the memory 8 in association with time information indicating the processing time for each frame. Details of the processing of the point cloud information processing block 73 will be described later. The point cloud information processing block 73 is an example of a "reflective member point detection means", a "surrounding point detection means", and a "forward vehicle detection means". In addition, a reflector is an example of a "reflective member", and a reflector point is an example of a "reflective member point".

[0041] The point cloud information stored in the memory 8 may be output to, for example, a device that controls driving assistance such as automatic driving of a vehicle (also called a "driving assistance device"). In this case, the driving assistance device may be, for example, an ECU (Electronic Control Unit) of the vehicle, or an in-vehicle device such as a car navigation device electrically connected to the vehicle.

[0042] (2) Overview of forward vehicle detection processing Next, an overview of the forward vehicle detection process executed by the control unit 7 will be described.

[0043] 2 is a flowchart showing an example of a procedure for a forward vehicle detection process. The point cloud information processing block 73 repeatedly executes the process shown in FIG.

[0044] First, the point cloud information generation block 72 generates point cloud information based on the detection signal Sg2 (step S11). In this case, the point cloud information generation block 72 generates point cloud information of a frame corresponding to the current processing time (also referred to as a "current frame") based on the detection signal Sg2 generated by one scan of the scanning target range of the LIDAR 100.

[0045] Next, the point cloud information processing block 73 executes a reflector point detection process (step S12) which is a process of detecting measurement points corresponding to reflector points from the measurement point cloud indicated by the point cloud information based on the point cloud information generated in step S11. A detailed process flow of the reflector point detection process will be described later with reference to FIG. 3. If the point cloud information further includes flag information indicating whether each measurement point is a valid point or not, the point cloud information processing block 73 detects only valid points as reflector points. In other words, the point cloud information processing block 73 excludes measurement points with a flag indicating that they are not valid points from candidates for reflector points. This effectively prevents noise points from being erroneously detected as reflector points.

[0046] Next, the point cloud information processing block 73 executes a surrounding point detection process (step S13) for detecting measurement points (also called "reflector surrounding points") that exist around the reflector points detected in step S12 and represent the same object as the reflector points (i.e., the forward vehicle). A detailed processing flow of the surrounding point detection process will be described later with reference to FIG. 8.

[0047] Next, the point cloud information processing block 73 executes a segmentation process (step S14) for generating a forward vehicle segment that is a group of measurement points constituting the forward vehicle, based on the reflector points detected in step S12 and the reflector surrounding points detected in step S13. A detailed processing flow of the segmentation process will be described later with reference to Figures 9 and 10. Then, the control unit 7 repeatedly executes the process of the flowchart shown in Figure 2 described above in accordance with a scanning period (i.e., a frame generation period).

[0048] 2, the control unit 7 can preferably detect a forward vehicle even if the number of measurement points for a distant vehicle traveling ahead is small. Also, because the forward vehicle is a color with low reflectance such as black, the forward vehicle can be preferably detected even if the number of measurement points for the forward vehicle is small.

[0049] (3) Reflector point detection process Next, the details of the reflector point detection process will be described. In summary, the point cloud information processing block 73 regards a pair of measurement points with high reflection intensity that satisfies either a long-distance detection condition, which is a reflector detection condition assuming that a forward vehicle exists at a long distance, or a short-distance detection condition, which is a reflector detection condition assuming that a forward vehicle exists at a short distance, as a reflector point. In addition, the point cloud information processing block 73 prevents reflection point detection failure by detecting reflection points based on the reflection point detection result in a frame corresponding to the immediately previous processing time (also called the "previous frame"). The reflection point detection condition based on the reflection point detection result in the previous frame is also called the "past detection condition".

[0050] (3-1) Processing flow Fig. 3 is an example of a flowchart showing details of the reflector point detection process executed by the point cloud information processing block 73 in step S12 in Fig. 2. In this flowchart, the point cloud information processing block 73 provides a flag (also called "forward vehicle flag Fv") indicating whether or not each measurement point indicated by the point cloud information of the current frame is a candidate point of a forward vehicle, and sets the forward vehicle flag Fv of a measurement point regarded as a reflector point to "valid" (here, "1"). On the other hand, the point cloud information processing block 73 sets the forward vehicle flag Fv of a measurement point (also called "non-reflector point") regarded as not a reflector point to "invalid" (here, "0").

[0051] First, the point cloud information processing block 73 sets the forward vehicle flags Fv of all measurement points indicated by the point cloud information corresponding to the frame at the current processing time to 0 (that is, invalid) (step S101).

[0052] Next, loop A consisting of steps S102 to S110 is executed for each of the measurement points for which the forward vehicle flag Fv has been set to 0. Hereinafter, the measurement point to be processed in each loop will also be referred to as a "point of interest."

[0053] First, the point cloud information processing block 73 determines whether a pair of the attention point and at least one measurement point existing on the same scanning line satisfies a long distance detection condition (step S102). In this case, the point cloud information processing block 73 determines whether the pair including the attention point satisfies the conditions regarding reflection intensity, X coordinate, Y coordinate, and Z coordinate defined as the long distance detection condition. Specific examples of the long distance detection condition will be described later with reference to Figs. 4(A) to (D) and the like.

[0054] Then, when the point cloud information processing block 73 determines that the pair including the attention point satisfies the long distance detection condition (step S102; Yes), it regards the pair including the attention point as a reflector point and sets the forward vehicle flag Fv of the pair to "1" (step S103). Note that when noise removal processing is performed, the point cloud information processing block 73 sets the forward vehicle flag Fv of the attention point Pi in the current frame to 1 only when the attention point Pi in the current frame is a valid point. The same applies to the following steps S107 and S110.

[0055] On the other hand, when it is determined that there is no pair including the attention point that satisfies the long distance detection condition (step S102; No), the point cloud information processing block 73 determines whether or not a pair of the attention point and at least one measurement point existing on the same scanning line satisfies the long distance non-detection condition (step S104). Here, the long distance non-detection condition is a condition provided to prevent erroneous detection of a measurement point of a highly reflective member (e.g., a sign, etc.) other than a reflector of a forward vehicle as a reflector point, and details will be described with reference to FIG. 5.

[0056] Then, if the point cloud information processing block 73 determines that there is a pair including a point of interest that satisfies the long-distance non-detection condition (step S104; Yes), it regards the pair including the point of interest as a non-reflector point and sets the forward vehicle flag Fv of these measurement points to "0" (step S105).

[0057] On the other hand, when the point cloud information processing block 73 determines that there is no pair including the attention point that satisfies the long-distance non-detection condition (step S104; No), it determines whether or not a pair of the attention point and at least one measurement point existing on the same scanning line satisfies the short-distance detection condition (step S106). In this case, the point cloud information processing block 73 determines whether or not the pair including the attention point satisfies the conditions regarding reflection intensity, X coordinate, Y coordinate, and Z coordinate defined as the short-distance detection condition. Specific examples of the short-distance detection condition will be described with reference to Figs. 6(A) to (K) etc.

[0058] Then, if the point cloud information processing block 73 determines that there is a pair including the attention point that satisfies the short-distance detection condition (step S106; Yes), it regards the pair including the attention point as a reflector point and sets the forward vehicle flag Fv of the pair to "1" (step S107).

[0059] On the other hand, if the point cloud information processing block 73 determines that there is no pair including a point of interest that satisfies the short-distance detection condition (step S106; No), it determines whether the point of interest satisfies the past detection condition (step S108). In this case, the point cloud information processing block 73 refers to the point cloud information of the previous frame stored in the memory 8 by the point cloud information processing block 73, and determines whether the point of interest satisfies the condition that allows it to be regarded as a reflector point. Details of the past detection condition will be described with reference to Figures 7(A) and (B).

[0060] If the point cloud information processing block 73 determines that the attention point satisfies the past detection condition (step S109; Yes), it further determines whether the surrounding points of the attention point satisfy the past detection condition (step S109). In this case, the point cloud information processing block 73 regards as surrounding points the measurement points whose vertical and horizontal positional differences from the attention point are within a predetermined number of pixels when the point cloud information is regarded as an image, and determines whether the surrounding points satisfy the past detection condition. If the point cloud information processing block 73 determines that the surrounding points satisfy the past detection condition (step S109; Yes), it regards the attention point and the surrounding points as reflector points, and sets the forward vehicle flags Fv of these measurement points to 1 (step S110).

[0061] On the other hand, if the focus point does not satisfy the past detection condition (step S108; No), or if the surrounding points of the focus point do not satisfy the past detection condition (step S109; No), the point cloud information processing block 73 ends the processing of loop A for the current focus point without executing step S110, while keeping the forward vehicle flag Fv of the focus point at 0.

[0062] (3-2) Long-distance detection conditions The point cloud information processing block 73 determines whether or not the long distance detection conditions related to the reflection intensity, the X coordinate, the Y coordinate, and the Z coordinate are satisfied.

[0063] First, the conditions related to reflection intensity will be described. In this case, the point cloud information processing block 73 judges whether the reflection intensity of the target point is greater than a first threshold value and whether the measurement points (also called "measurement point sequence") that are consecutively present on the same scanning line as the target point satisfy a predetermined condition.

[0064] 4(A) to (D) show examples of measurement point sequences that satisfy the long-distance detection condition regarding reflection intensity. Here, measurement points whose reflection intensity is greater than a first threshold (also called "high reflection measurement points") are indicated by hatched circles, and measurement points whose reflection intensity is less than a second threshold (also called "low reflection measurement points") are indicated by unhatched circles. The horizontal lines indicate scanning lines. The first threshold is set to a value greater than the second threshold, for example, the first threshold is set to a value two to four times the second threshold.

[0065] As shown in Fig. 4(A) to (D), a measurement point sequence that satisfies a long-distance detection condition for reflection intensity includes a pair of high reflection measurement points including a target point, and one to four low reflection measurement points between the pair. The long-distance detection condition is a reflector point detection condition that assumes that a forward vehicle is present at a long distance, and therefore the range of the number of low reflection measurement points present between the pair of high reflection measurement points is set to a smaller range than the short-distance detection condition described below. The range of the number of low reflection measurement points is determined in advance based on, for example, experiments, and is stored in memory 8. The number of low reflection measurement points that should be present between the pair of high reflection measurement points to satisfy the long-distance detection condition is an example of a "predetermined number".

[0066] Next, the condition related to the X coordinate for satisfying the long distance detection condition will be described. The point cloud information processing block 73 determines that the distance in the depth direction indicated by the measurement point sequence of the target (i.e., the X coordinate value) is longer than a first distance (e.g., 30 m) as a necessary condition for satisfying the long distance detection condition. The above-mentioned first distance is a threshold for determining whether the measurement point sequence of the target is a point sequence of an object existing at a long distance, and is stored in the memory 8 in advance. This first distance may be different for each type of the lidar 100. Note that the point cloud information processing block 73 may regard the distance indicated by the attention point as the distance in the depth direction indicated by the measurement point sequence, or may regard a representative value such as the average value of the distances indicated by each of the measurement points constituting the measurement point sequence as the distance in the depth direction indicated by the measurement point sequence.

[0067] Next, the condition related to the Y coordinate for satisfying the long-distance detection condition will be described. Considering that the interval between reflectors is equal to or less than the width of the vehicle, the point cloud information processing block 73 determines that the width in the Y coordinate direction indicated by the measurement point sequence is less than a predetermined distance (e.g., 2.5 m) as a necessary condition for satisfying the long-distance detection condition. The above-mentioned predetermined distance is set in advance in consideration of the expected vehicle width.

[0068] Next, the conditions related to the Z coordinate for satisfying the long-distance detection condition will be described. Considering that the reflector is located at a position lower than the vehicle height, the point cloud information processing block 73 determines that the height indicated by the measurement point sequence based on the ground is within a predetermined distance (e.g., 1.5 m) as a necessary condition for satisfying the long-distance detection condition. The above-mentioned predetermined distance is set in advance taking into consideration the expected vehicle height. In this case, the position information of the ground may be estimated based on point cloud information obtained by measuring the ground, or may be obtained based on information other than the point cloud information (e.g., map information, etc.).

[0069] Preferably, the point cloud information processing block 73 may further define, as a necessary condition for satisfying the long distance detection condition, that the depth difference (i.e., the difference in X coordinate value) between a pair of highly reflective measurement points regarded as reflector points is within a predetermined distance (e.g., 1 m). Information on threshold values ​​(predetermined distance, etc.) required for determining the various conditions described above is stored in advance in the memory 8.

[0070] When the various conditions described above are satisfied, the point cloud information processing block 73 determines that the long-distance detection condition is satisfied and regards the pair of target high reflection measurement points as reflector points. This makes it possible to detect the reflector points of a vehicle ahead that is far away with high accuracy.

[0071] (3-3) Long-distance non-detection conditions When there are successive high reflection measurement points indicating a long distance, the point cloud information processing block 73 regards these high reflection measurement points as non-reflector points.

[0072] FIG. 5 shows a measurement point sequence when the long-distance non-detection condition is satisfied. As shown in FIG. 5, in this case, the highly reflective measurement points are continuous on the same scanning line. Furthermore, this measurement point sequence indicates a position farther than a second distance (e.g., 35 m) in the depth direction. The above-mentioned second distance is a threshold value used to determine whether the long-distance non-detection condition is satisfied with respect to the X coordinate, and is stored in advance in the memory 8. In this case, the point cloud information processing block 73 determines that the highly reflective measurement points are continuous on the same scanning line and indicate a position farther than the above-mentioned predetermined distance, and detects these highly reflective measurement points as non-reflector points.

[0073] In this way, the point cloud information processing block 73 can preferably exclude measurement points of highly reflective objects, such as signs, that are located far away from the reflector points.

[0074] (3-4) Near-distance detection conditions The point cloud information processing block 73 determines whether or not the conditions for short-distance detection, including the reflection intensity, the X coordinate, the Y coordinate, and the Z coordinate, are satisfied.

[0075] First, the conditions related to reflection intensity will be described. In this case, the point cloud information processing block 73 determines whether the reflection intensity of the target point is greater than a first threshold value and whether the sequence of measurement points that is consecutively present on the same scanning line as the target point satisfies a predetermined condition.

[0076] 6(A) to (K) show examples of measurement point sequences that satisfy the short-distance detection condition for reflection intensity. Here, high reflection measurement points with reflection intensity greater than a first threshold are shown by hatched circles, and low reflection measurement points with reflection intensity less than a second threshold are shown by unhatched circles. The horizontal lines indicate scanning lines.

[0077] As shown in Fig. 6(A) to (K), a measurement point sequence that satisfies the short-distance detection condition for reflection intensity includes a pair of high reflection measurement points including a target point, and 2 to 12 low reflection measurement points between the pair. The short-distance detection condition is a reflector point detection condition that assumes that a vehicle ahead is located at a short distance, and therefore includes a range of cases in which the number of low reflection measurement points between a pair of high reflection measurement points is larger than that in the case of the long-distance detection condition described above. The number of low reflection measurement points is determined in advance, for example, based on experiments, and is stored in memory 8. The number of low reflection measurement points that should be between a pair of high reflection measurement points to satisfy the short-distance detection condition is an example of a "predetermined number."

[0078] Next, the condition related to the X coordinate for satisfying the short-distance detection condition will be described. The point cloud information processing block 73 sets, as a necessary condition for satisfying the short-distance detection condition, that the distance in the depth direction indicated by the target measurement point sequence is equal to or less than a first distance (e.g., 30 m). Next, the condition related to the Y coordinate for satisfying the short-distance detection condition will be described. Considering that the interval between reflectors is equal to or less than the width of the vehicle, the point cloud information processing block 73 determines, as a necessary condition for satisfying the short-distance detection condition, that the width in the Y coordinate direction indicated by the measurement point sequence is less than a predetermined distance (e.g., 2.5 m). Next, the condition related to the Z coordinate for satisfying the short-distance detection condition will be described. Considering that the reflector is located at a position lower than the vehicle height, the point cloud information processing block 73 determines, as a necessary condition for satisfying the short-distance detection condition, that the height indicated by the measurement point sequence with respect to the ground is within a predetermined distance (e.g., 1.5 m).

[0079] Preferably, the point cloud information processing block 73 may determine, as a necessary condition for the short-distance detection condition, that the depth difference (i.e., the difference in X-coordinate) between a pair of highly reflective measurement points regarded as reflector points is within a predetermined distance (e.g., 0.95 m). Information on the predetermined distance required for determining the various conditions described above is stored in advance in the memory 8.

[0080] When the various conditions described above are satisfied, the point cloud information processing block 73 determines that the short distance detection condition is satisfied and regards the pair of high reflection measurement points as reflector points. This makes it possible to detect the reflector points of a forward vehicle that is in a short distance with high accuracy. In addition, the point cloud information processing block 73 can suitably detect reflector points even for a forward vehicle that has a color such as black with low reflectance that is difficult to detect using general vehicle detection processing.

[0081] (3-5) Past detection conditions A specific description will be given of the past detection conditions, which are conditions for detecting a reflector point that does not satisfy both the long-distance detection condition and the short-distance detection condition.

[0082] In this case, in the first case, if the forward vehicle flag Fv of the attention point in the previous frame is 1 and the difference between the measured distances (i.e., X-coordinate values) of the attention point in the previous frame and the current frame is less than a predetermined distance difference (also called "distance difference dDth"), the point cloud information processing block 73 considers the attention point to be a reflector point in the current frame as well. The distance difference dDth is set to, for example, a fluctuation value of the measured distance of the forward vehicle that may occur during a time interval of one frame, and is stored in advance in the memory 8.

[0083] Fig. 7(A) is a diagram showing the attention point "Pi" in the current frame and its surrounding measurement points on a virtual scanning plane. Fig. 7(B) is a diagram showing the attention point Pi in the previous frame and its surrounding measurement points on a virtual scanning plane. Here, the distance (i.e., the measured distance) of the attention point Pi in the current frame is "x1", and the distance of the attention point Pi in the previous frame is "x2".

[0084] In this case, if the forward vehicle flag Fv of the attention point Pi in the previous frame is 1 and the absolute value of the difference between the distance x1 and the distance x2 is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of the attention point Pi in the current frame to 1. Note that, if noise removal processing has been performed, the point cloud information processing block 73 may set the forward vehicle flag Fv of the attention point Pi in the current frame to 1 only when the attention point Pi in the current frame is a valid point. The same applies to each of the cases described below.

[0085] Similarly, for the surrounding points of the attention point Pi in the current frame, if the forward vehicle flag Fv of the attention point Pi in the previous frame is 1 and the distance difference from the attention point Pi in the previous frame is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of the surrounding points to 1. Specific examples (second case to fifteenth case) will be described with continued reference to Figures 7(A) and 7(B).

[0086] In the second case, when the forward vehicle flag Fv of the attention point in the previous frame is 1 and the difference in distance between the attention point Pi in the previous frame and the measurement point P7 (the next to the attention point Pi on the left) in the current frame is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of the measurement point P7 in the current frame to 1. In the third case, when the forward vehicle flag Fv of the attention point in the previous frame is 1 and the difference in distance between the attention point Pi in the previous frame and the measurement point P6 in the current frame (the next to the attention point Pi on the left), the point cloud information processing block 73 sets the forward vehicle flag Fv of the measurement point P6 in the current frame to 1. In a fourth case, if the forward vehicle flag Fv of the attention point in the previous frame is 1 and the difference in distance between the attention point Pi in the previous frame and measurement point P8 (the point immediately to the right of the attention point Pi) in the current frame is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of measurement point P8 in the current frame to 1. In a fifth case, if the forward vehicle flag Fv of the attention point in the previous frame is 1 and the difference in distance between the attention point Pi in the previous frame and measurement point P9 (the point immediately to the right of the attention point Pi) in the current frame is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of measurement point P9 in the current frame to 1. In a sixth case, if the forward vehicle flag Fv of the attention point in the previous frame is 1 and the difference in distance between the attention point Pi in the previous frame and the measurement point P12 (directly below the attention point Pi) in the current frame is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of the measurement point P12 in the current frame to 1. In a seventh case, if the forward vehicle flag Fv of the attention point in the previous frame is 1 and the difference in distance between the attention point Pi in the previous frame and the measurement point P11 (to the lower left of the attention point Pi) in the current frame is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of the measurement point P11 in the current frame to 1.

[0087] In an eighth case, if the forward vehicle flag Fv of the attention point in the previous frame is 1 and the difference in distance between the attention point Pi in the previous frame and the measurement point P10 (two spaces to the left and below the attention point Pi) in the current frame is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of the measurement point P10 in the current frame to 1. In a ninth case, if the forward vehicle flag Fv of the attention point in the previous frame is 1 and the difference in distance between the attention point Pi in the previous frame and the measurement point P13 (to the lower right of the attention point Pi) in the current frame is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of the measurement point P13 in the current frame to 1. In a tenth case, if the forward vehicle flag Fv of the attention point in the previous frame is 1 and the difference in distance between attention point Pi in the previous frame and measurement point P14 (two points to the right below attention point Pi) in the current frame is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of measurement point P14 in the current frame to 1. In an eleventh case, if the forward vehicle flag Fv of the attention point in the previous frame is 1 and the difference in distance between attention point Pi in the previous frame and measurement point P3 (directly above attention point Pi) in the current frame is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of measurement point P3 in the current frame to 1. In a twelfth case, if the forward vehicle flag Fv of the attention point in the previous frame is 1 and the difference in distance between the attention point Pi in the previous frame and the measurement point P2 (upper left of the attention point Pi) in the current frame is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of the measurement point P2 in the current frame to 1. In a thirteenth case, if the forward vehicle flag Fv of the attention point in the previous frame is 1 and the difference in distance between the attention point Pi in the previous frame and the measurement point P1 (two spaces to the left and above the attention point Pi) in the current frame is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of the measurement point P1 in the current frame to 1. In the 14th case, if the forward vehicle flag Fv of the focus point in the previous frame is 1 and the difference in distance between the focus point Pi in the previous frame and the measurement point P4 (upper right of the focus point Pi) in the current frame is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of the measurement point P4 in the current frame to 1.In the fifteenth case, if the forward vehicle flag Fv of the focus point in the previous frame is 1 and the difference in distance between the focus point Pi in the previous frame and the measurement point P5 in the current frame (two points to the right and above the focus point Pi) is less than the distance difference dDth, the point cloud information processing block 73 sets the forward vehicle flag Fv of the measurement point P5 in the current frame to 1.

[0088] In this way, by setting the past detection condition, the point cloud information processing block 73 can preferably detect reflector points that do not meet both the long-distance detection condition and the short-distance detection condition.

[0089] (4) Surrounding point detection process Next, the surrounding point detection process will be described in detail. In summary, the point cloud information processing block 73 detects measurement points that exist around a reflector point and represent the same object as the reflector point (i.e., a forward vehicle) as the reflector surrounding points.

[0090] Fig. 8 is an example of a flowchart showing the procedure of the surrounding point detection process executed by the point cloud information processing block 73 in step S13 in Fig. 2. In this flowchart, generally speaking, the point cloud information processing block 73 regards the measurement points that do not satisfy the surrounding point non-detection condition described below and satisfy the surrounding point detection condition, among the surrounding points of the measurement points whose forward vehicle flags Fv are set to 1, as reflector surrounding points, and sets the forward vehicle flags Fv to 1.

[0091] First, the point cloud information processing block 73 sequentially regards each of all measurement points in the current frame as a point of interest, and executes a loop B consisting of step S201 and a loop C described later.

[0092] First, the point cloud information processing block 73 determines whether the forward vehicle flag Fv of the attention point is 1 or not (step S201). If the point cloud information processing block 73 determines that the forward vehicle flag Fv of the attention point is 1 (step S201; Yes), it executes Loop C consisting of steps S202 to S205. On the other hand, if the point cloud information processing block 73 determines that the forward vehicle flag Fv of the attention point is not 1 (i.e., 0) (step S201; No), it ends the processing of Loop B for the current attention point.

[0093] Next, in loop C, the point cloud information processing block 73 executes the following process for each of the surrounding points, which are measurement points existing around the point of interest. Here, the measurement points that become the surrounding points are determined depending on, for example, a parameter that specifies a search range. For example, when the parameter that specifies the search range (also called the "search range parameter") is "2", the measurement points P1 to P14 shown in FIG. 7A are determined as the surrounding points for the point of interest Pi. Also, when the search range parameter is "1", the measurement points P2 to P4, P7, P8, and P11 to P13 shown in FIG. 7A are determined as the surrounding points for the point of interest Pi.

[0094] First, the point cloud information processing block 73 determines whether the surrounding points of the target satisfy the surrounding point non-detection condition (step S202). Here, the surrounding point non-detection condition is a condition for excluding measurement points that are obviously inappropriate as measurement points that constitute a forward vehicle. The surrounding point non-detection condition is, for example, (a) The surrounding points of the target correspond to consecutive high reflectivity measurement points on the same scan line. (b) The distance of the surrounding point of the object is less than the first distance and corresponds to the ground point. The condition is determined to be satisfied when at least one of the following is satisfied. When the point cloud information processing block 73 determines that the surrounding point non-detection condition is satisfied (step S202; Yes), the point cloud information processing block 73 sets the forward vehicle flag Fv of the target surrounding point to 0 (step S203).

[0095] On the other hand, when it is determined that the surrounding point non-detection condition is not satisfied (step S202; No), the point cloud information processing block 73 determines whether the surrounding point detection condition is satisfied for the surrounding points of the target (step S204). Here, the surrounding point detection condition is, for example, (a) The difference in the X coordinate value (distance difference) from the point of interest is within a specified difference (for example, 1.0 m) (b) The difference in Y coordinate value from the point of interest is within a specified difference (for example, 1.5 m) (c) The difference in height from the ground (difference in Z coordinate value) is within a specified difference (for example, 1.5 m) It is a condition that is determined to be satisfied when all of the above are satisfied. Note that, if the point cloud information further includes flag information indicating for each measurement point whether it is a valid point or not, the point cloud information processing block 73 may determine that the target surrounding points are valid points as a necessary condition for satisfying the surrounding point detection condition. The threshold value (predetermined distance difference, etc.) used to determine the surrounding point non-detection condition and the surrounding point detection condition is stored in advance in the memory 8, for example.

[0096] Then, when the point cloud information processing block 73 determines that the surrounding points of the target satisfy the surrounding point detection condition (step S204; Yes), it regards the surrounding points of the target as reflector surrounding points and sets the forward vehicle flag Fv to 1 (step S205). On the other hand, when the point cloud information processing block 73 determines that the surrounding points of the target do not satisfy the surrounding point detection condition (step S204; No), it ends the processing of loop C for the surrounding points of the target.

[0097] After loop B is completed, the point cloud information processing block 73 sequentially regards each of all measurement points in the current frame as a point of interest, and starts loop D. Loop D includes steps S206 and S207.

[0098] First, the point cloud information processing block 73 determines whether the forward vehicle flag Fv of the attention point is 1, and the attention point is located closer than 10 m (i.e., the X coordinate value is less than 10 m), and is less than 10 cm above the ground (step S206). If the point cloud information processing block 73 determines that the forward vehicle flag Fv of the attention point is 1, and the attention point is located closer than 10 m (i.e., the X coordinate value is less than 10 m), and is less than 10 cm above the ground (step S206; Yes), the point cloud information processing block 73 sets the forward vehicle flag Fv of the attention point to 0 (step S207). This allows the point cloud information processing block 73 to set the forward vehicle flag Fv of attention points that clearly do not constitute a forward vehicle to 0. On the other hand, if the forward vehicle flag Fv for the point of interest is not 1, the point is not located more than 10 m away (i.e., the X coordinate value is less than 10 m), or the point is not located at a height less than 10 cm from the ground (step S206; No), the point cloud information processing block 73 ends the processing of loop D for the target point of interest.

[0099] In this way, by executing surrounding point detection processing, the point cloud information processing block 73 can detect measurement points that exist around the reflector point and are estimated to be the same object as the reflector point as reflector surrounding points, and set the forward vehicle flag Fv to 1.

[0100] (5) Segmentation Process Next, the segmentation process will be described in detail. In summary, the point cloud information processing block 73 generates a forward vehicle segment based on a group of measurement points for which the forward vehicle flag Fv is set to 1.

[0101] 9 and 10 are an example of a flowchart showing the procedure of the segmentation process executed by the point cloud information processing block 73 in step S14 in Fig. 2. In this flowchart, a front vehicle segment label Lv indicating the identification number of a front vehicle segment is defined. The front vehicle segment label Lv of a measurement point that does not correspond to a front vehicle segment is set to 0. Then, the point cloud information processing block 73 assigns the same front vehicle segment label Lv to the measurement point cloud clusters for each front vehicle by integrating and expanding clusters where the front vehicle flag Fv is 1.

[0102] First, the point cloud information processing block 73 sets the forward vehicle segment labels Lv of all measurement points in the current frame to 0 (i.e., initial values) (step S301). Then, the point cloud information processing block 73 sequentially regards each of all measurement points in the current frame as a point of interest and executes loop E. Loop E includes steps S302 to S305.

[0103] First, the point cloud information processing block 73 determines whether the forward vehicle flag Fv of the attention point is 1 and the forward vehicle segment label Lv is 0 (step S302). If the forward vehicle flag Fv of the attention point is 1 and the forward vehicle segment label Lv is 0 (step S302; Yes), the point cloud information processing block 73 sets a new label value to the forward vehicle segment label Lv of the attention point (step S303). In this case, for example, an unused serial number is assigned to the forward vehicle segment label Lv.

[0104] Next, the point cloud information processing block 73 determines whether or not there is a measurement point whose horizontal scan number (i.e., the scanning position on the scanning line) and depth (i.e., the distance measurement) are not far apart (step S304). If there is a measurement point whose horizontal scan number and depth are not far apart (step S304; Yes), the point cloud information processing block 73 sets the front vehicle segment label Lv of the corresponding measurement point to the same value as the front vehicle segment label Lv of the attention point (step S305). In this case, for example, the point cloud information processing block 73 sets the front vehicle segment label Lv of the measurement point whose horizontal scan number difference is within a predetermined value (e.g., 1) and whose depth difference is within a predetermined distance difference (e.g., 1.0 m) among the measurement points not processed as attention points to the same value as the front vehicle segment label Lv of the current attention point. As a result, the point cloud information processing block 73 suitably assigns the same front vehicle segment label Lv to the vertical block whose front vehicle flag Fv is 1.

[0105] After completing execution of loop E, the point cloud information processing block 73 calculates the maximum, minimum and average values ​​of the three-dimensional coordinate values ​​for the measurement point cloud for each forward vehicle segment label Lv (except when the forward vehicle segment label Lv is 0, the same applies below) (step S306). In this case, the point cloud information processing block 73 calculates the maximum, minimum and average values ​​of the X coordinate, Y coordinate and Z coordinate for each of the measurement point clouds having the same forward vehicle segment label Lv other than 0.

[0106] Next, the point cloud information processing block 73 determines whether or not there is a combination of measurement point clouds for each front vehicle segment label Lv that are close in distance in the XYZ coordinate system (step S307). If there is a combination of measurement point clouds for each front vehicle segment label Lv that are close in distance (step S307; Yes), the point cloud information processing block 73 integrates the measurement point clouds of the combination and sets them to the same front vehicle segment label Lv (step S308). For example, the point cloud information processing block 73 calculates the difference sum of the average values ​​of the XYZ coordinates for any pair of measurement point clouds for each front vehicle segment label Lv. If there is a pair for which the calculated difference sum is less than a predetermined threshold, the point cloud information processing block 73 sets the front vehicle segment label Lv of each measurement point cloud of the pair to the same value. The point cloud information processing block 73 executes this process for all combinations of pairs of measurement point clouds for each front vehicle segment label Lv. Thereafter, the point cloud information processing block 73 further calculates the maximum and minimum values ​​of the X, Y, and Z coordinates for the measurement point clouds having the same forward vehicle segment label Lv in preparation for step S313. On the other hand, if there is no combination of measurement point clouds for each forward vehicle segment label Lv that is close in distance (step S307; No), the point cloud information processing block 73 proceeds to step S309.

[0107] Next, the point cloud information processing block 73 determines whether or not there is a segment (also called an "existing segment") that has already been detected for the current frame in a process other than the forward vehicle detection process (step S309). For example, assume that general obstacle detection is performed before the forward vehicle detection process, and a label (also called a "segment label Ls") that identifies a segment for each measurement point cloud that forms a cluster of detected objects is assigned to each measurement point. In this case, the point cloud information processing block 73 refers to the segment label Ls and determines whether or not there is an existing segment. Note that a processing example in which obstacle detection processing is performed before the forward vehicle detection processing will be described in detail in the second embodiment.

[0108] If an existing segment exists (step S309; ​​Yes), the point cloud information processing block 73 recognizes the measurement point cloud for each forward vehicle segment label Lv that does not overlap with the existing segment as a new segment (step S310). In this case, the segment recognized as the new segment is a forward vehicle segment that does not correspond to an existing segment and corresponds to a forward vehicle segment detected based on the reflector point. Note that forward vehicle segments representing some forward vehicles are included in existing segments detected in a previous obstacle detection process or the like. On the other hand, if an existing segment does not exist (step S309; ​​No), the point cloud information processing block 73 recognizes each measurement point cloud for each forward vehicle segment label Lv as a segment (more specifically, a forward vehicle segment) (step S311).

[0109] Next, the point cloud information processing block 73 sequentially regards each of all measurement points in the current frame as a point of interest, and executes a loop H consisting of steps S312 and S313. First, the point cloud information processing block 73 determines whether the forward vehicle flag Fv of the point of interest is 1 (step S312). Then, if the forward vehicle flag Fv is 1 (step S312; Yes), the point cloud information processing block 73 ends the processing for the current point of interest.

[0110] On the other hand, when the forward vehicle flag Fv is 0 (step S312; No), the point cloud information processing block 73 executes step S313. In this case, the point cloud information processing block 73 sets the forward vehicle flag Fv of the measurement point existing between the minimum and maximum values ​​of the X coordinate value, the Y coordinate value, and the Z coordinate value of the measurement point cloud for each segment recognized as a segment in step S310 or step S311 to 1 (step S313). In other words, the point cloud information processing block 73 virtually sets the smallest rectangular parallelepiped (each side is assumed to be parallel to one of the coordinate axes X, Y, and Z) including the measurement point cloud in the XYZ coordinate system for each segment recognized as a segment. Then, the point cloud information processing block 73 sets the forward vehicle flag Fv of the measurement point existing within the rectangular parallelepiped formed by the measurement point cloud for each segment to 1. Thereby, the point cloud information processing block 73 can suitably set the forward vehicle flag Fv to 1 even for the measurement point of the forward vehicle that is not close to the reflector point. In this case, preferably, the point cloud information processing block 73 sets the forward vehicle flag Fv to 1 only for measurement points that are not noise (that is, valid points) and are not ground points.

[0111] As described above, the control unit 7 of the LIDAR 100 according to the first embodiment functions as an information processing device in the present invention, and includes a point cloud information generation block 72 functioning as an acquisition means, a reflective member point detection means, a surrounding point detection means, and a point cloud information processing block 73 functioning as a forward vehicle detection means. The point cloud information generation block 72 acquires point cloud information obtained by receiving reflected light of light emitted in a predetermined direction. The point cloud information processing block 73 detects reflector points representing a plurality of reflectors provided outside the vehicle from the measurement point cloud indicated by the point cloud information. The point cloud information processing block 73 also detects reflector surrounding points existing around the reflector points. Then, the point cloud information processing block 73 detects the forward vehicle segment formed by the reflector points and the reflector surrounding points as a vehicle existing in a predetermined direction. This allows the LIDAR 100 to suitably detect a forward vehicle that is far away and a forward vehicle having a color with low reflectance that is difficult to detect by general object detection.

[0112] <Second Example> In the second embodiment, the LIDAR 100 executes an object identification process that includes, as part of the process, the forward vehicle detection process described in the first embodiment. The configuration of the LIDAR 100 according to the second embodiment is similar to that shown in FIG. 1, and therefore the description thereof will be omitted.

[0113] FIG. 11 is an example of a flowchart showing a procedure of the object identification process in the second embodiment.

[0114] First, the point cloud information generation block 72 of the control unit 7 of the LIDAR 100 performs a process of generating point cloud information (step S1). In this case, the point cloud information generation block 72 performs the same process as step S11 in FIG. 2 to generate point cloud information.

[0115] Next, the point cloud information generation block 72 performs a noise removal process on the point cloud information generated in step S1 (step S2). In this case, the point cloud information generation block 72 detects noise generation points generated by erroneous detection of an object, and adds flag information that designates measurement points other than the noise generation points as valid points to the point cloud information.

[0116] Next, the point cloud information processing block 73 performs an obstacle / ground detection process (step S3) for detecting obstacles (i.e., objects other than the ground) and the ground based on the point cloud information after the noise removal process in step S2. In this case, the point cloud information processing block 73 performs a process for detecting valid points corresponding to the obstacles or the ground.

[0117] Then, the point cloud information processing block 73 executes a segment generation process, which is a process of dividing the measurement points detected in step S3 as points constituting the obstacle into segments for each object (step S4). In this case, the point cloud information processing block 73 assigns an identification number such as a serial number to each generated segment, and further adds a segment label Ls indicating the identification number of the corresponding segment for each measurement point to the point cloud information.

[0118] Then, the point cloud information processing block 73 executes a forward vehicle detection process (step S5). The forward vehicle detection process in the second embodiment will be described in detail with reference to FIG.

[0119] Thereafter, the point cloud information processing block 73 executes an object identification process, which is a process for identifying an object for each segment (step S6). In this case, the point cloud information processing block 73 executes a process for identifying an object for each segment based on the shape of each segment, etc.

[0120] Fig. 12 is an example of a flowchart showing the procedure of the forward vehicle detection process in the second embodiment. First, the point cloud information processing block 73 executes the reflector point detection process in step S12, the surrounding point detection process in step S13, and the segmentation process in step S14 shown in Fig. 2 in the same manner as in the first embodiment. After that, the point cloud information processing block 73 executes the segment integration process which is a process of integrating segments (step S15). Furthermore, the point cloud information processing block 73 executes post-processing (step S16).

[0121] Fig. 13 is an example of a flowchart showing the procedure of the segment integration process executed by the point cloud information processing block 73 in step S15 in Fig. 12. In this flowchart, it is assumed that the segment label Ls is set to an initial value of 0 for a segment newly recognized in the forward vehicle detection process (see steps S310 and S311 in Fig. 10).

[0122] First, the point cloud information processing block 73 sequentially sets each of all measurement points in the current frame as a point of interest, and executes the following loop J made up of steps S401 to S405.

[0123] First, the point cloud information processing block 73 determines whether the forward vehicle flag Fv of the attention point is 1 (step S401). If the forward vehicle flag Fv of the attention point is not 1 (step S401; No), the point cloud information processing block 73 ends the processing of the loop J for the target attention point. On the other hand, if the forward vehicle flag Fv of the attention point is 1 (step S401; Yes), it determines whether the segment label Ls of the attention point is an initial value (i.e., 0) indicating that the attention point does not belong to any segment (step S402). If the segment label Ls is not 0 (step S402; No), the point cloud information processing block 73 ends the processing of the loop J for the target attention point.

[0124] On the other hand, if the segment label Ls is 0 (step S402; Yes), the point cloud information processing block 73 determines whether or not a measurement point that is close to the focus point (for example, the three-dimensional distance is within a predetermined distance difference) and has a segment label Ls other than 0 exists within the search range (step S403). The search range is determined based on the search range parameters stored in the memory 8, for example, in the same way as when determining the surrounding points. Then, if a measurement point with a segment label Ls other than 0 does not exist within the search range (step S403; No), the point cloud information processing block 73 ends the processing of loop J for the current focus point.

[0125] On the other hand, if a measurement point that is close and has a segment label Ls other than 0 exists within the search range (step S403; Yes), the point cloud information processing block 73 sets the segment label Ls of the attention point to the same value as the segment label Ls of the measurement point (step S404). Then, the point cloud information processing block 73 sets the segment label Ls of the measurement point to which the same forward vehicle segment label Lv as the attention point is assigned to the same value as the segment label Ls of the attention point (step S405). In this way, when the distance between the existing segment (i.e., the segment generated in step S4) and the forward vehicle segment newly detected in the forward vehicle detection process is close, the point cloud information processing block 73 suitably integrates them as the same segment. Note that the point cloud information processing block 73 may set the segment label Ls of the measurement point whose distance measurement distance from the attention point is within a predetermined distance difference to the same value as the segment label Ls of the attention point.

[0126] Next, the point cloud information processing block 73 sequentially regards each of all measurement points in the current frame as a point of interest, and executes a loop K consisting of steps S406 to S408.

[0127] First, the point cloud information processing block 73 determines whether the forward vehicle flag Fv of the attention point is 1 and the segment label Ls is 0 (step S406). If the forward vehicle flag Fv of the attention point is 0 or the segment label Ls is other than 0 (step S406; No), the point cloud information processing block 73 ends the processing of the loop K for the current attention point.

[0128] On the other hand, if the forward vehicle flag Fv of the attention point is 1 and the segment label Ls is 0 (step S406; Yes), the point cloud information processing block 73 sets a new segment label Ls to the attention point (step S407). In this case, the point cloud information processing block 73 determines that the attention point constitutes a segment different from the existing segments, and assigns an unused segment label Ls to the attention point. Then, the point cloud information processing block 73 sets the segment label Ls of the measurement point that has been assigned the same forward vehicle segment label Lv as the attention point to the same value as the segment label Ls of the attention point (step S408).

[0129] According to the segment integrating process shown in FIG. 13, the point cloud information processing block 73 can suitably integrate an existing segment and a forward vehicle segment newly detected by the forward vehicle detection process into the same segment when the distance between them is close.

[0130] FIG. 14 is an example of a flowchart showing the procedure of post-processing executed by the point cloud information processing block 73 in step S16 of FIG.

[0131] First, the point cloud information processing block 73 calculates the maximum, minimum, average, etc. of the X coordinate value, Y coordinate value, and Z coordinate value of each segment identified by the segment label Ls (step S501). Then, the point cloud information processing block 73 determines whether or not there is a segment whose X length corresponding to the difference between the maximum and minimum values ​​of the X coordinate value, or whose Y length corresponding to the difference between the maximum and minimum values ​​of the Y coordinate value, is greater than a predetermined threshold value (step S502). This threshold value is, for example, a value corresponding to the maximum expected width length of an obstacle (including a forward vehicle) to be detected, and is stored in advance in the memory 8.

[0132] If there is a segment whose X length or Y length is greater than the predetermined threshold (step S502; Yes), the point cloud information processing block 73 sets the forward vehicle flag Fv of the current frame and the previous frame of each measurement point constituting the corresponding segment to 0 (step S503). On the other hand, if there is no segment whose X length or Y length is greater than the predetermined threshold (step S502; No), the point cloud information processing block 73 proceeds to the process of step S504.

[0133] Next, in step S504, the point cloud information processing block 73 determines whether or not a forward vehicle segment exists (step S504). In this case, for example, if a segment including a measurement point whose forward vehicle segment label Lv is set to a value other than 0 exists, the point cloud information processing block 73 determines that a forward vehicle segment exists. If a forward vehicle segment exists (step S504; Yes), the point cloud information processing block 73 determines whether or not a forward vehicle segment whose width is greater than a threshold exists (step S505). In this case, for example, the point cloud information processing block 73 determines whether or not a difference between the minimum and maximum values ​​of the Y coordinates of the forward vehicle segment is greater than a threshold corresponding to an expected vehicle width. If a forward vehicle segment whose width is greater than the threshold exists (step S505; Yes), the point cloud information processing block 73 regards the measurement points located at the ends of the corresponding forward vehicle segment as noise data (step S506). For example, the point cloud information processing block 73 continues the process of regarding the measurement points constituting the minimum and maximum values ​​of the Y coordinate of the forward vehicle segment as noise data until the difference between the minimum and maximum values ​​of the Y coordinate of the forward vehicle segment becomes equal to or less than a threshold. This makes it possible to suitably remove the noise data added to the forward vehicle segment. On the other hand, if the width is equal to or less than the threshold (step S505; No), the point cloud information processing block 73 proceeds to step S507.

[0134] Next, the point cloud information processing block 73 determines whether or not there is a segment that exists before 10 m (i.e., the X coordinate value is less than 10 m) and the total number of measurement points constituting the segment is less than 5 (step S507). Note that "5" and "10 m" here are just examples, and may be set to different values. Then, if there is a segment that exists before 10 m and the total number of measurement points constituting the segment is less than 5 (step S507; Yes), the point cloud information processing block 73 sets the forward vehicle flag Fv of the current frame and the previous frame of each measurement point constituting the target segment to 0 (step S508). This allows the point cloud information processing block 73 to suitably remove a mass of noise data determined to be a segment. On the other hand, if there is no segment that contains less than 5 measurement points constituting the segment (step S507; No), the point cloud information processing block 73 proceeds to the process of step S509.

[0135] Next, the point cloud information processing block 73 determines whether or not there is a segment with less than two total measurement points (step S509). If there is a segment with less than two total measurement points (step S507; Yes), the point cloud information processing block 73 sets the forward vehicle flag Fv of the current frame and the previous frame of each measurement point that constitutes the target segment to 0 (step S510). This allows the point cloud information processing block 73 to preferably remove segments that are clearly estimated to be noise. Note that "two" in step S509 is an example, and may be set to a different value. On the other hand, if there is no segment with less than two total measurement points that constitute the segment (step S509; No), the point cloud information processing block 73 advances the process to loop L.

[0136] Next, the point cloud information processing block 73 sequentially regards each of all measurement points in the current frame as a point of interest, and starts processing of a loop L consisting of steps S511 and S512. First, the point cloud information processing block 73 determines whether or not the forward vehicle flag Fv is 1 (step S511). Then, if the forward vehicle flag Fv is 1 (step S511; Yes), the point cloud information processing block 73 stores the XYZ coordinate values ​​of the point of interest in the memory 8 in association with the time information and the forward vehicle flag Fv (step S512). On the other hand, if the forward vehicle flag Fv is 0 (step S511; No), the point cloud information processing block 73 ends processing of the loop L for the current point of interest.

[0137] According to the second embodiment, the LIDAR 100 can suitably detect, in the forward vehicle detection process, a distant forward vehicle that is difficult to detect in the obstacle detection process or a forward vehicle having a color with low reflectance.

[0138] <Modification> Next, a description will be given of various modified examples that can be suitably applied to the first and second embodiments. The following modified examples may be applied to the first or second embodiment in any combination.

[0139] (First Modification) The configuration of the LIDAR 100 is not limited to the configuration shown in Fig. 1. For example, the point cloud information processing block 73 of the control unit 7 and a function corresponding to the point cloud information processing block 73 may be provided in a device other than the LIDAR 100.

[0140] 15 is a configuration diagram of a LIDAR system according to a first modified example. The LIDAR system includes a LIDAR 100X and an information processing device 200. In this case, the LIDAR 100X supplies point cloud information generated by a point cloud information generation block 72 to the information processing device 200.

[0141] The information processing device 200 has a control unit 7A and a memory 8. The memory 8 stores information necessary for the control unit 7A to execute processing. The control unit 7A functionally has a point cloud information acquisition block 72A and a point cloud information processing block 73. The point cloud information acquisition block 72A receives point cloud information generated by the point cloud information generation block 72 of the LIDAR 100X, and supplies the received point cloud information to the point cloud information processing block 73. The point cloud information processing block 73 executes the same processing as the point cloud information processing block 73 of the first or second embodiment for the point cloud information supplied from the point cloud information acquisition block 72A.

[0142] The information processing device 200 may be realized by a driving assistance device. Information on parameters required for the forward vehicle detection process or the like may be stored in another device having a memory that can be referenced by the information processing device 200. According to the configuration of this modification, the information processing device 200 can suitably detect a forward vehicle that is located far away or a forward vehicle having a color with low reflectance, based on the point cloud information generated by the lidar 100X.

[0143] (Second Modification) The rider 100 is not limited to being mounted on a vehicle, and may be fixed at a predetermined location.

[0144] 16(A) and (B) are diagrams showing an example of vehicle detection by a fixed lidar 100. Fig. 16(A) shows a state in which a distant vehicle 11A with a pair of reflectors 10A at its rear is detected in a parking lot or the like, and Fig. 16(B) shows a state in which a black vehicle 11B with a pair of reflectors 10B at its rear is detected in a parking lot or the like.

[0145] Even in this case, the lidar 100 detects distant vehicles 11A based on the reflectors 10A provided on each of the distant vehicles 11A by executing the process based on the first or second embodiment. Similarly, the lidar 100 detects black vehicles 11B with low reflectance based on the reflectors 10B provided on each of the black vehicles 11B by executing the process based on the first or second embodiment.

[0146] In this way, the lidar 100 does not need to be provided on a vehicle, and may be used for any vehicle detection process, not limited to the process of detecting a forward vehicle.

[0147] (Third Modification) The lidar 100 may be installed so that its detection range is in any direction other than the front of the vehicle.

[0148] Fig. 17 shows an overhead view of the periphery of a vehicle equipped with a lidar 100 in a third modified example. In Fig. 17, the lidar 100 is provided at the rear of the vehicle so that the rear of the vehicle is the detection range. In this case, a vehicle 11C with a pair of reflectors 10C provided at the rear is present within the detection range of the lidar 100. That is, in this case, the vehicle equipped with the lidar 100 and the vehicle 11C move, causing the vehicles to approach each other relatively, and the reflector 10C of the vehicle 11C is present within the detection range of the lidar 100.

[0149] Even in this case, the lidar 100 can preferably detect the vehicle 11C present behind based on the reflector 10C provided on each of the vehicles 11C by executing the process based on the first or second embodiment.

[0150] In this way, the lidar 100 does not need to detect the vehicle in the forward direction, but may be used to process vehicle detection in any direction.

[0151] (Fourth Modification) In the reflector point detection process, the point cloud information processing block 73 may detect, as a reflector point, a pair of measurement points that exist in a horizontal direction relative to the ground and satisfy a predetermined condition, regardless of whether they exist on the same scan line or not.

[0152] 18(A) and (B) show an example of a sequence of measurement points that is determined to satisfy a new long distance detection condition in this modified example, in addition to the examples of the sequence of measurement points in FIGS. 4(A) to (D).

[0153] 18(A) and (B), there is one high reflection measurement point that is a candidate for a reflector point on each of the adjacent scanning lines. Even in these cases, the point cloud information processing block 73 determines that the long distance detection condition is satisfied.

[0154] Here, a supplementary explanation of this modified example will be given. Due to the influence of the installation conditions (including tilt) of the lidar 100 and the tilt of the detection vehicle, there are cases where a pair of reflectors exists horizontally with respect to the ground (road, parking lot, etc.), but no reflector points exist on the same scanning line. Taking the above into consideration, in this modified example, a measurement point sequence that is not on the same scanning line as shown in Figures 18(A) and (B) is also added as an example that satisfies the long-distance detection condition. This allows the lidar 100 to suitably detect reflector points without being affected by the tilt of the lidar 100 or the tilt of the detection vehicle. Similarly, the point cloud information processing block 73 sets the short-distance detection condition and the long-distance non-detection condition so as to be conditions that take into account the case where a pair of reflectors does not exist on the same scanning line.

[0155] (Fifth Modification) In the reflector point detection process, when selecting high reflection measurement points and low reflection measurement points, the point cloud information processing block 73 may select based on the relative value of the reflection intensity instead of using thresholds (first threshold and second threshold).

[0156] As an example, a process of determining whether the three measurement points shown in Fig. 4A are high reflection measurement points will be described. Here, the reflection intensity of the left measurement point is "Intensity1", the reflection intensity of the center measurement point is "Intensity2", and the reflection intensity of the right measurement point is "Intensity3". In this case, for example, the point cloud information processing block 73 performs the following: Intensity1 > (Intensity2 × 3) Intensity3 > (Intensity2 × 3) is satisfied (i.e., when the ratio of reflection intensity is three times greater than that of the central measurement point), it is determined that the left and right measurement points have relatively higher reflection intensity values ​​than the central measurement point, and these measurement points are regarded as high reflection measurement points that are candidates for reflector points. Note that the parameter "3" relating to the ratio in the right-hand side of the above equation is an example, and is actually set to a suitable value determined in advance. In this way, the point cloud information processing block 73 can suitably select high reflection measurement points and other low reflection intensity points based on the ratio of reflection intensity between adjacent measurement points.

[0157] (Sixth Modification) The reflector that the lidar 100 detects as a reflector point is not limited to a reflector provided at the rear of the vehicle, and the lidar 100 may detect reflector points for multiple reflectors provided at a location other than the rear (for example, the side or front). In this case, the lidar 100 can suitably detect a large vehicle or the like that has multiple reflectors provided at the side or the like.

[0158] Although the present invention has been described above with reference to the embodiments, the present invention is not limited to the above-mentioned embodiments. Various modifications that can be understood by a person skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention. In other words, the present invention naturally includes various modifications and corrections that a person skilled in the art could make in accordance with the entire disclosure, including the claims, and the technical ideas. In addition, the disclosures of the above cited patent documents and the like are incorporated herein by reference. [Explanation of symbols]

[0159] 1. Transmitter 2. Receiving section 3 Beam splitter 5. Scanner 6 Piezo Sensor 7, 7A control section 8. Memory 100, 100X Lidar 200 Information processing device

Claims

1. an acquisition means for acquiring point cloud information obtained by receiving reflected light of light emitted in a predetermined direction; a reflecting member point detection means for detecting reflecting member points representing a plurality of reflecting members provided outside the vehicle from the measurement point cloud indicated by the point cloud information; a surrounding point detection means for detecting surrounding points existing around the reflecting member point; a vehicle detection means for detecting a segment in the measurement point cloud formed by the reflecting member points and the surrounding points as the vehicle present in the predetermined direction; having the reflecting member point detection means detects a pair of measurement points having a reflection intensity greater than a first threshold value and a predetermined number of measurement points having a reflection intensity less than a second threshold value in the measurement point cloud as a series of measurement points that are continuously present in a horizontal direction relative to the ground, the pair being the reflecting member point; The information processing device, wherein the reflecting member point detection means determines the predetermined number based on the distance in the depth direction indicated by the sequence of measurement points.

2. an acquisition means for acquiring point cloud information obtained by receiving reflected light of light emitted in a predetermined direction; a reflecting member point detection means for detecting reflecting member points representing a plurality of reflecting members provided outside the vehicle from the measurement point cloud indicated by the point cloud information; a surrounding point detection means for detecting surrounding points existing around the reflecting member point; a vehicle detection means for detecting a segment in the measurement point cloud formed by the reflecting member points and the surrounding points as the vehicle present in the predetermined direction; having the reflecting member point detection means detects a pair of measurement points as the reflecting member point when a pair of measurement points having a reflection intensity greater than other measurement points by a predetermined ratio or more and a predetermined number of measurement points between the pair having a reflection intensity smaller than the reflection intensity of the pair by a predetermined ratio or more exist in the measurement point cloud as a measurement point sequence that exists continuously in a horizontal direction with respect to the ground, The information processing device, wherein the reflecting member point detection means determines the predetermined number based on the distance in the depth direction indicated by the sequence of measurement points.

3. 3. The information processing device according to claim 1, wherein the reflective member point detection means changes the range of the predetermined number of points when the depth direction distance indicated by the measurement point sequence is longer than a first distance and when the distance indicated by the measurement point sequence is equal to or shorter than the first distance.

4. The information processing device according to any one of claims 1 to 3, wherein the reflective member point detection means determines whether the pair is the reflective member point based on at least one of the width of the pair, the distance difference in the depth direction of the pair, or the height of the pair from the ground.

5. An information processing device according to any one of claims 1 to 4, wherein the reflective member point detection means excludes a measurement point from candidates for the reflective member point if the measurement point has a depth distance greater than a second distance and a reflection intensity greater than a first threshold value and is continuous horizontally relative to the ground.

6. 6. The information processing apparatus according to claim 1, wherein the reflective member point detection means detects reflective member points at a current processing time based on a detection result of the vehicle at a previous processing time.

7. 7. The information processing device according to claim 6, wherein the reflective member point detection means detects as the reflective member point a measurement point where a difference in distance in the depth direction measured between the preprocessing time and the current processing time is within a predetermined distance difference and corresponds to a scanning position detected as the vehicle at the preprocessing time.

8. The acquiring means acquires, as the point cloud information, point cloud information in which a flag is attached to each of the measurement points indicating whether or not the measurement points are valid points that do not correspond to noise; The information processing apparatus according to any one of claims 1 to 7, wherein the reflecting member point detection means excludes a measurement point that is flagged as not being a valid point from candidates for the reflecting member point.

9. The information processing device according to any one of claims 1 to 8, wherein the surrounding point detection means detects measurement points that exist around the reflecting member point and whose depth direction distance difference from the reflecting member point is within a predetermined distance difference as the surrounding points.

10. 10. The information processing device according to claim 1, wherein the vehicle detection means regards measurement points within a rectangular parallelepiped formed based on the reflective member points and the surrounding points as part of the segment.

11. The information processing device according to any one of claims 1 to 10, wherein the plurality of reflective members are provided at a front part, a side part, or a rear part of an exterior of the vehicle.

12. A method executed by an information processing device, comprising: an acquisition step of acquiring point cloud information obtained by receiving reflected light of light emitted in a predetermined direction; a reflective member point detection step of detecting reflective member points representing a plurality of reflective members provided outside the vehicle from the measurement point cloud indicated by the point cloud information; a surrounding point detection step of detecting surrounding points existing around the reflecting member point; a vehicle detection step of detecting a segment in the measurement point cloud formed by the reflective member points and the surrounding points as the vehicle present in the predetermined direction; having The reflecting member point detection step detects a pair of measurement points having a reflection intensity greater than a first threshold value and a predetermined number of measurement points having a reflection intensity less than a second threshold value in the measurement point cloud as a measurement point sequence that exists continuously in a horizontal direction with respect to the ground, and detects the pair as the reflecting member point; The reflective member point detection step determines the predetermined number based on a distance in a depth direction indicated by the sequence of measurement points.

13. A method executed by an information processing device, comprising: an acquisition step of acquiring point cloud information obtained by receiving reflected light of light emitted in a predetermined direction; a reflective member point detection step of detecting reflective member points representing a plurality of reflective members provided outside the vehicle from the measurement point cloud indicated by the point cloud information; a surrounding point detection step of detecting surrounding points existing around the reflecting member point; a vehicle detection step of detecting a segment in the measurement point cloud formed by the reflective member points and the surrounding points as the vehicle present in the predetermined direction; having The reflecting member point detection step detects a pair of measurement points as the reflecting member point when a pair of measurement points having a reflection intensity greater than other measurement points by a predetermined ratio or more and a predetermined number of measurement points between the pair having a reflection intensity smaller than the reflection intensity of the pair by a predetermined ratio or more exist in the measurement point cloud as a measurement point sequence that exists continuously in a horizontal direction with respect to the ground, The reflective member point detection step determines the predetermined number based on a distance in a depth direction indicated by the sequence of measurement points.

14. an acquisition means for acquiring point cloud information obtained by receiving reflected light of light emitted in a predetermined direction; a reflecting member point detection means for detecting reflecting member points representing a plurality of reflecting members provided outside the vehicle from the measurement point cloud indicated by the point cloud information; a surrounding point detection means for detecting surrounding points existing around the reflecting member point; a computer that functions as a vehicle detection unit that detects a segment in the measurement point cloud formed by the reflective member points and the surrounding points as the vehicle present in the predetermined direction; the reflecting member point detection means detects a pair of measurement points having a reflection intensity greater than a first threshold value and a predetermined number of measurement points having a reflection intensity less than a second threshold value in the measurement point cloud as a series of measurement points that are continuously present in a horizontal direction relative to the ground, the pair being the reflecting member point; The reflecting member point detection means determines the predetermined number based on the distance in the depth direction indicated by the sequence of measurement points.

15. an acquisition means for acquiring point cloud information obtained by receiving reflected light of light emitted in a predetermined direction; a reflecting member point detection means for detecting reflecting member points representing a plurality of reflecting members provided outside the vehicle from the measurement point cloud indicated by the point cloud information; a surrounding point detection means for detecting surrounding points existing around the reflecting member point; a computer that functions as a vehicle detection unit that detects a segment in the measurement point cloud formed by the reflective member points and the surrounding points as the vehicle present in the predetermined direction; the reflecting member point detection means detects a pair of measurement points as the reflecting member point when a pair of measurement points having a reflection intensity greater than other measurement points by a predetermined ratio or more and a predetermined number of measurement points between the pair having a reflection intensity smaller than the reflection intensity of the pair by a predetermined ratio or more exist in the measurement point cloud as a measurement point sequence that exists continuously in a horizontal direction with respect to the ground, The reflecting member point detection means detects the predetermined number of reflecting members based on the distance in the depth direction indicated by the measurement point sequence. Determine the number, program.

16. A storage medium storing the program according to claim 14 or 15.

Images