JP6597352B2 - Object recognition device - Google Patents

Description

  The present invention relates to a technique for recognizing an object from distance measuring point data.

  In Patent Document 1, in an object recognition device that recognizes an object from distance measurement point data, the point cloud data is divided in parallel with the horizontal direction regardless of the distribution state in the point cloud data, and each of the divided point cloud data is divided. A technique for recognizing an object based on a feature amount is disclosed.

JP 2012-212456 A

However, since the object recognition apparatus always divides the point cloud data in parallel with the horizontal direction, there is a problem that the recognition accuracy is deteriorated depending on the shape of the object.
In view of these problems, it is an object of the present invention to improve the recognition accuracy of an object in an object recognition apparatus that recognizes an object from distance measurement point data.

The object recognition device according to one aspect of the present invention includes a point group acquisition unit, a distribution recognition unit, a point group division unit, and an object recognition unit.
The point cloud acquisition unit acquires point cloud data that is a set of distance measurement point data (S110, S310), and the distribution recognition unit recognizes the distribution state of the distance measurement point data constituting the point cloud data (S320). , S330, S350, S410). In addition, the point group dividing unit sets an orientation in a virtual plane representing one or more virtual planes that divide the area including the point cloud data into a plurality of areas according to the distribution state, and sets the direction in the virtual plane. The point cloud data is divided into a plurality of pieces (S340, S360, S420, S430, S440).

  Then, the object recognizing unit obtains a feature amount that represents the feature of the divided data numerically for each divided data that represents the point cloud data divided on the virtual plane, and based on the feature amount for each divided data, Is recognized (S200, S210, S220).

  According to such an object recognition device, since the orientation of the virtual plane for dividing the point cloud data is set according to the distribution state of the distance measuring point data, the point cloud data can be appropriately divided according to the object shape. The divided data obtained by dividing the point cloud data can be obtained in an appropriate state. Therefore, detailed feature description can be made for each divided data. Therefore, the recognition accuracy of the object can be improved.

  In the present invention, the “direction of the virtual plane” indicates the magnitude and direction of the inclination of the plane with respect to a reference plane such as a horizontal plane. Further, the reference numerals in parentheses described in this column and in the claims indicate the correspondence with the specific means described in the embodiment described later as one aspect, and the technical scope of the present invention is as follows. It is not limited.

It is a block diagram which shows schematic structure of a travel safety apparatus. It is a flowchart which shows the driving | running | working safety process which CPU of a control part performs. It is a flowchart which shows the point group division | segmentation process among driving | running | working safety processes. It is a figure which shows the example which divides | segments a cluster for every line in vertical direction division | segmentation. It is a figure which shows the example which divides | segments a cluster so that a predetermined number or more of ranging point data may be contained in one area | region in vertical direction division | segmentation. It is a figure which shows an example of a horizontal direction division | segmentation. It is a figure which shows an example of eyelid division | segmentation. It is a flowchart which shows the characteristic calculation process among driving | running | working safety processes. It is a figure which shows the example which uses the dispersion | distribution value of reflection intensity as one of the feature-values. It is a figure which shows the example which each uses the vertical direction division | segmentation and a horizontal direction division | segmentation.

Hereinafter, embodiments for carrying out the invention will be described with reference to the drawings.
[1. First Embodiment]
[1-1. Constitution]
The travel safety device 1 of this embodiment is mounted on a vehicle such as a passenger car, for example, and includes a travel safety function that suppresses the host vehicle from colliding with an object such as an obstacle. The traveling safety function includes a function for obtaining the type of object and the distance to the object.

Specifically, as shown in FIG. 1, the travel safety device 1 includes a control unit 10, a laser radar 21, a display unit 31, an alarm unit 32, and an operation instruction unit 33.
The laser radar 21 is configured as a well-known laser radar including a light emitting unit 22 and a light receiving unit 23. The laser radar 21 sets a region in which the vehicle may pass in the traveling direction of the vehicle as a detection region for detecting an object. Then, the light emitting unit 22 of the laser radar 21 sets the detection area as a number of irradiation areas divided in a grid shape in the horizontal direction and the vertical direction in advance, and irradiates laser light for each of the irradiation areas.

  In addition, what is arranged in the horizontal direction among a large number of irradiation areas divided in a lattice shape is called a line. The light receiving unit 23 of the laser radar 21 receives reflected light of the laser light obtained in each irradiation region. Then, the laser radar 21 outputs ranging point data. The distance measuring point data indicates the position where the object that reflects the laser beam exists according to the direction of the irradiation region and the timing when the reflected light is obtained. The distance measuring point data is data obtained every time the light emitting unit 22 issues, and includes information indicating the reflection intensity indicating the intensity of the reflected light.

  The display unit 31 displays an image for giving a warning or the like to the vehicle occupant in accordance with a command from the control unit 10. The warning unit 32 emits a warning sound or the like for warning a vehicle occupant in accordance with a command from the control unit 10.

  When the vehicle is about to collide with an object, the operation instructing unit 33 instructs the brake or steering of the vehicle according to a command from the control unit 10 to perform an operation for avoiding a collision with the object.

  The control unit 10 is configured around a well-known microcomputer having a CPU 11 and a semiconductor memory (hereinafter, memory 12) such as a RAM, a ROM, and a flash memory. Various functions of the control unit 10 are realized by the CPU 11 executing a program stored in a non-transitional physical recording medium. In this example, the memory 12 corresponds to a non-transitional tangible recording medium that stores a program. Further, by executing this program, a method corresponding to the program is executed. The number of microcomputers constituting the control unit 10 may be one or plural.

  The control unit 10 includes a well-known function as an electronic control device in addition to a function for executing a travel safety process, which will be described later, as a function configuration realized by the CPU 11 executing a program. The method of realizing these elements constituting the control unit 10 is not limited to software, and some or all of the elements may be realized using hardware that combines a logic circuit, an analog circuit, and the like.

[1-2. processing]
In the travel safety device 1 configured as described above, the control unit 10 performs the travel safety process shown in FIG. The travel safety process is a process that is started when the vehicle is turned on, for example, and is then repeatedly performed at regular intervals.

  In the driving safety process, as shown in FIG. 2, first, point cloud data is acquired in S110. Here, the point cloud data indicates a set of distance measuring point data in each irradiation area obtained by the laser radar 21.

  Subsequently, in S120, clustering is performed. Clustering represents a process for collecting distance measuring point data representing an object. Specifically, the distance between the position indicated by a certain distance measurement point data and the position indicated by other distance measurement point data is grouped by grouping them closer than a preset reference distance. Are extracted as clusters (objects).

  In S130, tracking is performed. In this process, the cluster is tracked using the latest point cloud data and the point cloud data acquired one cycle before. Specifically, for example, a Kalman filter, a particle filter, or the like is used to monitor the movement state of the clusters, and settings are made so that clusters having similar sizes and shapes before and after movement are tracked as representing the same object.

  Subsequently, a point group division process is performed in S200. The point group dividing process is a process of setting how to divide the point group data into a plurality of points and dividing the point group data. A plurality of divided data generated by dividing the point cloud data is referred to as divided data.

Details of the point group division processing are shown in FIG. That is, first, in S310, a cluster is acquired. The cluster here indicates point cloud data after clustering.
Subsequently, in S320, the state of distribution of distance measuring point data constituting the cluster is recognized. The “distribution state of distance measuring point data” is a numerical value indicating how a plurality of distance measuring point data are distributed. For example, in the “distribution state of ranging point data”, the size of a figure such as a rectangular parallelepiped or a rectangle circumscribing the cluster, the distance from the center of gravity position of this figure to the end point, the aspect ratio of this figure, and the cluster are configured. The number of ranging point data can be included.

  Here, first, a rectangle (cuboid) circumscribing the cluster is obtained as a circumscribed rectangle. In this process, when the horizontal direction of the host vehicle is the x axis, the traveling direction of the host vehicle is the y axis, and the vertical direction is the z axis, a rectangular parallelepiped that passes through the minimum and maximum cluster coordinate values on each axis. Is a circumscribed rectangle. At this time, the difference between the minimum value and the maximum value on the x-axis is defined as the “horizontal length” of the cluster, and the difference between the minimum value and the maximum value on the z-axis is defined as the “vertical direction length” of the cluster.

  Subsequently, in S330, a value obtained by dividing the vertical length of the cluster by the horizontal length is compared with a predetermined constant. Here, the predetermined constant is represented by “1 + ε”. Note that 0 <ε <1.

  In S330, if the value obtained by dividing the vertical length of the cluster by the horizontal length is larger than this constant, it indicates that the cluster is sufficiently vertical. In this case, vertical division is performed in S340. In the vertical division, the direction of the virtual plane for dividing the cluster is set parallel to the horizontal plane, and divided data is obtained by dividing the cluster on this virtual plane.

  When dividing in the vertical direction, for example, as shown in FIG. 4, the cluster may be set to be divided for each line in the laser radar 21. For example, as shown in FIG. 5, the cluster may be set to be divided so that the number of distance measurement point data more than a certain number (for example, 6 or more) is included in the divided area by the virtual plane. Moreover, you may set a virtual plane for every length of the predetermined vertical direction.

  If the value obtained by dividing the vertical length of the cluster by the horizontal length is equal to or smaller than the above-described constant in the process of S330, the vertical length of the cluster is divided by the horizontal length in S350. The value is compared with a predetermined constant different from the aforementioned constant. The constant here is represented by “1-ε”. Note that 0 <ε <1.

  In S350, if the value obtained by dividing the vertical length of the cluster by the horizontal length is smaller than this constant, it indicates that the cluster is sufficiently horizontal. In this case, horizontal direction division is performed in S360. Horizontal division means that the orientation of the virtual plane is set along the depth direction and parallel to the vertical direction, the cluster is set to be divided into multiple parts on this virtual plane, and the cluster is divided on this virtual plane. Indicates that the divided data is obtained. The depth direction indicates a direction that coincides or substantially coincides with the traveling direction of the host vehicle and the laser irradiation direction.

  When dividing in the horizontal direction, for example, as shown in FIG. 6, it is set to divide the cluster so that more than a certain number (for example, six or more) of distance measuring point data is included in the divided area on the virtual plane. Alternatively, a virtual plane may be set for each preset lateral length.

  In S350, if the value obtained by dividing the vertical length of the cluster by the horizontal length is equal to or larger than this constant, the cluster cannot be said to be vertically long or wide, and is within the range indicating that the aspect ratio is substantially equal. I can say that. That is, “the aspect ratio is substantially equal” indicates that the ratio of the longitudinal length and the lateral length of the object is within a preset ratio. In this case, in S410, the number of ranging point data constituting the cluster is compared with the variable C. The variable C is a value set experimentally. For example, the variable C is set to increase as the distance to the cluster or the length of the cluster in any direction increases.

  If the number of distance measuring point data is larger than the variable C, the eyelet division is performed in S420. For example, as shown in FIG. 7, the eyelet division is performed by dividing the cluster in the vertical direction and further dividing the cluster in the vertical direction by further dividing in the horizontal direction (divided data). It is set that the cluster is divided in the direction of, and divided data is obtained by dividing the cluster on this virtual plane.

  In S410, if the number of ranging point data is equal to or smaller than variable C, in S430, the vertical division similar to the process in S340 is performed, and in S440, the horizontal division similar to the process in S360 is performed. To do.

When such a process is completed, the point group dividing process is terminated and the process returns to FIG.
Subsequently, in S210, feature calculation processing is performed. The feature calculation process is a process for extracting the feature amount of the cluster. As the cluster feature amount, for example, a divided feature amount is extracted. The divided feature amount represents a feature amount for divided data obtained by dividing a cluster. The feature amount is obtained by quantifying the feature obtained from the variance, standard deviation, average value, etc. of the distance measuring points constituting the divided data.

In the feature calculation process, as shown in FIG. 8, first, in S510, the divided data generated by the point group division process is acquired.
Subsequently, in S520, a divided feature amount is extracted from the divided data. In the eyelet division, feature quantities are extracted from the divided data obtained by dividing the cluster in a plurality of directions. In this process, not only the divided feature amount but also other feature amounts, for example, histograms of all point groups included in the cluster may be extracted as feature amounts. These methods are effective in a configuration for detecting a feature amount of a vertically long cluster.

  When both vertical direction division and horizontal direction division are performed in the point group division processing, respective divided feature amounts are obtained for the divided data obtained by the respective division methods. A divided feature value obtained from the divided data divided in the vertical direction is defined as a vertical feature, and a divided feature value obtained from the divided data divided in the horizontal direction is defined as a horizontal feature.

  Here, in the process of S520, the feature amount of the reflection intensity is also calculated for each divided data. That is, the degree of variation in reflection intensity is obtained as one of the feature amounts. The “degree of reflection intensity variation” may be a value such as the dispersion or standard deviation of the reflection intensity, or the average intensity with respect to the distance. When the reflection intensity dispersion is used, as shown in FIG. 9, the reflection intensity dispersion values v1 to vN for each divided data are used when the object is identified as the feature quantity f.

Note that S530 is performed only when both the vertical and horizontal features are obtained in the processing of S520. In this process, the vertical feature and the horizontal feature are integrated.
That is, as shown in FIG. 10, the vertical division and the horizontal division are separately performed on the cluster so that the object recognition can be performed using both of the feature amounts obtained by these processes. To do. Note that as the integration method in S530, any method such as a weighted average can be used.

  Subsequently, the object type is identified (S200, S210, S220). In this process, the object type is identified by a well-known identification method using the feature amount obtained so far. For example, discriminators such as pattern matching, SVM (support vector machine), and Ababoost (Adaptive Boosting) can be used. By performing this process, it is possible to recognize what type of object is moving to which position and in what direction.

  Subsequently, in S240, it is determined whether or not to issue an alarm. In this process, it is necessary to determine the probability that the vehicle will collide with the object corresponding to the cluster based on the relative speed or relative distance to the object, and to issue an alarm if the probability of collision is greater than or equal to the first threshold value. judge.

  In S250, it is determined whether or not collision avoidance control needs to be performed. In this process, it is determined that it is necessary to perform the avoidance control if the probability that the vehicle and the object collide is equal to or higher than the second threshold set to a value higher than the first threshold.

  Subsequently, in S260, it is determined whether it is determined that an alarm or avoidance control is necessary. If alarm or avoidance control is necessary, in S270, a command for alarm or avoidance control is transmitted to the necessary portions of the display unit 31, the alarm unit 32, and the operation instruction unit 33, and the driving safety process is performed. finish.

If there is no need for warning or avoidance control, the traveling safety process is terminated.
[1-3. effect]
According to the first embodiment described in detail above, the following effects can be obtained.

  (1a) In the traveling safety device 1 described above, the control unit 10 acquires point cloud data that is a set of distance measuring point data, and recognizes the distribution state of the distance measuring point data constituting the point cloud data. Further, an orientation in a virtual plane representing one or a plurality of virtual planes that divide an area including point cloud data into a plurality of areas according to a distribution state is set, and a plurality of point cloud data is set in the virtual plane. Divide into

  And the control part 10 calculates | requires the feature-value which represents the characteristic of this division | segmentation data numerically for every division | segmentation data showing the point group data divided | segmented on the virtual plane, and based on the feature-value for every division | segmentation data, a point cloud The object represented by the data is recognized (S200, S210, S220).

  According to such a travel safety device 1, since the orientation of the virtual plane for dividing the point cloud data is set according to the distribution state of the distance measuring point data, the point cloud data is appropriately divided according to the object shape. The divided data obtained by dividing the point cloud data can be obtained in an appropriate state. Therefore, a detailed feature description, that is, a feature amount can be obtained for each divided data. Therefore, the recognition accuracy of the object can be improved. Specifically, when a vertically long object such as an electric pole is divided in the horizontal direction, the number of distance measurement points included in the divided data is reduced, or inappropriate parts are cut out (for example, if the electric pole is used, only the part of the scaffold that has jumped out is cut out) Etc.) and the feature amount may not be obtained appropriately, but according to the configuration of the present embodiment, the feature amount can be obtained appropriately.

In the present invention, the “direction of the virtual plane” indicates the magnitude and direction of the inclination of the plane with respect to a reference plane such as a horizontal plane.
(1b) Further, in the travel safety device 1, the control unit 10 recognizes, as the distribution state, an aspect ratio that represents a ratio between a vertical length of an object indicated by the point cloud data and a horizontal length orthogonal to the vertical direction. To do. When the aspect ratio indicates a long vertical length with respect to the horizontal length, the orientation of the virtual plane is set parallel to the horizontal plane, and the aspect ratio is vertical with respect to the horizontal length. When the horizontal length indicates a short landscape, the direction of the virtual plane is set along the depth direction and parallel to the vertical direction.

  According to such a traveling safety device 1, since the object indicated by the point cloud data is divided into a plurality of planes perpendicular to the direction having a long aspect ratio, the ranging point data is distributed in a balanced manner in the divided areas. Therefore, the feature amount for each divided data can be calculated correctly, and the object identification accuracy can be improved.

  (1c) Further, in the travel safety device 1 described above, the control unit 10 determines the ratio of the vertical length of the object indicated by the point cloud data and the horizontal length orthogonal to the vertical direction as the distribution state. Recognize If the aspect ratio is within the range indicating that the aspect ratios are substantially equal, the number of distance measuring point data constituting the point cloud data is compared with a preset threshold value, and the number of distance measuring point data is equal to or greater than the threshold value. In this case, the point cloud data is divided into a plurality of pieces using both the direction of the virtual plane parallel to the horizontal plane and the direction of the virtual plane to be set along the depth direction and parallel to the vertical direction.

  According to such a travel safety device 1, when the number of distance measuring point data is sufficiently large and is equal to or greater than the threshold value, the point cloud data is divided on the respective virtual planes along the vertical direction and the horizontal direction. The feature amount can be easily recognized.

  (1d) In the traveling safety device 1 described above, when the number of distance measuring point data constituting the point cloud data is less than the threshold, the control unit 10 divides the point cloud data with the direction of the virtual plane parallel to the horizontal plane. Further, the point cloud data is divided such that the direction of the virtual plane is along the depth direction and parallel to the vertical direction. Then, a feature amount is obtained for each divided data divided by virtual planes in different directions, and an object is recognized based on the feature amount.

  According to such a travel safety device 1, when the number of distance measuring point data is not sufficient and is less than the threshold value, the point cloud data is divided on the respective virtual planes along the vertical direction and the horizontal direction. Even if the number of distance measuring point data is not sufficient, the feature amount can be easily recognized.

(1f) Further, in the travel safety device 1 described above, the control unit 10 obtains the degree of variation in the reflection intensity as one of the feature amounts because the distance measurement point data also includes information on the reflection intensity.
According to such a traveling safety device 1, since the degree of variation in reflection intensity is one of the feature quantities, for example, when identifying a tree and a person, or identifying a person and a utility pole, If the degree of reflection intensity variation for a plurality of objects having a difference in reflection intensity variation is learned as a feature amount using SVM or the like, the plurality of objects can be well identified based on the degree of reflection intensity variation. Can do. The “degree of reflection intensity variation” may be a value such as the dispersion or standard deviation of the reflection intensity, or the average intensity with respect to the distance.

[2. Other Embodiments]
As mentioned above, although the form for implementing this invention was demonstrated, this invention is not limited to the above-mentioned embodiment, It can implement in various deformation | transformation.

  (2a) In the above embodiment, the present invention is applied to a vehicle, but the present invention is not limited to this. For example, the present invention may be applied to a fixed object such as a utility pole or a traffic light.

  (2b) The functions of one constituent element in the above embodiment may be distributed as a plurality of constituent elements, or the functions of a plurality of constituent elements may be integrated into one constituent element. Moreover, you may abbreviate | omit a part of structure of the said embodiment. Further, at least a part of the configuration of the above embodiment may be added to or replaced with the configuration of the other embodiment. In addition, all the aspects included in the technical idea specified only by the wording described in the claim are embodiment of this invention.

  (2c) In addition to the above-described travel safety device 1, a system including the travel safety device 1 as a constituent element, a program for causing a computer to function as the travel safety device 1, a non-transitive semiconductor memory or the like in which the program is recorded The present invention can also be realized in various forms such as an actual recording medium and a driving safety method.

[3. Relationship between Configuration of Present Invention and Configuration of Embodiment]
In the above embodiment, the travel safety device 1 corresponds to the object recognition device referred to in the present invention. In addition, the processing of S110 and S310 among the processing executed by the control unit 10 in the above embodiment corresponds to the point cloud acquisition unit referred to in the present invention, and the processing of S200, S210 and S220 in the above embodiment refers to the present invention. It corresponds to an object recognition unit.

  In the above embodiment, the processes of S320, S330, S350, and S410 correspond to the distribution recognition unit in the present invention. In the above embodiment, the processes of S340, S360, S420, S430, and S440 are the point group division in the present invention. It corresponds to the part.

  DESCRIPTION OF SYMBOLS 1 ... Travel safety device, 10 ... Control part, 11 ... CPU, 12 ... Memory, 21 ... Laser radar, 22 ... Light emission part, 23 ... Light receiving part, 31 ... Display part, 32 ... Alarm part, 33 ... Operation instruction part

Claims (5)

An object recognition device for recognizing an object based on ranging point data,
A point cloud acquisition unit (S110, S310) for acquiring point cloud data that is a set of ranging point data;
A distribution recognition unit (S320, S330, S350, S410) for recognizing the distribution state of the distance measuring point data constituting the point cloud data;
An orientation in a virtual plane representing one or a plurality of virtual planes that divide the area including the point cloud data into a plurality of areas according to the distribution state is set, and the point cloud data is set on the virtual plane. A point group dividing unit (S340, S360, S420, S430, S440) for dividing into a plurality of parts;
For each piece of divided data representing the point cloud data divided on the virtual plane, a feature amount representing the feature of the divided data as a numerical value is obtained, and the object represented by the point cloud data based on the feature amount for each divided data An object recognition unit (S200, S210, S220) for recognizing
Equipped with a,
The distribution recognition unit recognizes an aspect ratio representing a ratio between a vertical length of an object indicated by the point cloud data and a horizontal length orthogonal to the vertical direction as the distribution state,
The point group dividing unit sets the orientation of the virtual plane parallel to a horizontal plane when the aspect ratio indicates a length that is long in the vertical direction with respect to the length in the horizontal direction, and the aspect ratio is When the length in the vertical direction is short with respect to the length in the horizontal direction, the direction of the virtual plane is set along the depth direction and parallel to the vertical direction.
Object recognition device. The object recognition device according to claim 1 ,
The distribution recognition unit recognizes an aspect ratio representing a ratio between a vertical length of an object indicated by the point cloud data and a horizontal length orthogonal to the vertical direction as the distribution state,
When the point group dividing unit is within a range indicating that the aspect ratio is substantially equal, the point group dividing unit compares the number of ranging point data constituting the point group data with a preset threshold value, and the ranging point When the number of data is equal to or greater than a threshold, both the direction of the virtual plane is parallel to the horizontal plane and the direction of the virtual plane to be set along the depth direction and parallel to the vertical direction are used. An object recognition apparatus that divides the point cloud data into a plurality of pieces. An object recognition device for recognizing an object based on ranging point data,
A point cloud acquisition unit (S110, S310) for acquiring point cloud data that is a set of ranging point data;
A distribution recognition unit (S320, S330, S350, S410) for recognizing the distribution state of the distance measuring point data constituting the point cloud data;
An orientation in a virtual plane representing one or a plurality of virtual planes that divide the area including the point cloud data into a plurality of areas according to the distribution state is set, and the point cloud data is set on the virtual plane. A point group dividing unit (S340, S360, S420, S430, S440) for dividing into a plurality of parts;
For each piece of divided data representing the point cloud data divided on the virtual plane, a feature amount representing the feature of the divided data as a numerical value is obtained, and the object represented by the point cloud data based on the feature amount for each divided data An object recognition unit (S200, S210, S220) for recognizing
Equipped with a,
The distribution recognition unit recognizes an aspect ratio representing a ratio between a vertical length of an object indicated by the point cloud data and a horizontal length orthogonal to the vertical direction as the distribution state,
When the point group dividing unit is within a range indicating that the aspect ratio is substantially equal, the point group dividing unit compares the number of ranging point data constituting the point group data with a preset threshold value, and the ranging point When the number of data is equal to or greater than a threshold, both the direction of the virtual plane is parallel to the horizontal plane and the direction of the virtual plane to be set along the depth direction and parallel to the vertical direction are used. Dividing the point cloud data into a plurality of pieces
Object recognition device. The object recognition device according to claim 2 or 3,
The point group dividing unit divides the point group data by setting the direction of the virtual plane parallel to a horizontal plane when the number of ranging point data constituting the point group data is less than the threshold, and the virtual Dividing the point cloud data with the plane direction along the depth direction and parallel to the vertical direction,
The object recognition unit obtains the feature amount for each divided data divided by virtual planes in different directions, and recognizes the object based on the feature amount. The object recognition device according to any one of claims 1 to 4, wherein:
The distance measuring point data also includes reflection intensity information,
The object recognition unit obtains a degree of variation in the reflection intensity as one of feature quantities.