Information processing device, method, program, and storage medium

Abstract

To accurately determine erroneously generated false data.SOLUTION: A controller 13 of an information processing device 1 detects a vehicle segment, which is a segment representing a vehicle, from the measurement data output by a Lidar 3, which is a measuring device. The controller 13 also sets a shadow area Rs, which is an area serving as a blind spot for the Lidar 3 due to the presence of the vehicle. Furthermore, if a second vehicle segment is present in the shadow area Rs for the first vehicle segment, the controller 13 determines that the second vehicle segment is erroneously generated false data.SELECTED DRAWING: Figure 8

Description

[Technical Field]

[0001] The present disclosure relates to processing of measured data. [Background technology]

[0002] Laser radar devices that irradiate a detection space with pulses of laser light and detect objects within the detection space based on the level of the reflected light have been known for some time. For example, Patent Document 1 discloses a lidar that scans the surrounding space by appropriately controlling the emission direction (scanning direction) of repeatedly emitted light pulses and observes the returned light to generate point cloud data representing information about surrounding objects, such as distance and reflectivity. Furthermore, Patent Document 2 discloses a technology that compares observation data obtained by a distance measuring device with distance data of background candidates stored in a background candidate data storage unit and counts up the number of observations of background candidates whose distance data matches the observation data, in order to determine an appropriate number of backgrounds based on the observation data even when the observation data from the distance measuring device is unstable. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2018-009831 [Patent Document 2] Japanese Patent Application Publication No. 2017-207365 Summary of the Invention [Problem to be solved by the invention]

[0004] When a laser beam emitted by a measuring device such as a lidar is reflected by an object and irradiates another object, there is a problem that measurement data of the other object is generated erroneously. When such measurement data is used for object detection processing, the number of detected objects, etc., is calculated erroneously.

[0005] The present disclosure has been made to solve the above-mentioned problems, and has as its main object to provide an information processing device that can accurately determine false data that has been generated by mistake. [Means for solving the problem]

[0006] The claimed invention is a segment detection means for detecting segments representing objects from the measurement data output by the measurement device, and detecting vehicle segments representing vehicles from among the segments based on conditions related to the size or shape of the segments; an area setting means for setting an area that is a blind spot of the measurement device due to the vehicle for each vehicle segment; a determination means for determining, when a second segment, which is a vehicle segment different from the first segment, exists within the area for the first segment selected from the vehicle segments, that the second segment is false data generated by mistake; The information processing device is provided with:

[0007] The claimed invention also includes: 1. A computer-implemented method comprising: Detecting segments representing objects from the measurement data output by the measurement device, and detecting vehicle segments representing vehicles from among the segments based on conditions related to the size or shape of the segments; a blind spot area of ​​the measurement device due to the vehicle is set for each vehicle segment; If a second segment, which is a vehicle segment different from the first segment, exists within the area for the first segment selected from the vehicle segments, the second segment is determined to be false data generated by mistake. It is a method.

[0008] The claimed invention also includes: Detecting segments representing objects from the measurement data output by the measurement device, and detecting vehicle segments representing vehicles from among the segments based on conditions related to the size or shape of the segments; a blind spot area of ​​the measurement device due to the vehicle is set for each vehicle segment; This is a program that causes a computer to execute a process of determining that a second segment, which is a vehicle segment different from the first segment, is false data that has been generated in error, when the second segment exists within the area for a first segment selected from any of the vehicle segments. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a schematic configuration of a rider unit according to an embodiment. [Figure 2] An example of a lidar installation for measuring vehicles on a road is shown. [Figure 3] FIG. 2 is a block diagram illustrating an example of a hardware configuration of an information processing device. [Figure 4] An example of generating fake data is shown below. [Figure 5] An example of setting a shadow area is shown below. [Figure 6] 10 is a diagram clearly showing end points A and B of a large vehicle, which serve as base points when generating a shadow area of ​​the large vehicle. [Figure 7] FIG. 10 is a diagram illustrating a method for determining whether or not a virtual vehicle is included in a shadow area. [Figure 8] 10 is a flowchart illustrating an example of a procedure for vehicle detection processing. DETAILED DESCRIPTION OF THE INVENTION

[0010] According to a preferred embodiment of the present invention, an information processing device includes: a segment detection means for detecting a segment representing an object from measurement data output by a measurement device; an area setting means for setting an area that is a blind spot of the measurement device due to the object; and a determination means for determining, when a second segment detected by the segment detection means is present within the area for a first segment detected by the segment detection means, that the second segment is false data that has been erroneously generated. According to this aspect, the information processing device can accurately determine that the second segment is false data when the second segment has been erroneously generated due to an object corresponding to the first segment.

[0011] In one aspect of the information processing device, the measurement device generates, as the measurement data, point cloud data indicating the positions of the measurement points for each direction in which the light is emitted by receiving reflected light from the measurement device that is emitted in a plurality of directions, and the segment detection means detects a group of adjacent measurement points as the segment. With this aspect, when the light from the measurement device is reflected by an object corresponding to the first segment and irradiated onto another object, resulting in an erroneous generation of the second segment, the information processing device can accurately determine that the second segment is false data.

[0012] In another aspect of the information processing device, the information processing device further includes an object information output means for outputting information about the number of the objects based on the segments other than the segment corresponding to the second segment. This aspect allows the information processing device to output accurate information about the number of the objects present.

[0013] In another aspect of the information processing device, the area setting means sets, as the area, an area on a horizontal plane that is a blind spot of the measuring device due to the object. With this aspect, the information processing device can suitably reduce the amount of processing required for area setting.

[0014] In another aspect of the information processing device, the area setting means sets the area based on size information of the object corresponding to the first segment. With this aspect, the information processing device can accurately set the area even if measurement data of the object corresponding to the first segment is incomplete.

[0015] In another aspect of the information processing device, the information processing device further includes an object tracking unit that tracks the object detected from the measurement data obtained in a time series based on the segments other than the segment corresponding to the second segment. This aspect enables the information processing device to accurately track the real object.

[0016] In another aspect of the information processing device, the segment detection means detects a segment representing a vehicle as the segment representing the object. In the case of a vehicle, the presence of an aluminum loading platform or reflector can cause specular reflection or retroreflection, which can result in erroneous measurement data. Even in this case, the information processing device can accurately determine that the erroneously generated measurement data is false data. In a preferred example, the area setting means may set, as the area, an area that is a blind spot of the measurement device due to a vehicle determined to be a large vehicle among the vehicles.

[0017] According to another preferred embodiment of the present invention, a computer-executed control method includes: detecting a segment representing an object from measurement data output by a measurement device; setting a blind spot area of ​​the measurement device due to the object; and, if a detected second segment exists within the blind spot area for a detected first segment, determining that the second segment is false data that has been erroneously generated. By executing this control method, an information processing device can accurately determine that the second segment is false data when the second segment is erroneously generated due to an object corresponding to the first segment.

[0018] According to another preferred embodiment of the present invention, a computer is caused to execute a process of detecting a segment representing an object from measurement data output by a measurement device, setting an area that is a blind spot of the measurement device due to the object, and determining that a detected second segment is false data that has been erroneously generated if the detected second segment is present within the area for the detected first segment. By executing this program, the computer can accurately determine that the second segment is false data if the second segment was erroneously generated due to an object corresponding to the first segment. Preferably, the program is stored in a storage medium. [Example]

[0019] Preferred embodiments of the present invention will now be described with reference to the drawings.

[0020] (1) Overview of the Lidar Unit FIG. 1 shows a schematic configuration of a LIDAR unit 100 according to an embodiment. The LIDAR unit 100 includes an information processing device 1 that processes data generated by the sensor group 2, and a sensor group 2 that includes at least a LIDAR (Light Detection and Ranging, or Laser Illuminated Detection and Ranging) 3. FIG. 2 shows an example of the installation of a LIDAR 3 that measures vehicles on a road. The LIDAR 3 shown in FIG. 2 is installed so that the road is included in a field of view "Rv," which is the range over which the LIDAR 3 can measure distances. Note that the field of view Rv actually has a shape corresponding to the maximum measurement distance of the LIDAR 3. The LIDAR unit 100 then detects vehicles on the road and outputs information related to the vehicle detection results. Note that the LIDAR 3 may be installed so that a parking lot or the like is included in the field of view Rv, instead of the road.

[0021] The information processing device 1 is electrically connected to the sensor group 2 and processes data output by various sensors included in the sensor group 2. In this embodiment, the information processing device 1 performs a process of detecting vehicles on a road (also referred to as a "vehicle detection process") based on point cloud data output by the LIDAR 3. The information processing device 1 is fixedly installed, for example, in a state where it is housed together with the LIDAR 3 in a housing. Note that the information processing device 1 may be provided integrally with the LIDAR 3 as an electronic control device for the LIDAR 3, or may be provided in a location remote from the LIDAR 3 and capable of data communication with the LIDAR 3.

[0022] The LIDAR 3 discretely measures the distance to an object in the external world by emitting a pulsed laser, which is an infrared laser, while changing the angle within a predetermined angular range in the horizontal and vertical directions. In this case, the LIDAR 3 has an irradiation unit that irradiates laser light while changing the irradiation direction (i.e., scanning direction), a light receiving unit that receives reflected light (scattered light) of the irradiated laser light, and an output unit that outputs data based on a light receiving signal output by the light receiving unit. The data measured by the LIDAR 3 for each irradiation direction of the pulsed laser is generated based on the irradiation direction corresponding to the laser light received by the light receiving unit and the response delay time (so-called time of flight) of the laser light identified based on the above-mentioned light receiving signal. Then, for each scanning period, the LIDAR 3 generates a group of data corresponding to points irradiated by the pulsed laser within the field of view Rv of the LIDAR 3 (i.e., the irradiation range of the pulsed laser) as point cloud data for one frame. The LIDAR 3 is an example of a "measurement device" in the present invention. The LIDAR 3 is not limited to the above-mentioned scan-type LIDAR, but may be a flash-type LIDAR that generates three-dimensional data by irradiating a laser beam diffusely into the field of view of a two-dimensional array sensor. Hereinafter, the points (and the measurement data) measured by irradiation with the pulsed laser emitted from the irradiation unit will also be referred to as "measured points."

[0023] Here, when a vehicle (especially a large vehicle) is irradiated with the laser light of the LIDAR 3, the laser light of the LIDAR 3 may be specularly reflected from the vehicle due to the presence of a loading platform made of aluminum, or the laser light of the LIDAR 3 may be retroreflected from the vehicle due to the presence of a reflector made of retroreflecting material. In such cases, the laser light is reflected from the vehicle due to its high reflectivity, and the reflected light is irradiated onto another vehicle, resulting in the generation of point cloud data (also referred to as "false data") that indicates erroneous measurement results for the other vehicle. Taking the above into consideration, the information processing device 1 according to this embodiment accurately determines this false data and outputs accurate vehicle detection results that eliminate the influence of the false data.

[0024] The sensor group 2 may include various external sensors and / or internal sensors in addition to the lidar 3. For example, the sensor group 2 may include a GNSS (Global Navigation Satellite System) receiver or the like required for generating position information.

[0025] (2) Configuration of information processing device 3 is a block diagram showing an example of the hardware configuration of the information processing device 1. The information processing device 1 mainly includes an interface 11, a memory 12, and a controller 13. These elements are connected to each other via a bus line.

[0026] The interface 11 performs interface operations related to the exchange of data between the information processing device 1 and an external device. In this embodiment, the interface 11 acquires output data from a sensor group 2 such as a lidar 3, and supplies the data to the controller 13. The interface 11 may be a wireless interface such as a network adapter for wireless communication, or may be a hardware interface for connecting to an external device via a cable or the like. The interface 11 may also perform interface operations with various peripheral devices such as an input device, a display device, and a sound output device.

[0027] The memory 12 is configured by various types of volatile and non-volatile memory, such as a RAM (Random Access Memory), a ROM (Read Only Memory), a hard disk drive, and a flash memory. The memory 12 stores programs for the controller 13 to execute predetermined processes. The programs executed by the controller 13 may be stored in a storage medium other than the memory 12.

[0028] The memory 12 also stores information necessary for the controller 13 to execute predetermined processing. For example, in this embodiment, the memory 12 stores a vehicle list LV. The vehicle list LV is a list of detected vehicles generated at each vehicle detection processing cycle. The vehicle detection processing cycle is, for example, the cycle at which the LIDAR 3 generates one frame of point cloud data (also referred to as the "scanning cycle" or the "frame cycle"). In this case, for example, the detected vehicles are assigned identification information (detected vehicle IDs), and the vehicle list LV is a list of detected vehicle IDs. Here, if the same detected vehicle exists at different processing times in the vehicle detection processing, when vehicle tracking information in the subsequent stage is used, the detected vehicle is assigned the same detected vehicle ID. Note that the vehicle list LV may also associate various information about the detected vehicle based on the detection results of the detected vehicle (for example, classification information indicating whether the detected vehicle is a large vehicle or not) with the detected vehicle ID.

[0029] The controller 13 includes one or more processors such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), and a TPU (Tensor Processing Unit), and controls the entire information processing device 1. In this case, the controller 13 executes various processes, which will be described later, by executing programs stored in the memory 12, etc. Functionally, the controller 13 includes a segment detection unit 15, a filtering unit 16, a vehicle tracking unit 17, and a vehicle information output unit 18.

[0030] The segment detection unit 15 extracts adjacent data groups (also called "segments") from the point cloud data generated in the frame period corresponding to the current processing time, and detects segments (also called "vehicle segments") that are estimated to represent vehicles from the extracted segments.The segment detection unit 15 then generates a vehicle list LV based on the detection results of the vehicle segments.

[0031] In this case, the segment detection unit 15 extracts one or more segments from the point cloud data based on any clustering technique, such as Euclidean distance clustering, and detects a segment that satisfies a predetermined condition from the extracted segments as a vehicle segment. The predetermined condition may be, for example, a condition related to the size or shape of the segment. For example, the segment detection unit 15 detects a segment that has a predetermined number of measured points or more as a vehicle segment. The segment detection unit 15 may detect vehicle segments using a vehicle detection model trained based on deep learning (neural network). In this case, the vehicle detection model is, for example, a model based on instance segmentation, and is trained in advance to output a vehicle segment when point cloud data expressed in a predetermined tensor format is input. The segment detection unit 15 may further determine whether the detected vehicle is a large vehicle. In this case, the segment detection unit 15 may recognize a vehicle segment that satisfies a predetermined condition (for example, a condition regarding the number of measured points) among the vehicle segments as a vehicle segment representing a large vehicle, or may recognize a vehicle segment representing a large vehicle based on the output result of the above-mentioned vehicle detection model that has been trained to output classification information as to whether the detected vehicle is a large vehicle or not.

[0032] The filtering unit 16 filters (removes) false data that has been mistakenly detected as a vehicle segment. In this embodiment, the filtering unit 16 accurately determines false data that is generated by the reflected light of a vehicle that has been irradiated with laser light from the LIDAR 3 and the presence of another vehicle that has been irradiated with laser light. Details of the method for determining false data will be described later. Then, the filtering unit 16 deletes information about vehicles corresponding to vehicle segments that have been determined to be false data from the vehicle list LV.

[0033] The vehicle tracking unit 17 tracks vehicles in a time series based on vehicle segments corresponding to vehicles in the vehicle list LV. In this case, the vehicle tracking unit 17 uses the time series point cloud data and an arbitrary object tracking model to determine whether vehicle segments detected in consecutive frame periods represent the same vehicle. The object tracking model may be a model based on a Kalman filter or a model based on deep learning. The vehicle tracking unit 17 may determine a representative point from the vehicle segment and perform tracking based on the representative point. In this case, the representative point may be the center of gravity of the vehicle segment or a measured point corresponding to a specific part of the vehicle. In another example, the vehicle tracking unit 17 may set a bounding box for the vehicle segment and perform tracking based on the bounding box.

[0034] The vehicle information output unit 18 outputs information about vehicles detected by the LIDAR 3. In this case, the vehicle information output unit 18 may output, for example, information about the number of vehicles detected within a predetermined period based on the results of the vehicle tracking process executed by the vehicle tracking unit 17, or may output information about the position and / or speed of each detected vehicle. In this case, as one form of output, the vehicle information output unit 18 may store the above-mentioned information in the memory 12, or may transmit the information via the interface 11 to a server device that collects information about vehicles traveling on each road.

[0035] The controller 13 functions as a "segment detection means," "area setting means," "determination means," "object tracking means," "object information output means," and a computer that executes a program.

[0036] The processes executed by the controller 13 are not limited to being realized by software programs, but may be realized by any combination of hardware, firmware, and software. Furthermore, the processes executed by the controller 13 may be realized by using a user-programmable integrated circuit, such as an FPGA (Field-Programmable Gate Array) or a microcomputer. In this case, the programs executed by the controller 13 in this embodiment may be realized by using this integrated circuit.

[0037] (3) Identifying fake data Next, a method for determining false data by the filtering unit 16 will be described. In summary, the filtering unit 16 sets an area (also referred to as a "shadow area Rs") that is a blind spot for the LIDAR 3 caused by each vehicle corresponding to the vehicle segment detected by the segment detection unit 15, and determines that other vehicle segments present within the shadow area Rs are false data. In this way, the filtering unit 16 accurately detects false data generated due to the presence of a vehicle illuminated by the reflected light of a vehicle illuminated by the laser light of the LIDAR 3.

[0038] 4 shows an example of generating false data. In this example, a vehicle 90, which is a large vehicle, and a vehicle 91, which is a standard-sized vehicle, are present within the field of view Rv, and the laser light of the lidar 3, which is specularly reflected by the vehicle 90, is irradiated onto the vehicle 91, thereby generating false data representing a non-existent virtual vehicle 92 (i.e., a pseudo vehicle of the vehicle 91). Then, this false data is erroneously detected as a vehicle segment by the segment detection unit 15.

[0039] Here, a supplementary explanation will be given of the process by which the false data shown in Fig. 4 is generated. The laser light emitted from the lidar 3 along line L1 is reflected by the vehicle 90, and the reflected light is irradiated onto the vehicle 91 along line L2. In this case, the laser light reflected by the vehicle 91 is incident on the lidar 3 again along lines L2 and L1. In this case, the lidar 3 generates data that sets the endpoint of line L3, which has the same length as line L2 (more specifically, the endpoint opposite line L1), as the measurement point, based on the emission direction of the laser light and the return time of the reflected light.

[0040] In this way, when laser light reflected from a part of the first vehicle with high specular reflectivity is irradiated onto the second vehicle, a vehicle segment corresponding to the second vehicle is erroneously generated in the area behind the first vehicle (specifically, the area that is the blind spot formed by the first vehicle for the rider 3).

[0041] Taking the above into consideration, the filtering unit 16 according to this embodiment sets a shadow area Rs corresponding to each detected vehicle segment, and determines that other vehicle segments present within the shadow area Rs are false data. Note that, since the shadow area Rs is an area that is a blind spot for the LIDAR 3, the point cloud data within the shadow area Rs is uniquely determined to be false data.

[0042] FIG. 5 shows an example of setting a shadow area Rs in the situation shown in FIG. 4. In this example, the area behind the vehicle 90, which is a large vehicle, when the lidar 3 is used as a reference is set as the shadow area Rs. In other words, the shadow area Rs is a blind spot formed by the vehicle 90 for the lidar 3. The shadow area Rs contains false data representing a virtual vehicle 92, which is generated by irradiating a vehicle 91 with light reflected from the vehicle 90. Therefore, the filtering unit 16 regards the point cloud data (i.e., vehicle segments) present in the shadow area Rs as false data and filters (i.e., excludes) it. This prevents the information processing device 1 from performing tracking based on vehicle segments corresponding to non-existent vehicles or outputting vehicle information that takes non-existent vehicles into consideration.

[0043] Next, a specific method for setting the shadow area Rs will be described in detail. Below, an example will be described in which the shadow area Rs is set only for vehicles determined to be large vehicles, and the shadow area Rs is set in a two-dimensional coordinate system (i.e., a two-dimensional coordinate system on a horizontal plane) in which the height axis is deleted. Here, by setting the shadow area Rs only for vehicles determined to be large vehicles, the amount of processing can be reduced compared to when the shadow area Rs is set for all vehicles. Furthermore, for passenger vehicles that are not large vehicles, the shadow area Rs is small, the specular reflectance is lower than that of large vehicles, and there are many curved surfaces (few flat surfaces), so there is a low probability that the laser light from the LIDAR 3 reflected by the passenger vehicle will be irradiated onto other vehicles and false data will be generated. Furthermore, by setting the shadow area Rs in a two-dimensional coordinate system, the amount of processing can be suitably reduced compared to when the shadow area Rs is set in a three-dimensional coordinate system.

[0044] FIG. 6 is a diagram clearly showing two end points (end point A and end point B) of a vehicle 90, which are used as base points when generating a shadow area Rs of a vehicle 90 that is a large vehicle.

[0045] First, the filtering unit 16 sets an xy coordinate system, which is a two-dimensional coordinate system on the water surface with the position of the lidar 3 as the origin. Here, the filtering unit 16 sets the x axis to be parallel to the width direction of the lane, and sets the y axis to be parallel to the direction of vehicle movement (i.e., the direction in which the lane extends). In this way, to simplify the subsequent processing, the filtering unit 16 sets the xy coordinate system based on the direction of vehicle movement.

[0046] Next, the filtering unit 16 recognizes two endpoints (end point A and end point B) required to set the shadow area Rs from the vehicle segment for the vehicle 90. Here, end point A corresponds to the measured point of the vehicle 90 based on the laser beam whose emission angle in the two-dimensional coordinate system is smallest when the clockwise direction is used as the reference (or largest when the counterclockwise direction is used as the reference), among the measured points of the vehicle 90. Line 95 is a line connecting end point A and the origin of the two-dimensional coordinate system representing the position of the lidar 3. Furthermore, end point B corresponds to the measured point of the vehicle 90 based on the laser beam whose emission angle in the two-dimensional coordinate system is largest when the clockwise direction is used as the reference, among the measured points of the vehicle 90. Line 96 is a line connecting end point B and the origin of the two-dimensional coordinate system representing the position of the lidar 3.

[0047] Here, the filtering unit 16 determines the endpoints A and B by identifying the maximum x-coordinate value and the maximum y-coordinate value at the measured points of the vehicle 90. Here, the x-axis is parallel to the width direction of the vehicle, and the y-axis is parallel to the direction of movement of the vehicle (i.e., the length direction of the vehicle). Therefore, if the vehicle 90 is considered to be a rectangle whose longitudinal direction coincides with the direction of movement, the x-coordinate value of endpoint A will be the minimum value (A_xmin) among the measured points of the vehicle 90, and the y-coordinate value of endpoint A will be the maximum value (A_ymax) among the measured points of the vehicle 90. Furthermore, the x-coordinate value of endpoint B will be the maximum value (A_xAmax) among the measured points of the vehicle 90, and the y-coordinate value of endpoint B will be the minimum value (A_ymin) among the measured points of the vehicle 90.

[0048] Furthermore, the slope "slop_a" of the line 95 is calculated based on the pair of x and y coordinate values ​​of the end point A as follows: slop_a=A_ymax / A_xmin

[0049] Furthermore, the slope "slop_b" of the line 96 is calculated based on the pair of x and y coordinate values ​​of the end point B as follows: slop_a=A_ymin / A_xmax

[0050] In this case, the shadow area Rs is the area specified by the line segment connecting the end points A and B, and the lines 95 and 96.

[0051] The filtering unit 16 may further perform processing to determine the depth of the shadow area Rs. For example, when the filtering unit 16 determines the shadow area Rs using four points, namely, end points A and B, point C on line 95, and point D on line 96, the filtering unit 16 determines the positions of points C and D based on the installation height of the LIDAR 3 and the vehicle height of the vehicle 90. In this case, the filtering unit 16 determines point C so that, for example, the ratio of the length from the origin (i.e., the position of the LIDAR 3) to end point A to the length from the origin to point C is the ratio of the vehicle height of the vehicle 90 to the installation height of the LIDAR 3. Similarly, the filtering unit 16 determines point D so that, for example, the ratio of the length from the origin to end point B to the length from the origin to point D is the ratio of the vehicle height of the vehicle 90 to the installation height of the LIDAR 3. This makes it possible to determine a shadow area Rs whose depth corresponds to the blind spot area that is actually formed. The method of determining points C and D is not limited to this, and the filtering unit 16 may determine points C and D so that the length from the origin to point C and the length from the origin to point D are predetermined lengths, or if depth is not particularly taken into consideration, points C and D may be set at positions that are further away from the origin than the maximum ranging distance of the lidar 3.

[0052] Here, a supplementary explanation will be given of a case where either end point A or end point B cannot be accurately detected. Generally, depending on the type of vehicle, an end point far from the sensor (e.g., end point A having "A_ymax" as its y-coordinate value) may not be accurately detected. In this case, the filtering unit 16 may store size information including the vehicle length of a typical large vehicle (e.g., 12 m) in the memory 12 or the like as prior knowledge information, and estimate the coordinate values ​​of the undetected end points from the detected end points based on the size information. Note that instead of storing the vehicle size information in the memory 12 in advance, the filtering unit 16 may generate the above-mentioned size information by measuring the vehicle size based on measurement data output by an external sensor other than the lidar 3. Furthermore, end point A far from the sensor may be close to the sensor in a previous frame and may be detectable. Therefore, the filtering unit 16 may estimate the coordinate value of end point A in the current frame from the position of end point A in the previous frame and the movement speed of the segment.

[0053] Next, a specific example of a method for determining whether or not a vehicle segment is included in the shadow area Rs will be described. Fig. 7 is a diagram that schematically shows a method for determining whether or not a virtual vehicle 92 is included in the shadow area Rs of the vehicle 90. Here, as an example, the filtering unit 16 determines whether or not a representative point p (= (px, py)) of the vehicle segment representing the virtual vehicle 92 is included in the shadow area Rs, without taking depth into consideration.

[0054] In this case, the filtering unit 16 first sets a point q on the line 95 whose x-coordinate value coincides with that of the representative point p, and a point r on the line 96 whose x-coordinate value coincides with that of the representative point p. In this case, the coordinate value of point q is "(px, slop_a*px)" and the coordinate value of point r is "(px, slop_b*px)".

[0055] Then, the filtering unit 16 determines that the representative point p is included in the shadow area Rs when the representative point p exists between the points q and r and is farther from the origin than the vehicle 90. In this case, the filtering unit 16 determines that the representative point p exists between the points q and r because the following two equations, which focus on the y coordinate values ​​of the points q, r, and representative point p, are satisfied. slop_a*px>py slop_b*px <py

[0056] In addition, the filtering unit 16 determines that the representative point p is farther from the origin than the vehicle 90 based on the following two equations that represent the magnitude relationship between the coordinate values ​​of the point E of the vehicle 90 that is closest to the origin and the representative point p. A_xmin <px A_ymin <py

[0057] 7 is just an example, and any geometric method may be used to determine whether the representative point p is included in the shadow area Rs. In this case, the filtering unit 16 may determine whether the representative point p is included in the shadow area Rs by taking into account the depth of the shadow area Rs (i.e., the positions of points C and D). Furthermore, instead of determining whether the representative point p is included in the shadow area Rs, the filtering unit 16 may determine whether the entire vehicle segment or the bounding box of the virtual vehicle 92 is included in the shadow area Rs. In this case, if the filtering unit 16 determines that a predetermined percentage or more of the measured points or the bounding box that make up the vehicle segment are included in the shadow area Rs, for example, the filtering unit 16 determines that the vehicle segment is false data.

[0058] Furthermore, instead of setting the shadow area Rs in a two-dimensional coordinate system on a horizontal plane and determining whether or not a vehicle segment exists within the shadow area Rs, the filtering unit 16 may set the shadow area Rs in a three-dimensional coordinate system including the vertical direction and determine whether or not a vehicle segment exists within the shadow area Rs. In this case, the filtering unit 16 determines, based on a geometric method, a blind spot space within the field of view Rv of the LIDAR 3 generated by the vehicle in the three-dimensional coordinate system based on the LIDAR 3, as the shadow area Rs. Then, based on a general analytical method, the filtering unit 16 determines whether or not a vehicle segment of another vehicle is included within the shadow area Rs in the three-dimensional coordinate system.

[0059] (4) Processing Flow 8 is an example of a flowchart showing the procedure of the vehicle detection process executed by the information processing device 1. The information processing device 1 repeatedly executes the process of this flowchart at every predetermined processing cycle (for example, every frame cycle of the lidar 3).

[0060] First, the controller 13 of the information processing device 1 acquires point cloud data measured by the LIDAR 3 via the interface 11 (step S11). Then, the controller 13 detects vehicle segments based on the point cloud data acquired in step S11, and generates a vehicle list LV, which is a list of detected vehicles to be used in the current processing cycle, based on the detection results (step S12).

[0061] Next, the controller 13 determines whether or not two or more detected vehicles exist in the vehicle list LV generated in step S12 (step S13). If two or more detected vehicles do not exist in the vehicle list LV (step S13; No), the controller 13 determines that no vehicle segment containing false data has occurred, and ends the processing of the flowchart.

[0062] On the other hand, if there are two or more detected vehicles in the vehicle list LV (step S13; Yes), a vehicle (also referred to as a "target vehicle") to be targeted in setting the shadow area Rs is selected from the vehicle list (step S14). At this time, the target vehicles are selected starting from the vehicle closest to the sensor (here, LIDAR 3). In this way, if a non-target vehicle of this time is within the shadow area of ​​the target vehicle of this time, by setting the shadow area Rs starting from the vehicle closest to this sensor, the non-target vehicle of this time can be deleted from the vehicle list LV and the process of setting the shadow area Rs for the non-target vehicle of this time can be reduced. Then, the controller 13 determines whether the target vehicle is a large vehicle (step S15). If the controller 13 determines that the target vehicle is not a large vehicle (step S15; No), it determines that there is no need to set a shadow area Rs for the target vehicle, and proceeds to step S19.

[0063] On the other hand, if the target vehicle is a large vehicle (step S15; Yes), the controller 13 sets a shadow area Rs of the target vehicle (step S16). Then, the controller 13 determines whether or not a vehicle other than the target vehicle (also referred to as a "non-target vehicle") exists within the shadow area Rs set in step S16 (step S17). In this case, the controller 13 determines that a non-target vehicle exists within the shadow area Rs, for example, if a vehicle segment whose entirety or representative point is included within the shadow area Rs exists.

[0064] If the controller 13 determines that a non-target vehicle exists in the shadow area Rs (step S17; Yes), it deletes the non-target vehicle in the shadow area Rs from the vehicle list LV (step S18). Then, the controller 13 regards the vehicle segment corresponding to the non-target vehicle in the shadow area Rs as false data and excludes the false data from the point cloud data. Alternatively, the controller 13 determines a representative point from the vehicle segment, and when tracking is performed based on the representative point, it excludes the representative point of the vehicle segment of the false data. Furthermore, the controller 13 deletes the vehicle from the vehicle list LV. On the other hand, if a non-target vehicle does not exist in the shadow area Rs (step S17; No), the controller 13 proceeds to step S19.

[0065] Then, the controller 13 determines whether all vehicles in the vehicle list LV have been designated as target vehicles (step S19). If all vehicles in the vehicle list LV have been designated as target vehicles (step S19; Yes), the controller 13 ends the processing of the flowchart. On the other hand, if there is a vehicle in the vehicle list LV that has not been designated as a target vehicle (step S19; No), the controller 13 returns the processing to step S14 and selects a vehicle that has not been designated as a target vehicle as a target vehicle.

[0066] (5) Variations The LIDAR unit 100 may be mounted on a vehicle. In this case, the LIDAR 3 is provided on the vehicle, and the information processing device 1 is an on-board device of the vehicle or an electronic control unit (ECU) built into the vehicle. In this case, the information processing device 1 detects vehicles around the vehicle on which the LIDAR unit 100 is mounted based on point cloud data generated by the LIDAR 3. Even in this case, the information processing device 1 can accurately detect false data generated due to the presence of a vehicle illuminated with reflected light from a vehicle illuminated with laser light from the LIDAR 3, thereby preventing erroneous detection of the vehicle.

[0067] As described above, the controller 13 of the information processing device 1 according to this embodiment detects vehicle segments, which are segments representing a vehicle, from the measurement data output by the LIDAR 3, which is a measurement device. The controller 13 then sets a shadow area Rs, which is an area that is a blind spot for the LIDAR 3 due to the vehicle. If a second vehicle segment exists within the shadow area Rs for the first vehicle segment, the controller 13 determines that the second vehicle segment is false data that has been generated in error. This allows the information processing device 1 to accurately determine false data.

[0068] In the above-described embodiments, the program can be stored using various types of non-transitory computer-readable media and supplied to a controller or the like that is a computer. Non-transitory computer-readable media include various types of tangible storage media. Examples of non-transitory computer-readable media include magnetic storage media (e.g., flexible disks, magnetic tapes, hard disk drives), magneto-optical storage media (e.g., magneto-optical disks), CD-ROMs (Read Only Memory), CD-Rs, CD-R / Ws, and semiconductor memories (e.g., mask ROMs, PROMs (Programmable ROMs), EPROMs (Erasable PROMs), flash ROMs, and RAMs (Random Access Memory)).

[0069] Although the present invention has been described above with reference to the examples, the present invention is not limited to the above examples. Various modifications that would be understood by those 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 alterations that would be possible for those skilled in the art in accordance with the entire disclosure, including the claims, and the technical ideas. Furthermore, the disclosures of the above cited patent documents, non-patent documents, etc. are incorporated herein by reference. [Explanation of symbols]

[0070] 1. Information processing equipment 2 Sensor group 3 Rider 100 rider unit

Claims

1. a segment detection means for detecting segments representing objects from the measurement data output by the measurement device, and detecting vehicle segments representing vehicles from among the segments based on conditions related to the size or shape of the segments; an area setting means for setting an area that is a blind spot of the measurement device due to the vehicle for each vehicle segment; a determination means for determining, when a second segment, which is a vehicle segment different from the first segment, exists within the area for the first segment selected from the vehicle segments, that the second segment is false data generated by mistake; An information processing device comprising:

2. the measurement device receives reflected light of light emitted in a plurality of directions, and generates, as the measurement data, point cloud data indicating positions of measurement points for each direction in which the light is emitted; The information processing apparatus according to claim 1 , wherein the segment detection means detects a group of adjacent measurement points as the segment.

3. The information processing apparatus according to claim 1 , further comprising a vehicle information output unit that outputs information about the number of vehicles based on the vehicle segments other than the second segment.

4. 4. The information processing device according to claim 1, wherein the area setting means sets, as the area, an area on a horizontal plane that is a blind spot of the measuring device due to the vehicle.

5. 5. The information processing device according to claim 1, wherein the area setting means sets the area based on size information of a vehicle corresponding to the first segment.

6. The information processing device according to any one of claims 1 to 5, further comprising a vehicle tracking means for tracking the vehicle detected by the measurement data obtained in a time series based on the segments other than the segment corresponding to the second segment.

7. The information processing device according to any one of claims 1 to 6, wherein the area setting means sets, as the area, an area that is a blind spot of the measuring device due to a vehicle determined to be a large vehicle among the vehicles.

8. 1. A computer-implemented method comprising: Detecting segments representing objects from the measurement data output by the measurement device, and detecting vehicle segments representing vehicles from among the segments based on conditions related to the size or shape of the segments; a blind spot area of ​​the measurement device due to the vehicle is set for each vehicle segment; If a second segment, which is a vehicle segment different from the first segment, exists within the area for the first segment selected from the vehicle segments, the second segment is determined to be false data generated by mistake. method.

9. Detecting segments representing objects from the measurement data output by the measurement device, and detecting vehicle segments representing vehicles from among the segments based on conditions related to the size or shape of the segments; a blind spot area of ​​the measurement device due to the vehicle is set for each vehicle segment; A program that causes a computer to execute a process of determining that a second segment, which is a vehicle segment different from the first segment, is false data that was generated in error when the second segment exists within the area for a first segment selected from any of the vehicle segments.

10. A storage medium storing the program according to claim 9.

Images