JP2018185785A - Difference detection device and difference detection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a difference detection device which reduces annoyance felt by a user when setting parameters. A difference detection device (10) is a measurement box cell group in which a measurement unit 110 that measures the positions of a plurality of points in a local space and a measurement box cell group that is three-dimensionally arranged in association with the local space from the measurement results of the measurement unit 110. An extraction unit 120 that extracts attribute information for each measurement box cell, and a map management unit 130 that manages attribute information for each map box cell in a map box cell group arranged three-dimensionally in association with a space represented by a map. The parameter setting unit 150 that sets a numerical value in the parameter according to the operation mode set by the user, and the attribute information about the measurement box cell and the position corresponding to the measurement box cell by using the numerical value set in the parameter. The map box cell is provided with a difference detection unit 140 for detecting the presence or absence of a difference from the attribute information of the box cell, and an output unit 160 for outputting the detection result of the difference detection unit 140. [Selection diagram] Fig. 2

Description

  The present disclosure relates to a difference detection device and a difference detection method for detecting a difference in space.

  Conventionally, a difference detection device that detects a difference in space is known (see, for example, Patent Document 1).

JP 2017-16474 A

  When using a conventional difference detection device, it is necessary to set the parameters used in the difference detection performed by the difference detection device to an appropriate numerical value according to the environment in which the difference detection device is used before using the difference detection device. There is. A user who uses a conventional difference detection device may feel annoying the setting of the parameters.

  Therefore, an object of the present disclosure is to provide a difference detection device and a difference detection method that can reduce the annoyance felt by the user when setting parameters used in the difference detection device, as compared with the related art.

  The difference detection device according to the present disclosure is a difference detection device that detects a difference in space, and includes a measurement unit, an extraction unit, a map management unit, a parameter setting unit, a difference detection unit, and an output unit. The measurement unit measures the position of each of a plurality of points in the surrounding local space. The extraction unit extracts attribute information about each measurement voxel in the voxel group arranged in a three-dimensional manner in association with the local space from the measurement result by the measurement unit. The map management unit manages attribute information about each map voxel in the map voxel group arranged in a three-dimensional manner in association with the space represented by the map. The parameter setting unit sets a numerical value for the parameter according to the operation mode set by the user. The difference detection unit detects the presence / absence of a difference between the attribute information about the measurement voxel and the attribute information about the map voxel at the position corresponding to the measurement voxel using a numerical value set in the parameter. The output unit outputs the detection result of the difference detection unit.

  The difference detection method in the present disclosure is a difference detection method for detecting a difference in space, and includes a measurement step, an extraction step, a reading step, a parameter setting step, a difference detection step, and an output step. . In the measurement step, the position of each of a plurality of points in the surrounding local space is measured. In the extraction step, attribute information about each measurement voxel in the measurement voxel group arranged in a three-dimensional manner in association with the local space is extracted from the measurement result in the measurement step. In the reading step, the attribute information about each map voxel in the map voxel group arranged in a three-dimensional manner in association with the space represented by the map is read. In the parameter setting step, a numerical value is set for the parameter according to the operation mode of the difference detection device set by the user of the difference detection device. In the difference detection step, the presence or absence of a difference between the attribute information about the measurement voxel and the attribute information about the map voxel at the position corresponding to the measurement voxel is detected using a numerical value set in the parameter. In the output step, the detection result in the difference detection step is output.

  According to the difference detection device and the difference detection method of the present disclosure, the user can set parameters used in difference detection by setting the operation mode, and the user feels troublesome when setting the parameters. This can be reduced as compared with the prior art.

FIG. 1 is a perspective view schematically showing an example of a scene in which the difference detection apparatus according to Embodiment 1 is used. FIG. 2 is a block diagram illustrating a functional configuration of the difference detection apparatus according to the first embodiment. FIG. 3 is a conceptual diagram showing a process of extracting measurement voxel data. FIG. 4 is a diagram illustrating an example of measurement point data. FIG. 5 is a diagram illustrating an example of measurement voxel data. FIG. 6 is a flowchart illustrating an example of attribute information extraction processing. FIG. 7 is a flowchart illustrating an example of the difference detection process. FIG. 8 is a flowchart illustrating an example of the voting process in the difference detection process. FIG. 9 is a conceptual diagram showing how voxel data is classified into one of a spherical shape, a sheet shape, and a line shape. FIG. 10 is a diagram illustrating an example of weighting in the voting process. FIG. 11 is a diagram illustrating an example of calculation of the approximation βi in the normal direction indicated by the third principal component of the covariance matrix. FIG. 12A is a data configuration diagram of an example of a parameter setting table set in accordance with operation modes having different moving speeds. FIG. 12B is a data configuration diagram of an example of a parameter setting table set in accordance with operation modes having different measurement cycles. FIG. 12C is a data configuration diagram of an example of a parameter setting table set in accordance with operation modes with different usage locations. FIG. 12D is a data configuration diagram of an example of a parameter setting table in which the distance measuring apparatus is set according to different operation modes. FIG. 13 is a conceptual diagram showing that the number of measurements and the number of votes for the object 1a change depending on the moving speed of the difference detection apparatus according to the first embodiment. FIG. 14 is a flowchart illustrating an example of the setting process. FIG. 15 is a block diagram illustrating a functional configuration of the difference detection apparatus according to the second embodiment. FIG. 16 is a data configuration diagram of an example of a parameter setting table for setting parameters according to the voxel size. FIG. 17 is a conceptual diagram exemplarily showing an object assumed as a cause of the detected difference. FIG. 18 is a conceptual diagram exemplarily showing a correspondence relationship between measurement voxel data and map voxel data.

  Hereinafter, embodiments will be described in detail. Note that each of the embodiments described below shows a preferred specific example of the present disclosure. Numerical values, shapes, materials, components, arrangement positions and connection forms of components, steps, order of steps, and the like shown in the following embodiments are merely examples, and are not intended to limit the present disclosure. The present disclosure is limited only by the claims. Therefore, among the constituent elements in the following embodiments, constituent elements that are not described in the independent claims of the present disclosure are not necessarily required to achieve the problems of the present disclosure, but constitute more preferable embodiments. It is explained as a thing.

(Embodiment 1)
Here, the difference detection apparatus 10 which detects the difference in the space around the own apparatus is demonstrated, referring FIGS. 1-14.

[1-1. Constitution]
FIG. 1 is a perspective view schematically showing an example of a scene in which the difference detection device 10 is used.

  As illustrated in FIG. 1, as an example, the difference detection device 10 is mounted on a moving body 20 (for example, a mobile robot), and detects a difference in a space around the own device. The difference detection device 10 moves in a space where the object 1a and the object 1b exist as the moving body 20 moves.

  The difference detection device 10 includes a computer including a memory, a processor, a communication interface, and the like. And the difference detection apparatus 10 implement | achieves various functions, when a processor runs the program memorize | stored in memory.

  FIG. 2 is a block diagram illustrating a functional configuration of the difference detection apparatus 10.

  As shown in FIG. 2, the difference detection device 10 includes a measurement unit 110, an extraction unit 120, a map management unit 130, a difference detection unit 140, a parameter setting unit 150, and an output unit 160. Is done.

[1-1-1. Measuring unit 110]
The measurement unit 110 measures the position of each of a plurality of points (that is, measurement points) in the surrounding local space, and acquires measurement point data that is a measurement result by measurement.

  The measurement unit 110 includes, for example, a distance measuring device, and is realized by a processor that executes a program stored in a memory controlling the distance measuring device.

  Here, as an example, the measurement unit 110 will be described as including a stereo camera and a depth sensor as a distance measuring device.

  A stereo camera is a camera that measures the distance to a target in units of pixels by simultaneously shooting the target from a plurality of different directions.

  A depth sensor is a sensor that measures the distance to a target in units of pixels by irradiating the target with laser light such as infrared rays and measuring the time until the reflected light is received (TOF: Time Of Flight). .

  For example, the measurement unit 110 may use either a stereo camera or a depth sensor as a distance measuring device in accordance with the operation mode of the difference detection device 10.

  Further, for example, the measurement unit 110 may perform the above measurement every certain period.

[1-1-2. Extractor 120]
The extraction unit 120 extracts attribute information about each voxel in the voxel group arranged in a three-dimensional manner in association with the local space around the measurement unit 110 from the measurement result (observation point data) by the measurement unit 110.

  For example, the extraction unit 120 is realized by a processor that executes a program stored in a memory.

  The measurement point data obtained by measuring the local space includes a position represented by coordinates in the three-dimensional coordinate system determined by the measurement unit 110. The extraction unit 120 extracts data (voxel data) for each voxel of the voxel group obtained by dividing the local space into rectangular parallelepiped voxels along the three-dimensional coordinate system from the measurement point data related to the measurement point group. . The voxel data includes position information in the three-dimensional coordinate system. The voxel data extracted from the measurement point data by the extraction unit 120 is referred to as measurement voxel data, and the voxel that will have the measurement voxel data is referred to as a measurement voxel.

  In the extraction (extraction process) of the measurement voxel data for the measurement voxel by the extraction unit 120, the measurement voxel is compressed from the measurement point data group for a plurality of measurement points included in the measurement voxel in terms of information amount. The measured voxel data is calculated. This compression of the amount of information corresponds to the problem that the amount of information increases more than the two-dimensional plane in the three-dimensional space, so that comparing the measurement points as they are increases the amount of calculation and makes real-time position estimation difficult. It becomes means to do.

  The extraction of the measurement voxel data by the extraction unit 120 is performed, for example, by determining the position information of the measurement voxel for each measurement voxel so as to indicate the average value of the positions for each point in the measurement voxel measured by the measurement unit 110. Is done. That is, the extraction unit 120 averages the positions in the measurement point data (positions represented by three-dimensional coordinates x, y, and z) for each measurement point included in the measurement voxel, and measures the voxel for the measurement voxel. Calculate location information in the data.

  The extraction of the measurement voxel data by the extraction unit 120 is performed, for example, by calculating a 3 × 3 covariance matrix of the point group in the voxel for each measurement voxel. That is, the extraction unit 120 calculates a three-dimensional covariance from the position (the position represented by the three-dimensional coordinates x, y, z) in the measurement point data for each measurement point included in the measurement voxel, A covariance matrix that is attribute information about the measurement voxel is calculated.

  FIG. 3 is a conceptual diagram showing a process of extracting measurement voxel data for measurement voxels based on measurement point data of measurement points in space. In FIG. 3, each measurement point data (measurement point data 201a and the like) is represented by an arrangement associated with the position of the measurement point, and the measurement voxel data (measurement voxel data 203a and the like) possessed by the measurement voxel is measured. This is represented by an arrangement associated with the position indicated by the position information in the voxel data. The measurement unit 110 performs measurement in a local space that is a part of the space around the moving body 20, and obtains measurement point data 201a, 201b, and the like for a plurality of measurement points including the object 1a and the object 1b.

  The extraction unit 120 divides the measured local space into voxels such as measurement voxels 202a and 202b (for example, cubes each having several meters) (that is, a three-dimensional array of voxels in the voxel group) Associate). The measurement voxel 202a has measurement voxel data 203a, and the measurement voxel 202b has measurement voxel data 203b. Measurement point data 201a, 201b and the like corresponding to the measurement points included in the measurement voxel 202a are reflected in the measurement voxel data 203a for the measurement voxel 202a.

  As illustrated in FIG. 4, the measurement point data represents a position by three-dimensional coordinates x, y, z, and a color at the position by an RGB value. FIG. 4 shows measurement point data for one measurement point for each row.

  FIG. 5 is a diagram showing an example of measurement voxel data, and shows measurement voxel data for one measurement voxel for each row. Measurement voxel data as shown in FIG. 5 is temporarily stored in a storage medium such as a memory. As shown in FIG. 5, the measurement voxel data includes an index indicating the arrangement of the corresponding measurement voxel in the three-dimensional array of measurement voxels in the measurement voxel group associated with the local space, and each measurement in the measurement voxel. It includes position information indicating an average value of the position of the point data and a covariance matrix that is attribute information of the measurement voxel. For example, the position information in the measurement voxel data 203a in FIG. 3 indicates the average value of the positions of measurement point data (measurement point data 201a, 201b, etc.) corresponding to each measurement point included in the measurement voxel 202a, and the measurement voxel data 203a. The attribute information in indicates a 3 × 3 covariance matrix of the measurement point group included in the measurement voxel 202a.

[1-1-3. Map management unit 130]
The map management unit 130 associates with a space represented by a map (that is, a map that defines a three-dimensional coordinate system representing a space), and voxel data for each voxel in a voxel group arranged in a three-dimensional manner, that is, a position It has a function of managing information and attribute information (specifically, a covariance matrix). As an example, the map management unit 130 is realized by a processor that executes a program stored in a memory.

  For example, the map management unit 130 may have a function of managing voxel data for a plurality of voxel groups having different voxel sizes.

  Voxel data managed by the map management unit 130 is referred to as map voxel data, and voxels that have map voxel data are referred to as map voxels. Further, measurement voxels and map voxels may be collectively referred to as voxels, and measurement voxel data and map voxel data may be collectively referred to as voxel data. The map management unit 130 manages map information including map voxel data (position information and attribute information) for each map voxel in a map voxel group formed by dividing a space.

  The management of the map information by the map management unit 130 is to make the map information usable. Specifically, the map information is stored in a storage medium such as a memory, or the map information is obtained from an external device or the like. Is to get. The map information is, for example, information generated in advance by measuring a space according to a predetermined three-dimensional coordinate system. The map information is generated, for example, by measuring the position of each measurement point in space and extracting map voxel data (position information and attribute information) for each map voxel based on the measurement result. The extraction of the map voxel data based on the measurement result may be performed by a method similar to the method in which the extraction unit 120 extracts the measurement voxel data for the measurement voxel from the measurement point data, for example.

  In the present embodiment, it is assumed that the data structure of the map voxel data is the same as that of the measurement voxel data shown in FIG. That is, the map voxel data includes an index (index) indicating the arrangement of the map voxel and position information indicating an average value of the measurement result of the position of each point (that is, each location) in a part of the space corresponding to the map voxel. , And attribute information indicating a covariance matrix for each point (that is, each location) in a part of the space corresponding to the map voxel.

[1-1-4. Difference detection unit 140]
The difference detection unit 140 uses a parameter set by the parameter setting unit 150 described later to determine whether there is a difference between the attribute information about each measurement voxel and the attribute information about the map voxel at the position corresponding to each measurement voxel Is detected.

  As an example, the difference detection unit 140 is realized by a processor that executes a program stored in a memory.

  For example, in the detection of the presence or absence of the difference, the difference detection unit 140 detects that there is a difference when the map voxel at the position corresponding to the measurement voxel having significant attribute information does not have significant attribute information. It is good. Thereby, the difference detection apparatus 10 can detect the presence or absence of a difference with a relatively small processing amount.

  In addition, the difference detection unit 140, for example, in the detection of the presence or absence of the difference, only when the map voxel located in the vicinity of the position corresponding to the target measurement voxel does not have significant attribute information, The detection that there is a difference may be performed. Thereby, the difference detection apparatus 10 comes to be able to detect the presence or absence of a difference more accurately.

  Further, the difference detection unit 140, for example, in the detection of the presence or absence of the difference, attribute information indicating that the map voxel at the position corresponding to the measurement voxel having the attribute information (covariance matrix) indicating the shape type indicates the shape type. In the case of having (covariance matrix), when the shape type indicated by the attribute information of the measurement voxel is different from the shape type indicated by the attribute information of the map voxel, it is detected that there is a difference. Also good. Thereby, the difference detection apparatus 10 comes to be able to detect the presence or absence of a difference more accurately.

  Further, the difference detection unit 140 detects, for example, whether or not the map voxel at the position corresponding to the measurement voxel having the attribute information (covariance matrix) indicating the normal direction of the plane or the line segment is the plane or When the attribute information (covariance matrix) indicating the normal direction of the line segment is included, the degree of approximation between the normal direction indicated by the attribute information of the measurement voxel and the normal direction indicated by the attribute information of the map voxel is predetermined. When the value is smaller than the value, it may be detected that there is a difference. Thereby, the difference detection apparatus 10 comes to be able to detect the presence or absence of a difference more accurately.

  In addition, for example, the measurement unit 110 performs position measurement for each of a plurality of points in the surrounding local space N (N is an integer of 2 or more) times, and the extraction unit 120 performs each measurement by the measurement unit 110 for each measurement. You may extract the attribute information about a measurement voxel. In such a case, each time the extraction unit 120 performs the above extraction, the difference detection unit 140 performs a difference when the map voxel at the position corresponding to the measurement voxel having significant attribute information does not have significant attribute information. The vote concerning the presence or absence of the difference is performed so that a positive vote is made. When the voting is repeated N times, the difference detection unit 140 may detect the presence / absence of the difference based on the results of the N votings. Thereby, the difference detection apparatus 10 can detect the presence or absence of the difference while being mounted on the moving body 20 and moving.

  In addition, for example, in the voting, the difference detection unit 140 determines that the difference exists only when the map voxel located in the vicinity of the position corresponding to the target measurement voxel does not have significant attribute information. A positive vote may be made. Thereby, the difference detection apparatus 10 can detect the presence or absence of the difference with higher accuracy while being mounted on the moving body 20 and moving.

  In addition, for example, the measurement unit 110 performs position measurement for each of a plurality of points in the surrounding local space N (N is an integer of 2 or more) times, and the extraction unit 120 performs each measurement by the measurement unit 110 for each measurement. You may extract the attribute information about a measurement voxel. In such a case, each time the extraction unit 120 performs the extraction, the difference detection unit 140 indicates that the map voxel at the position corresponding to the measurement voxel having attribute information (covariance matrix) indicating the shape type indicates the shape type. When there is attribute information (covariance matrix), when the shape type indicated by the attribute information of the measurement voxel and the shape type indicated by the attribute information of the map voxel are different from each other, an affirmative with a difference The voting concerning the presence / absence of the difference is performed so that the voting is performed. When the voting is repeated N times, the difference detection unit 140 may detect the presence / absence of the difference based on the results of the N votings. Thereby, the difference detection apparatus 10 can detect the presence or absence of the difference with higher accuracy while being mounted on the moving body 20 and moving.

  In addition, for example, the measurement unit 110 performs position measurement for each of a plurality of points in the surrounding local space N (N is an integer of 2 or more) times, and the extraction unit 120 performs each measurement by the measurement unit 110 for each measurement. You may extract the attribute information about a measurement voxel. In such a case, each time the extraction unit 120 performs the above extraction, the difference detection unit 140 determines that the map voxel at the position corresponding to the measurement voxel having attribute information (covariance matrix) indicating the normal direction of the plane or line segment is When having attribute information (covariance matrix) indicating the normal direction of a plane or line segment, the degree of approximation between the normal direction indicated by the attribute information of the measurement voxel and the normal direction indicated by the attribute information of the map voxel When the value is smaller than the predetermined value, the vote relating to the presence or absence of the difference is performed so that a positive vote is made with a difference. When the voting is repeated N times, the difference detection unit 140 may detect the presence / absence of the difference based on the results of the N votings. Thereby, the difference detection apparatus 10 can detect the presence or absence of the difference with higher accuracy while being mounted on the moving body 20 and moving.

[1-1-5. Parameter setting unit 150]
The parameter setting unit 150 sets numerical values for one or more parameters according to the operation mode of the difference detection device 10 set by the user of the difference detection device 10.

  The parameter setting unit 150 includes, for example, an input interface for accepting setting of an operation mode of the difference detection device 10 by a user, and is realized by a processor that executes a program stored in a memory controlling the input interface. .

  Examples of the input interface include a keyboard, a mouse, a touch panel, and a push button switch.

  The parameter whose numerical value is set by the parameter setting unit 150 may be, for example, a voting threshold used when the difference detection unit 140 detects the presence or absence of a difference based on the result of voting.

[1-1-6. Output unit 160]
The output unit 160 outputs the detection result of the difference detection unit 140.

  The output unit 160 includes, for example, an output interface for outputting a display result to a user, and is realized by a processor that executes a program stored in a memory controlling the output interface.

  Examples of the output interface include an output port, a display, a speaker, and the like.

  In addition, when a display device exists outside, the output unit 160 may perform output so that an image indicating the detection result is displayed on the display device. Thereby, the difference detection apparatus 10 can notify a user of a detection result using an image.

  The operation performed by the difference detection apparatus 10 having the above configuration will be described below with reference to the drawings.

[1-2. Operation]
[1-2-1. Operation of the extraction unit]
The extraction unit 120 performs an extraction process for extracting measurement voxel data as a characteristic operation.

  FIG. 6 is a flowchart illustrating an example of extraction processing by the extraction unit 120. The extraction process will be described with reference to FIG.

  The extraction process is started when, for example, a user using the difference detection device 10 performs an operation for starting the extraction process on the difference detection device 10.

  When the extraction process is started, the extraction unit 120 determines a voxel size for the measurement voxel (step S11). For example, the lengths in the x, y, and z directions in the three-dimensional coordinate system (xyz three-dimensional coordinate system) related to the measurement are determined as the voxel size. As a result, the measurement voxel is determined to be, for example, a cube having a side of 1 meter. In addition, the extraction unit 120 adds an offset x, which is an addition value used to convert a value (x, y, z coordinate value) within a measurement range in a three-dimensional coordinate system related to measurement of a measurement voxel into an index. The y and z coordinates are determined (step S12). Note that the voxel size of the measurement voxel does not necessarily need to match the voxel size of the map voxel.

  Subsequently, the extraction unit 120 acquires measurement point data for one of the measurement points corresponding to the surrounding local space obtained by measurement by the measurement unit 110 as a processing target (step S13). Then, the measurement voxel corresponding to the acquired measurement point data (measurement voxel including the position related to the measurement point data) is specified, and the index of the measurement voxel is calculated (step S14). For example, the index (index) of the measurement voxel is calculated by the following equations 1 to 3 using the coordinate value of the measurement point data. This index is a number indicating the arrangement in the three-dimensional array for each measurement voxel of the measurement voxel group arranged three-dimensionally in the local space.

x (index) = (x coordinate of measurement point + x coordinate of offset) / voxel size (Formula 1)
y (index) = (y coordinate of measurement point + y coordinate of offset) / voxel size (Formula 2)
z (index) = (z coordinate of measurement point + z coordinate of offset) / voxel size (Formula 3)
Next, the extraction unit 120 uses the position in the measurement point data acquired in step S13 to update the position information (average value of x, y, z coordinates) for the measurement voxel specified in step S14 (step S15). . Further, the extraction unit 120 calculates a three-dimensional covariance using the position in the measurement point data acquired in step S13, and updates the attribute information about the measurement voxel specified in step S14 (step S16). Through the processing in step S15 and step S16, measurement voxel data (position information and attribute information) as shown in FIG. 5 temporarily stored in a storage medium such as a memory is updated. Then, the processing in steps S13 to S16 is repeated until the measurement point data not yet processed is sequentially processed and all the measurement point data is processed, and the processing is completed (step S17).

  Each time the measurement unit 110 obtains a measurement point data group for each measurement point in the surrounding local space (for example, each measurement point such as the entire circumference in the horizontal direction), this extraction process (steps S11 to S17), or , Some of the processes in steps S13 to S17 may be performed. When the extraction unit 120 extracts measurement voxel data (position information and attribute information) of each measurement voxel after a certain amount of measurement point data group is obtained in this way, the measurement points located in each measurement voxel The average value of the positions in each measurement point data for the measurement point data may be the position information for the measurement voxel, and the three-dimensional covariance in each measurement point data may be the attribute information for the measurement voxel. The position of the measurement voxel (for example, the x-coordinate, y-coordinate, and z-coordinate of one vertex corresponding to the minimum position in the measurement voxel) can be specified by the following Expressions 4 to 6 based on the index.

X coordinate of measurement voxel = x (index) × voxel size−x coordinate of offset (Formula 4)
Y coordinate of measurement voxel = y (index) × Voxel size−y coordinate of offset (Formula 5)
Z coordinate of measurement voxel = z (index) × voxel size−z coordinate of offset (Formula 6)
[1-2-2. Operation of difference detection unit]
The difference detection unit 140 performs a difference detection process for detecting whether or not there is a difference between the attribute information about each measurement voxel and the attribute information about the map voxel at the position corresponding to each measurement voxel.

  FIG. 7 is a flowchart illustrating an example of a difference detection process performed by the difference detection unit 140. The difference detection process will be described with reference to FIG.

  Here, it is assumed that the measurement unit 110 acquires measurement point data at a predetermined frame rate, and the extraction unit 120 extracts attribute information for each measurement voxel at the predetermined frame rate.

  The difference detection process is started, for example, when the extraction unit 120 extracts attribute information a predetermined number of times (for example, N (N is an integer of 2 or more) times).

  When the difference detection process is started, the difference detection unit 140 acquires measurement voxel data for one frame that has not yet been selected among the measurement voxel data for N frames extracted by the extraction unit 120 (step S1). S21).

  Then, the difference detection unit 140 compares the measurement voxel of the measurement voxel data with the map voxel located in the same vicinity in the measurement voxel data that is not yet selected among the measurement voxel data for one frame, A voting process for voting the difference value to the map voxel is performed (step S22).

  FIG. 8 is a flowchart illustrating an example of the voting process performed by the difference detection unit 140. Details of the voting process will be described with reference to FIG.

  When the voting process is started, the difference detection unit 140 includes the measurement voxel data of the target measurement voxel, the map voxel having the same index as the target measurement voxel, and the 26 map voxels adjacent to the map voxel. The map voxel data of the following 27 map voxels (hereinafter also referred to as “27 neighboring map voxels”) are compared (step S31).

  In the process of step S31, first, the difference detection unit 140 reads map voxel data of 27 neighboring map voxels. Then, the difference detection unit 140 converts the measurement voxel data of the target measurement voxel and the map voxel data of the 27 neighboring map voxels into eigenvalues λ1, λ2, λ3 (λ1 + λ2 + λ3 = 1, λ3> λ2> λ1 of the covariance matrix. ), Any one of a spherical shape (λ3≈λ2≈λ1 >> 0), a sheet shape (λ3≈λ2 >> λ1≈0), or a linear shape (λ3 >> λ2≈λ1≈0) Classify into: Then, the classified shapes are compared with each other.

  FIG. 9 shows a state where the voxel data is classified into one of a spherical shape, a sheet shape, and a line shape. In the example shown in FIG. 9, the voxel data 302a, 302b, and 302c of the voxel 301 have a spherical shape ((a) in FIG. 9), a sheet shape ((b) in FIG. 9), and a linear shape ((c) in FIG. 9). ) Are conceptually shown as being classified.

  Here, the map voxels may overlap. The overlap means that the voxels are arranged so that a part of the voxels overlaps, for example, when the voxels are arranged so as to overlap each other by half a voxel. That is, the map voxel group managed by the map management unit 130 may be overlapped. Overlapping is effective in reducing positional deviation and error between the map point group and the measurement point group due to position estimation deviation and sensor distance measurement error. Thereby, the difference detection apparatus 10 comes to be able to detect the presence or absence of a difference more accurately while moving.

  Next to step S31, the difference detection unit 140 determines whether one or more shapes that are the same as the shape indicated by the measurement voxel data of the target measurement voxel are present in the shape indicated by the map voxel data of the 27 adjacent map voxels. Is determined (step S32).

  If it is not determined in step S32 that one or more of the same shapes exist (step S32: No), the difference detection unit 140 weights the 27 neighboring map voxels and performs a vote with a difference. (Step S36), the voting process is terminated.

  FIG. 10 is a diagram illustrating an example of weighting when performing voting with a difference between the target measurement voxel 401 and the map voxels 402a and 402b in the vicinity thereof.

  The difference detection unit 140 uses, for example, a three-dimensional normal distribution equation with an average of 0 and a variance σ as shown by an equation 403 in FIG. 10 to indicate the index in the x, y, and z directions with reference to the target measurement voxel. The weight f (x, y, z) based on the distance may be calculated.

  When it is determined in step S32 in FIG. 8 that one or more of the same shapes are present (step S32: Yes), the difference detection unit 140 checks whether or not the shape is a spherical shape (step S33). ).

  In the process of step S33, when the shape is a spherical shape (step S33: Yes), the difference detection unit 140 ends the voting process without voting.

  In the process of step S33, when the shape is not a spherical shape (step S33: No), that is, when the shape is a sheet shape or a line shape, the difference detection unit 140 uses the method indicated by the third principal component of the covariance matrix. The degree of approximation βi in the line direction is calculated (step S34).

  FIG. 11 is a diagram illustrating an example of calculation of the degree of approximation βi in the normal direction.

  For example, as illustrated in FIG. 11, the difference detection unit 140 performs the first principal component analysis and the first principal component analysis when the covariance matrix of each of the measurement voxel data 502 a and the map voxel data 502 b corresponding to the voxel 501 is analyzed. From the normal vectors Nmap = (Nmap.x, Nmap.y, Nmap.z) and Ni = (Nix, Niy, Niz) of the normals 503a and 503b with respect to the representative surface composed of two principal components, The degree of approximation βi shown may be calculated.

  When the normality approximation βi is calculated in step S34 of FIG. 8, the difference detection unit 140 has one or more map voxels in which the approximation βi is equal to or greater than a predetermined value (that is, the approximation is high). Is determined (step S35).

  In the process of step S35, when there is a map voxel with a high degree of approximation (step S35: Yes), the difference detection unit 140 ends the voting process without voting.

  In the process of step S35, when there is no map voxel having a high degree of approximation (step S35: No), the difference detection unit 140 weights the 27 neighboring map voxels and performs a vote with a difference (step S36). ) And finish the voting process.

  In FIG. 7, when the voting process (step S22) is completed, the difference detection unit 140 determines whether or not the voting process has been completed for all measurement voxel data for one frame (step S23).

  In the process of step S23, when the voting process has not been completed for all the measurement voxel data for one frame (step S23: No), the difference detection unit 140 returns to the process of step S22 again to return one frame. Until the voting process is completed for all the measurement voxel data of the minute, the processes after step S22 are repeated.

  In the process of step S23, when the voting process is completed for all measurement voxel data for one frame (step S23: Yes), the difference detection unit 140 determines all measurement voxel data for N frames. It is determined whether or not the voting process has been completed (step S24).

  In the process of step S24, when the voting process has not been completed for all measurement voxel data for N frames (step S24: No), the difference detection unit 140 returns to the process of step S21 again. The processes after step S21 are repeated until the voting process is completed for all the measurement voxel data for N frames.

  In the process of step S24, when the voting process has been completed for all measurement voxel data for N frames (step S24: Yes), the difference detection unit 140 includes the parameter setting unit 150 of the map voxels. A map voxel having a voting value larger than the voting threshold value, which is a parameter set numerically by, is specified. Then, the difference detection unit 140 determines that there is a difference in attribute information between the specified map voxel and the corresponding measurement voxel (step S25).

[1-2-3. Operation of parameter setting unit]
For example, the parameter setting unit 150 stores in advance a parameter setting table that associates the operation mode of the difference detection device 10 with a numerical value to be set for a parameter. Then, with reference to the stored parameter setting table, a setting process for setting a numerical value to the parameter is performed in cooperation with the difference detection unit 140.

  FIG. 12A is a data configuration diagram of the parameter setting table 151 set in accordance with operation modes having different moving speeds. As illustrated in FIG. 12A, the parameter setting table 151 associates the operation mode of the difference detection device 10 with the parameter value (numerical value) set to the voting threshold (parameter).

  In this example, as the operation mode of the difference detection device 10, the moving speed of the mobile body 20 on which the difference detection device 10 is mounted (that is, the movement speed of the difference detection device 10) is relatively high (for example, 7 km / h). A high-speed movement mode (an example of the first mode) in a certain case and a low-speed movement mode (an example of the second mode) in which the movement speed of the difference detection device 10 is relatively low (for example, 3 km / h). Contains. The high-speed movement mode is associated with “15” as a relatively low voting threshold, and the low-speed movement mode is associated with “35” as a relatively high voting threshold.

  The parameter setting unit 150 refers to the parameter setting table 151 to be stored. When the operation mode input by the user is the high-speed movement mode, the parameter setting unit 150 sets the voting threshold to “15” and the operation input by the user. When the mode is the low speed movement mode, the voting threshold is set to “35”.

  FIG. 13 is a conceptual diagram showing that the number of measurements and the number of votes for the object 1a (see FIG. 1) change depending on the moving speed of the difference detection device 10.

  As shown in FIG. 13, when the moving speed of the difference detection apparatus 10 is relatively high ((b) in FIG. 13), the number of votes for the object 1a is relatively small. Thereby, the vote value with respect to the object 1a becomes comparatively low. For this reason, when the moving speed of the difference detection apparatus 10 is relatively high, the voting threshold is set to be relatively low.

  On the other hand, when the moving speed of the difference detection device 10 is relatively low ((a) in FIG. 13), the number of votes for the object 1a is relatively large. Thereby, the vote value with respect to the object 1a becomes comparatively high. For this reason, when the moving speed of the difference detection apparatus 10 is relatively low, the voting threshold is set to be relatively high.

  FIG. 12B is a data configuration diagram of the parameter setting table 152 set in accordance with operation modes having different measurement cycles. In this example, the operation mode of the difference detection apparatus 10 includes a low frame rate mode (an example of the first mode) when the frame rate at which the measurement unit 110 acquires measurement point data is relatively low (for example, 20 fps), and measurement. And a high frame rate mode (an example of the second mode) when the frame rate at which the unit 110 acquires measurement point data is relatively high (for example, 60 fps). The low frame rate mode is associated with “10” as a relatively low vote threshold, and the high frame rate mode is associated with “30” as a relatively high vote threshold.

  The parameter setting unit 150 refers to the parameter setting table 152 to be stored. When the operation mode input by the user is the low frame rate mode, the parameter setting unit 150 sets the voting threshold to “10” and is input by the user. When the operation mode is the high frame rate mode, the voting threshold is set to “30”.

  When the frame rate at which the measurement unit 110 acquires measurement point data is relatively low, the number of votes for the object 1a is relatively small. Thereby, the vote value with respect to the object 1a becomes comparatively low. For this reason, when the frame rate at which the measurement unit 110 acquires measurement point data is relatively low, the voting threshold is set to be relatively low.

  On the other hand, when the frame rate at which the measurement unit 110 acquires measurement point data is relatively high, the number of votes for the object 1a is relatively large. Thereby, the vote value with respect to the object 1a becomes comparatively high. For this reason, when the frame rate at which the measurement unit 110 acquires the measurement point data is relatively high, the voting threshold is set to be relatively high.

  FIG. 12C is a data configuration diagram of the parameter setting table 153 set in accordance with operation modes with different usage locations.

  In this example, as an operation mode of the difference detection device 10, an indoor mode in which the difference detection device 10 is used indoors (an example of a first mode) and an outdoor mode in which the difference detection device 10 is used outdoors (second mode). Example). The indoor mode is associated with “10” as a relatively low vote threshold, and the outdoor mode is associated with “30” as a relatively high vote threshold.

  The parameter setting unit 150 refers to the parameter setting table 153 to be stored. When the operation mode input by the user is the indoor mode, the parameter setting unit 150 sets the voting threshold to “10” and the operation mode input by the user. If is the outdoor mode, the voting threshold is set to “30”.

  When the measurement unit 110 uses a stereo camera as a distance measuring device, the amount of the measurement point data group to be acquired varies depending on the amount of texture of the subject due to the nature of the stereo camera.

  In general, when the difference detection device 10 is used indoors, the subject tends to include a uniform wall or floor, and thus is acquired by the measurement unit 110 using a stereo camera. The amount of measurement point data group tends to be relatively small. For this reason, when the difference detection apparatus 10 is used indoors, the voting threshold value is set to be relatively low.

  On the other hand, in general, when the difference detection device 10 is used outdoors, the subjects tend to be diverse. Therefore, the amount of measurement point data group acquired by the measurement unit 110 using a stereo camera is relatively small. It tends to increase. For this reason, when the difference detection apparatus 10 is used outdoors, the voting threshold is set to be relatively high.

  FIG. 12D is a data configuration diagram of the parameter setting table 154 in which the distance measuring apparatus is set according to different operation modes.

  In this example, when the difference detection device 10 is used indoors, as an operation mode of the difference detection device 10, the measurement unit 110 uses a stereo camera (an example of a first sensor) as a distance measuring device. Stereo camera mode (an example of a first mode) and a depth sensor mode (an example of a second mode) for indoor use in which the measurement unit 110 uses a depth sensor (an example of a second sensor) as a distance measuring device. . The stereo camera mode for indoor use is associated with “10” as a relatively low vote threshold, and the depth sensor mode for indoor use is associated with “30” as a relatively high vote threshold. Yes.

  The parameter setting unit 150 refers to the parameter setting table 154 to be stored. When the operation mode input by the user is the stereo camera mode for indoor use, the parameter setting unit 150 sets the voting threshold to “10” and When the input operation mode is the depth sensor mode for indoor use, the voting threshold is set to “30”.

  In general, when the difference detection device 10 is used indoors, when the measurement unit 110 uses a stereo camera as a distance measuring device, the amount of measurement point data groups acquired by the measurement unit 110 is compared as described above. Tend to be less. For this reason, when the measurement unit 110 uses a stereo camera as a distance measuring device, the voting threshold is set to be relatively low.

  On the other hand, generally, when the difference detection device 10 is used indoors, when the measurement unit 110 uses a depth sensor as a distance measuring device, the amount of the measurement point data group acquired by the measurement unit 110 is the difference detection device. 10 tends to be the same as when used outdoors. For this reason, when the measurement unit 110 uses a depth sensor as a distance measuring device, the voting threshold is set to be relatively high.

  FIG. 14 is a flowchart illustrating an example of setting processing by the parameter setting unit 150 and the difference detection unit 140.

  Here, the operation mode of the difference detection apparatus 10 includes a high-speed movement mode and a low-speed movement mode, and the parameter setting unit 150 is described as being stored with a parameter setting table 151. The same applies when the operation mode includes the low frame rate mode and the high frame rate mode, and the parameter setting table 152 is stored in the parameter setting unit 150. The same applies when the operation mode includes an indoor mode and an outdoor mode, and the parameter setting table 153 is stored in the parameter setting unit 150. The same applies when the operation mode includes the stereo camera mode and the depth sensor mode, and the parameter setting table 154 is stored in the parameter setting unit 150.

  The setting process is started when, for example, a user who uses the difference detection apparatus 10 performs an operation for starting the setting process on the difference detection apparatus 10.

  When the setting process is started, the parameter setting unit 150 receives the operation mode of the difference detection device 10 input by the user of the difference detection device 10 (step S41). Then, the parameter setting unit 150 determines whether or not the received operation mode is the high speed movement mode, that is, whether the operation mode is the high speed movement mode or the low speed movement mode (step S42).

  In the process of step S42, when the received operation mode is the high-speed movement mode (step S42: Yes), the parameter setting unit 150 refers to the stored parameter setting table 151 and sets the voting threshold value to “15”. (Step S43). Then, the difference detection unit 140 sets the voting threshold value to “15” and performs a difference detection process to be performed later (step S44).

  In the process of step S42, when the received operation mode is the low speed movement mode (step S42: No), the parameter setting unit 150 refers to the stored parameter setting table 151 and sets the voting threshold value to “35”. (Step S45). Then, the difference detection unit 140 sets the voting threshold value to “35” and performs a difference detection process to be performed later (step S46).

[1-4. Effect]
The difference detection apparatus 10 has, for example, a high-speed movement mode and a low-speed movement mode respectively corresponding to different movement speeds of the moving body as operation modes, and the parameter setting unit 150 is an operation set by the user in the setting. Different values are set for the parameters depending on whether the mode is the high-speed movement mode or the low-speed movement mode. The difference detection unit 140 detects that there is a difference when the relationship between the numerical value of the parameter set by the parameter setting unit 150 and the result of N votes is a predetermined relationship. Thereby, the difference detection apparatus 10 can set different numerical values as parameters in operation modes having different moving speeds.

  Further, the difference detection device 10 has, for example, the measurement unit 110 performs each measurement in N measurements in a predetermined cycle, and has a low frame rate mode and a high frame rate mode with different predetermined cycles as operation modes, The parameter setting unit 150 sets different numerical values for the parameter when the operation mode set by the user is the low frame rate mode and when the operation mode is the high frame rate mode. Even if the difference detection unit 140 detects the presence or absence of a difference, the difference detection unit 140 detects that there is a difference when the relationship between the numerical value of the parameter set by the parameter setting unit and the result of N votes is a predetermined relationship. Good. Thereby, the difference detection apparatus can set different numerical values as parameters in operation modes having different measurement cycles.

  Further, for example, the operation mode includes an indoor mode and an outdoor mode corresponding to a place where the moving object is used, and the parameter setting unit 150 sets the operation mode set by the user in the indoor mode. Different values are set for the parameters in some cases and in the outdoor mode. The difference detection unit 140 detects that there is a difference when the relationship between the numerical value of the parameter set by the parameter setting unit 150 and the result of N votes is a predetermined relationship.

  As a result, the difference detection device 10 can set different values for the parameters in the operation mode according to the place where the moving object is used.

  Further, for example, the measurement unit 110 includes a stereo camera and a depth sensor as a distance measuring device capable of performing position measurement, and the operation mode includes a stereo camera mode that uses a measurement result by the stereo camera and a measurement by the depth sensor. And depth sensor mode using the result. In the extraction, when the operation mode set by the user is the stereo camera mode, the extraction unit 120 extracts attribute information from the measurement result by the stereo camera, and the operation mode set by the user is the depth sensor mode. In some cases, attribute information is extracted from the measurement result of the depth sensor. The parameter setting unit 150 sets different numerical values for the parameter in the setting when the operation mode set by the user is the stereo camera mode and the depth sensor mode. The difference detection unit 140 detects that there is a difference when the relationship between the numerical value of the parameter set by the parameter setting unit 150 and the result of N votes is a predetermined relationship.

  Thus, the difference detection device can set different values for the parameters in the operation modes in which the distance measuring devices to be used are different from each other.

  For example, the stereo camera mode and the depth sensor mode may be operation modes in which the difference detection device is used indoors. As a result, the difference detection device is an operation mode used indoors, and different numerical values can be set as parameters in an operation mode in which different distance measuring devices are used.

  According to the difference detection device 10 configured as described above, the user who uses the difference detection device 10 sets the parameters used in the difference detection performed by the difference detection device 10 by setting the operation mode of the difference detection device 10. Will be able to.

  For this reason, the user who uses the difference detection apparatus 10 can reduce the troublesomeness which a user feels when setting the parameter utilized in the difference detection apparatus 10 conventionally.

(Embodiment 2)
Here, the difference detection apparatus 10A according to the second embodiment in which a part of the functions is changed from the difference detection apparatus 10 according to the first embodiment will be described.

  The difference detection device 10A can change the size of the voxel to be used in accordance with the operation mode set in the own device.

  Hereinafter, the difference detection apparatus 10A will be described focusing on differences from the difference detection apparatus 10 according to the first embodiment.

[2-1. Constitution]
The difference detection device 10A has the same hardware configuration as the difference detection device 10 according to the first embodiment, but a part of the executed program is changed. For this reason, some of the functions realized by the difference detection device 10A are changed from the functions realized by the difference detection device 10 according to the first embodiment.

  In addition to the operation mode of the difference detection device 10 according to the first embodiment, the difference detection device 10A includes a spherical shape detection mode (third mode) in which the shape of the object assumed as the cause of the detected difference is a spherical shape. And a sheet shape / line shape detection mode (an example of a fourth mode) that is a sheet shape or a line shape.

  FIG. 15 is a block diagram illustrating a functional configuration of the difference detection apparatus 10A.

  As shown in FIG. 15, in the difference detection device 10A, the extraction unit 120 is changed to the extraction unit 120A and the map management unit 130 is changed to the map management unit 130A from the difference detection device 10 according to the first embodiment. The difference detection unit 140 is changed to the difference detection unit 140A, and the parameter setting unit 150 is changed to the parameter setting unit 150A.

  150 A of parameter setting parts memorize | store the parameter setting table 155 in addition to the parameter setting table which the parameter setting part 150 which concerns on Embodiment 1 memorize | stores.

  FIG. 16 is a data configuration diagram of a parameter setting table 155 that is an example of a parameter setting table stored in the parameter setting unit 150A.

  As shown in FIG. 16, the parameter setting table 155 associates the operation mode of the difference detection apparatus 10 with the parameter value (numerical value) set for the voxel size (parameter).

  In this example, as an operation mode of the difference detection apparatus 10A, a spherical shape detection mode in which the shape of an object assumed as a cause of a detected difference is a spherical shape, and a sheet shape / line shape detection in which a sheet shape or a linear shape is detected. Includes modes. The spherical shape detection mode is associated with a numerical value (for example, 0) indicating the first size (for example, a cube with each side being 1 meter), and the sheet shape / line shape detection mode is associated with the second size (for example, a cube). , Each side is associated with a numerical value (for example, 1) indicating a cube of 50 centimeters).

  The parameter setting unit 150A refers to the parameter setting table 155 to be stored. If the operation mode input by the user is the spherical shape detection mode, the parameter setting unit 150A sets the voxel size as a parameter to 0 and is input by the user. If the operation mode is the sheet shape / line shape detection mode, 1 is set to the voxel size as a parameter.

  In FIG. 15, the extraction unit 120 </ b> A performs the following operation in addition to the operation performed by the extraction unit 120 according to the first embodiment. That is, when the operation mode set by the user is the spherical shape detection mode, the extraction unit 120A extracts the first type attribute information for each first measurement voxel in the first size voxel group, and the user When the operation mode set by is the sheet shape / line shape detection mode, the second type attribute information about each second measurement voxel in the second size voxel group is extracted.

  The map management unit 130A performs management of attribute information performed by the map management unit 130 according to Embodiment 1 for the first size voxel group and the second size voxel group.

  That is, the map management unit 130A obtains the first type attribute information about each first map voxel in the first size voxel group and the second type attribute information about each second map voxel in the second size voxel group. to manage.

  For example, the map management unit 130A may further manage attribute information about each map voxel in one or more voxel groups having a size different from both the first size and the second size.

  In addition to the operation performed by the difference detection unit 140 according to Embodiment 1, the difference detection unit 140A further performs the following operation.

  That is, when the operation mode set by the user is the spherical shape detection mode (that is, when the value of the voxel size as a parameter is set to 0), the difference detection unit 140A sets each first measurement voxel. The presence / absence of a difference between the first type attribute information about and the first type attribute information about the first map voxel at the position corresponding to the first measurement voxel is detected. When the operation mode set by the user is the sheet shape / line shape detection mode (that is, when the value of voxel size as a parameter is set to 1), the second type for each second measurement voxel The presence / absence of a difference between the attribute information and the second type attribute information about the second map voxel at the position corresponding to the second measurement voxel is detected.

[2-2. Operation]
Hereinafter, operations performed by the difference detection apparatus 10A having the above-described configuration will be described using specific examples.

  FIG. 17 shows a case where the object assumed as the cause of the detected difference is a rectangular parallelepiped 802 including a lot of planes (sheets) (FIG. 17A) and a sphere 803 (FIG. 17B). )) Is a conceptual diagram exemplarily showing.

  FIG. 18 shows the correspondence between the measured voxel data and the map voxel data when the voxel size is the first size, and the correspondence between the measured voxel data and the map voxel data when the voxel size is the second size. It is a conceptual diagram shown as an example.

  In FIG. 18, measurement voxel data 805a is measurement voxel data corresponding to a rectangular parallelepiped 802 of the first measurement voxel 804a in the first size voxel group, and is classified into a sheet shape.

  The measurement voxel data 805b is measurement voxel data corresponding to the sphere 803 of the first measurement voxel 804a in the first size voxel group, and is classified into a spherical shape.

  The map voxel data 805c is map voxel data of the map voxel 804c in the first size voxel group, and is classified into a sheet shape.

  Here, the first map voxel 804c is a map voxel at the same position as the first measurement voxel 804b adjacent to the first measurement voxel 804a.

  The measurement voxel data 805d is measurement voxel data corresponding to the rectangular parallelepiped 802 of the second measurement voxel 804d in the second size voxel group, and is classified into a sheet shape.

  The measurement voxel data 805e is measurement voxel data corresponding to the sphere 803 of the second measurement voxel 804d in the second size voxel group, and is classified into a spherical shape.

  The map voxel data 805f is map voxel data of the second map voxel 804f in the second size voxel group, and is classified into a sheet shape.

  Here, the second map voxel 804f is a map voxel at the same position as the second measurement voxel 804e adjacent to the second measurement voxel 804e.

  In the following, the degree of approximation between the normal direction of the measurement voxel data 805a and the normal direction of the map voxel data 805c is high, and the normal direction of the measurement voxel data 805d and the normal direction of the map voxel data 805f The description will be made assuming that the degree of approximation is high.

  In the voting process of FIG. 8, when the voxel size is the first size and the measured voxel data 805b and the map voxel data 805c are compared (step S31), the shape indicated by the measured voxel data 805b and the map voxel Since the shape indicated by the data 805c is different, there is a high possibility that a negative determination is made in the process of step S32 (step S32: No). For this reason, in the process of step S36, there is a high possibility that the 27 neighborhood map voxels are weighted and voted with a difference.

  Therefore, when the shape of the object assumed as the cause of the detected difference is a spherical shape, the difference detection device 10A can detect the difference using the first size voxel.

  In the voting process, when the voxel size is the first size and the measured voxel data 805a and the map voxel data 805c are compared (step S31), the shape indicated by the measured voxel data 805a and the map voxel data Since both of the shapes indicated by 805c are sheet shapes, there is a high possibility that a positive determination is made in the process of step S32 (step S32: Yes).

  And since the measurement voxel data 805a is a sheet | seat shape, possibility of being negatively determined in the process of step S33 (step S33: No) is high.

  Furthermore, since the degree of approximation between the normal direction of the measurement voxel data 805a and the normal direction of the map voxel data 805c is high, there is a high possibility that a positive determination is made in the process of step S35 (step S35: Yes). .

  For this reason, there is a high possibility that the process of step S36 is not performed and the 27-neighboring map voxels are weighted and a vote with a difference is not performed.

  Therefore, when the shape of the object assumed as the cause of the detected difference is a sheet shape, it is difficult for the difference detection device 10A to detect the difference using the first size voxel.

  On the other hand, in the voting process, when the voxel size is the second size, the map voxel data 805f is not included in the 27 neighboring map voxels corresponding to the measurement voxel data 805d. Is highly likely to be determined (step S32: No). For this reason, in the process of step S36, there is a high possibility that the 27 neighborhood map voxels are weighted and voted with a difference.

  Therefore, when the shape of the object assumed as a cause of the detected difference is a sheet shape, the difference detection device 10A detects the difference using a voxel having a second size smaller than the first size. Can do.

[2-4. Effect]
The operation mode of the difference detection apparatus 10A may include a spherical shape detection mode and a sheet shape / line shape detection mode in which the shape of an object assumed as a cause of the detected difference is different. When the operation mode set by the user in the extraction is the spherical shape detection mode, the extraction unit 120A extracts the first type attribute information about each first measurement voxel in the first size voxel group, When the operation mode set by the user is the sheet shape / line shape detection mode, the second type attribute information for each second measurement voxel in the voxel group of the second size different from the first size is extracted. In the management, the map management unit 130A manages the first type attribute information for each first map voxel in the first size voxel group and the second type attribute information for each second map voxel in the second size voxel group. Manage. In the difference detection, when the operation mode set by the user is the spherical shape detection mode, the difference detection unit 140A includes the first type attribute information about each first measurement voxel and the first measurement voxel. The presence / absence of a difference from the first type attribute information about the first map voxel at the corresponding position is detected. Further, in the difference detection, the difference detection unit 140A, when the operation mode set by the user is the sheet shape / line shape detection mode, the second type attribute information for each second measurement voxel, The presence / absence of a difference from the second type attribute information about the second map voxel at the position corresponding to the second measurement voxel is detected.

  Thereby, 10 A of difference detection apparatuses can detect the presence or absence of a difference using the voxel of a different size in the operation mode from which the shape of the object assumed as a cause of the detected difference differs.

  According to the difference detection apparatus 10A, a user who uses the difference detection apparatus 10A can set parameters used in difference detection performed by the difference detection apparatus 10A by setting an operation mode of the difference detection apparatus 10A. become.

  For this reason, the user who uses 10 A of difference detection apparatuses can reduce the troublesomeness which a user feels when setting the parameter utilized in 10 A of difference detections conventionally.

(Other embodiments)
As described above, the first embodiment and the second embodiment have been described as examples of the technology disclosed in the present application. However, the technology in the present disclosure is not limited to this, and can also be applied to embodiments in which changes, replacements, additions, omissions, and the like are appropriately performed.

  In the first embodiment, as an example, the difference detection device 10 has been described as being mounted on the moving body 20 and used as the moving body 20 moves. However, the difference detection device 10 is not necessarily limited to an example in which the difference detection device 10 is mounted on the moving body 20 and used. For example, the difference detection device 10 may be used by being fixed to a ceiling in a building.

  Further, for example, the measurement unit 110 and the extraction unit 120 have a wireless communication function, and only the measurement unit 110 of the difference detection device 10 is mounted on the moving body 20, and other parts including the extraction unit 120 are measured. An example in which it is stored and used in a casing installed on the floor, for example, within a range in which communication with the unit 110 can be considered.

  In the first embodiment, the measurement unit 110 has been described as including a stereo camera and a depth sensor as a distance measuring device. However, the measurement unit 110 does not necessarily include both a stereo camera and a depth sensor as a distance measuring device, and may include only one of them.

  In addition, the distance measuring device included in the measurement unit 110 is not necessarily limited to a stereo camera and a depth sensor as long as the device has a distance measuring function.

  In the first embodiment, for example, the case where the difference detection unit 140 detects that there is a difference when the types of shapes indicated by the measurement voxel data and the map voxel data are different has been described, but it is significant for the RGB value indicating the color. When there is a difference, the detection may be performed with the difference.

  In Embodiment 1, when the difference detection unit 140 detects the presence or absence of a difference, the map management unit 130 updates the stored map voxel data using information related to the presence or absence of the detected difference. It is good.

  Each functional component (functional block) in the difference detection device 10 or 10A may be individually made into one chip by a semiconductor device such as an IC (Integrated Circuit), LSI (Large Scale Integration), or a part or all of them. It may be made into one chip so that it may be included. Further, the method of circuit integration is not limited to LSI's, and implementation using dedicated circuitry or general purpose processors is also possible. An FPGA (Field Programmable Gate Array) that can be programmed after manufacturing the LSI, or a reconfigurable processor that can reconfigure the connection and setting of circuit cells inside the LSI may be used. Furthermore, if integrated circuit technology that replaces LSI appears as a result of progress in semiconductor technology or other derived technology, functional blocks may be integrated using this technology. Biotechnology can be applied as a possibility.

  In addition, all or part of the various processes described above (for example, the procedures shown in FIGS. 6 to 8 and 14) may be realized by hardware such as an electronic circuit or may be realized by using software. Note that the processing by software is realized by a processor included in the difference detection apparatus executing a control program stored in a memory. Further, the control program may be recorded on a recording medium and distributed or distributed. For example, by installing the distributed control program in the difference detection device and causing the processor of the difference detection device to execute it, the difference detection device performs various processes (the procedures shown in FIGS. 6 to 8, 14, etc.). It becomes possible. This difference detection device may not include a distance measuring device, for example, and may be a computer that acquires a measurement result by the distance measuring device and performs difference detection.

  In addition, forms realized by arbitrarily combining the components and functions described in the above-described embodiments are also included in the scope of the present disclosure.

  The present disclosure can be widely used in an apparatus that detects a difference in space.

10, 10A Difference detection device 110 Measurement unit 120, 120A Extraction unit 130, 130A Map management unit 140, 140A Difference detection unit 150, 150A Parameter setting unit 160 Output unit 202a, 202b, 401, 804a, 804b, 804d, 804e Measurement voxel 402a, 402b, 804c, 804f Map voxels

Claims (19)

A difference detection device for detecting a difference in space,
A measurement unit that measures the position of each of a plurality of points in the surrounding local space;
An extraction unit for extracting attribute information about each measurement voxel in the measurement voxel group arranged in a three-dimensional manner in association with the local space from the measurement result by the measurement unit;
A map management unit that manages attribute information about each map voxel in the map voxel group arranged in a three-dimensional manner in association with the space represented by the map;
A parameter setting unit for setting numerical values to parameters according to the operation mode set by the user;
A difference detection unit that detects whether or not there is a difference between the attribute information about the measurement voxel and the attribute information about the map voxel at a position corresponding to the measurement voxel using the numerical value set in the parameter;
An output unit that outputs a detection result of the difference detection unit,
Difference detection device. In the detection of the presence or absence of the difference, the difference detection unit performs the detection as having a difference when a map voxel at a position corresponding to a measurement voxel having significant attribute information does not have significant attribute information. ,
The difference detection apparatus according to claim 1. In the case where the map voxel at the position corresponding to the measurement voxel having the significant attribute information does not have the significant attribute information, the difference detection unit further corresponds to the measurement voxel having the significant attribute information. When the map voxel located in the vicinity of has no significant attribute information, the detection is performed with the difference.
The difference detection apparatus according to claim 2. The map voxels are overlapped,
The difference detection apparatus according to claim 3. In the detection of the presence / absence of the difference, the difference detection unit is different from a shape type indicated by the attribute information of the measurement voxel and a shape type indicated by the attribute information of the map voxel at a position corresponding to the measurement voxel. Sometimes the detection is performed with a difference,
The difference detection apparatus according to claim 1. In the detection of the presence / absence of the difference, the difference detection unit, when the measurement voxel and a map voxel at a position corresponding to the measurement voxel have attribute information indicating a plane or a line segment, the attribute information of the measurement voxel When the degree of approximation between the normal direction of the plane or line segment to be displayed and the normal direction of the plane or line segment indicated by the attribute information of the map voxel is smaller than a predetermined value, the detection is performed with a difference.
The difference detection apparatus according to claim 1. The measurement is performed N (N is an integer of 2 or more) times,
The extraction unit performs the extraction every time the measurement unit performs the measurement,
Each time the extraction unit performs the extraction, the difference detection unit is positive if there is a difference when a map voxel at a position corresponding to a measurement voxel having significant attribute information does not have significant attribute information. Voting related to the presence or absence of the difference, and detecting the presence or absence of the difference based on the result of the voting related to the presence or absence of the difference N times.
The difference detection apparatus according to claim 1. The difference detection unit further corresponds to the measurement voxel having the significant attribute information when the map voxel at the position corresponding to the measurement voxel having the significant attribute information does not have the significant attribute information. If the map voxel located in the vicinity of the position to have does not have significant attribute information, perform a positive vote with the difference,
The difference detection apparatus according to claim 7. The map voxels are overlapped,
The difference detection apparatus according to claim 8. The measurement is performed N (N is an integer of 2 or more) times,
The extraction unit performs the extraction every time the measurement unit performs the measurement,
The difference detection unit, each time the extraction unit extracts the attribute information of the measurement voxel, the type of shape indicated by the attribute information of the measurement voxel and the attribute information of the map voxel at the position corresponding to the measurement voxel When the types of shapes indicated by are different from each other, a vote relating to the presence / absence of the difference is performed so that a positive vote is made with a difference, and based on the result of voting relating to the presence / absence of the difference of N times Detecting the presence or absence of the difference,
The difference detection apparatus according to claim 1. The measurement is performed N (N is an integer of 2 or more) times,
The extraction unit performs the extraction every time the measurement unit performs the measurement,
The difference detection unit, when the extraction unit extracts attribute information of the measurement voxel, the measurement voxel and the map voxel at a position corresponding to the measurement voxel have attribute information indicating a plane or a line segment In addition, there is a difference when the degree of approximation between the normal direction of the plane or line segment indicated by the attribute information of the measurement voxel and the normal direction of the plane or line segment indicated by the attribute information of the map voxel is smaller than a predetermined value. Voting on the presence or absence of the difference so that a positive vote is made, and detecting the presence or absence of the difference based on the results of N votes;
The difference detection apparatus according to claim 1. The measurement unit is mounted on a moving body,
The operation mode includes a first mode and a second mode in which the moving speed of the moving body is different from each other,
The parameter setting unit sets different values for the parameters when the operation mode is the first mode and when the operation mode is the second mode,
The difference detection unit performs the detection with a difference when the set numerical value of the parameter and the result of the N votes are in a predetermined relationship.
The difference detection apparatus of any one of Claims 7-11. The measurement unit performs each measurement in the N measurements in a predetermined cycle,
The operation mode includes a first mode and a second mode having different predetermined cycles,
The parameter setting unit sets different values for the parameters when the operation mode is the first mode and when the operation mode is the second mode,
The difference detection unit performs the detection with a difference when the set numerical value of the parameter and the result of the N votes are in a predetermined relationship.
The difference detection apparatus of any one of Claims 7-11. The operation mode includes a first mode and a second mode corresponding to a place where the difference detection device is used,
The parameter setting unit sets different values for the parameters when the operation mode is the first mode and when the operation mode is the second mode,
The difference detection unit performs the detection with a difference when the set numerical value of the parameter and the result of the N votes are in a predetermined relationship.
The difference detection apparatus of any one of Claims 7-11. The measurement unit includes a first sensor and a second sensor capable of performing the measurement,
The operation mode includes a first mode using a measurement result by the first sensor and a second mode using a measurement result by the second sensor,
The extraction unit extracts attribute information from the measurement result of the first sensor when the operation mode is the first mode, and when the operation mode is the second mode, the second sensor Extract attribute information from the measurement results by
The parameter setting unit sets different values for the parameters when the operation mode is the first mode and when the operation mode is the second mode,
The difference detection unit performs the detection with a difference when the set numerical value of the parameter and the result of the N votes are in a predetermined relationship.
The difference detection apparatus of any one of Claims 7-11. The first mode and the second mode are operation modes in which the difference detection device is used indoors.
The difference detection apparatus according to claim 15. The operation mode includes a third mode and a fourth mode in which the shapes of objects assumed as causes of the detected difference are different from each other,
The extraction unit extracts attribute information about a measurement voxel of a first size when the operation mode is the third mode, and extracts the first information when the operation mode is the fourth mode. Extract attribute information about measurement voxels of second size different from size,
The map management unit manages attribute information about the first size map voxels and attribute information about the second size map voxels;
When the operation mode is the third mode, the difference detection unit includes attribute information about the measurement voxel of the first size and the first size of the position corresponding to the measurement voxel of the first size. If the difference between the map voxel and attribute information is detected and the operation mode is the fourth mode, the attribute information about the second size measurement voxel and the measurement voxel of the second size Detecting the presence or absence of a difference from the attribute information about the second size map voxel at the position corresponding to
The difference detection apparatus of any one of Claims 1-16. The output unit performs the output to display an image indicating the detection result on a display device.
The difference detection apparatus of any one of Claims 1-17. A difference detection method for detecting a difference in space,
A measuring step for measuring a position for each of a plurality of points in the surrounding local space;
An extraction step of extracting attribute information about each measurement voxel in the measurement voxel group arranged in a three-dimensional manner in association with the local space from the measurement result of the measurement step;
Reading out the attribute information about each map voxel in the map voxel group arranged in a three-dimensional manner in association with the space represented by the map;
A parameter setting step for setting a numerical value to the parameter according to the operation mode set by the user;
Using the numerical value set in the parameter, the attribute information about the measurement voxel extracted by the extraction step and the map voxel at the position corresponding to the measurement voxel read by the read step A difference detection step for detecting the presence or absence of a difference from the attribute information;
An output step of outputting a detection result by the difference detection step,
Difference detection method.

Images