Measurement error correction system and measurement error correction method

Abstract

To measure a mobile body with high accuracy.SOLUTION: A measurement error correction system includes a control apparatus which executes a predetermined process, and a storage apparatus which can be accessed by the control apparatus. The control apparatus includes: a measurement simulation unit which performs, by the control apparatus, simulation to measure a mobile body on a virtual space that reproduces a real space; a correction model generation unit which generates, by the control apparatus, a correction model for correcting measurement error of the mobile body, using a parameter as an explanatory variable, on the basis of a result of the simulation and the parameter set in the simulation; and a measurement error correction unit which corrects, by the control apparatus, a result of measuring the mobile body in the real space, on the basis of the correction model.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a technique for measuring a moving object.

Background Art

[0002] There is an increasing need to accurately measure the flow of people or objects indoors or outdoors using sensors such as cameras or LiDAR. However, since the measurement accuracy varies depending on dynamically changing environments such as lighting conditions, the number of measurement targets, and the distance from the measurement targets, correction is required.

[0003] As a technique for correcting the count of the number of people, for example, there is a technique described in Japanese Unexamined Patent Application Publication No. 2016-91326 (Patent Document 1). Patent Document 1 describes, "The present invention is a method for measuring the number of objects moving within an image from a camera image, including steps of estimating the movement of the moving object, setting a line for determining the movement state of the moving object within the image, setting a mesh parallel to the line for determining the size of the moving object, correcting the distortion of the scene of the camera image, when the moving objects overlap within the image, correcting the overlap of the moving objects for the estimated movement to extract particles of the moving object, correcting the size of the moving object for the extracted particles of the moving object, and estimating the number of moving objects in a speed-to-number conversion unit based on the particles of the moving object whose size has been corrected."

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] According to the above-mentioned Patent Document 1, in order to accurately count the number of people, in addition to conventional preprocessing, camera image processing includes correction of coordinate system distortion, apparent size correction, and congestion level correction. However, in actual measurements, it has been difficult to automatically identify when the measurement accuracy has decreased due to the measurement environment and to perform corrections on the captured video data according to the situation. [Means for solving the problem]

[0006] To solve at least one of the above problems, a representative example of the invention disclosed in this application is a measurement error correction system comprising a control device that performs a predetermined process and a storage device accessible by the control device, The storage device holds spatial information that reproduces the real space, and sensor information that reproduces the position, measurement direction, and specifications of the measurement sensors placed in the real space. The control device, Based on the spatial information Recreating the real world did virtual space An agent simulation that generates a moving object, and the agent simulation that generated Mobile Based on the spatial information and the sensor information Measure measurement simulation and, Execute Then, the difference between the number of moving objects generated by the agent simulation and the number of moving objects measured by the measurement simulation is obtained as the measurement error. The measurement simulation unit and the control device are The agent simulation and The aforementioned measurement Simulation results and , the agent simulation and The aforementioned measurement Based on the parameters set in the simulation, the parameters are used as explanatory variables. Measurement error The control device is characterized by comprising a correction model generation unit that generates a correction model for correcting the error, and a measurement error correction unit that corrects the number of moving objects measured in the real space based on the correction model. [Effects of the Invention]

[0007] According to one aspect of the present invention, measurement accuracy can be improved by supplementing missing values ​​according to the measurement environment. Problems, configurations, and effects other than those described above will be clarified by the following description of the embodiments. [Brief explanation of the drawing]

[0008] [Figure 1]This is an explanatory diagram illustrating the overview of measurement error correction in an embodiment of the present invention. [Figure 2] This is a block diagram showing the configuration of the measurement error correction system in an embodiment of the present invention. [Figure 3] This flowchart shows the process performed by the measurement simulation unit of the measurement error correction device in an embodiment of the present invention. [Figure 4] This is an explanatory diagram showing the three-dimensional spatial information held by the measurement error correction device in an embodiment of the present invention. [Figure 5] This is an explanatory diagram showing the equipment and sensor information held by the measurement error correction device in an embodiment of the present invention. [Figure 6] This is an explanatory diagram showing the simulation model held by the measurement error correction device in an embodiment of the present invention. [Figure 7] This is an explanatory diagram showing the simulation conditions held by the measurement error correction device in an embodiment of the present invention. [Figure 8A] This is a plan view illustrating an example of a simulation performed by the measurement error correction device in an embodiment of the present invention. [Figure 8B] This is a front view illustrating an example of a simulation performed by the measurement error correction device in an embodiment of the present invention. [Figure 9] This is an explanatory diagram showing the simulation results held by the measurement error correction device in an embodiment of the present invention. [Figure 10] This flowchart shows the process executed by the correction model generation unit of the measurement error correction device in an embodiment of the present invention. [Figure 11] This is an explanatory diagram showing the measurement simulation results related to a specific sensor included in the simulation results held by the measurement error correction device in an embodiment of the present invention. [Figure 12] This is an explanatory diagram showing the importance of measurement error occurrences held by the measurement error correction device in an embodiment of the present invention. [Figure 13] This is an explanatory diagram showing the correction model information held by the measurement error correction device in an embodiment of the present invention. [Figure 14A]It is an explanatory diagram showing the performance data acquired by the measurement error correction device in an embodiment of the present invention. [Figure 14B] It is an explanatory diagram showing a first example of the generation and concentration parameters of pedestrians generated by the measurement error correction device in an embodiment of the present invention based on performance data. [Figure 14C] It is an explanatory diagram showing a second example of the generation and concentration parameters of pedestrians generated by the measurement error correction device in an embodiment of the present invention based on performance data. [Figure 14D] It is an explanatory diagram showing the lighting parameters generated by the measurement error correction device in an embodiment of the present invention based on performance data. [Figure 15] It is a flowchart showing the processing executed by the measurement error correction unit of the measurement error correction device in an embodiment of the present invention. [Figure 16] It is an explanatory diagram showing the measurement environment information acquired by the measurement error correction device in an embodiment of the present invention. [Figure 17] It is an explanatory diagram showing the measurement data acquired by the measurement error correction device in an embodiment of the present invention. [Figure 18] It is an explanatory diagram showing the measurement simulation setting screen displayed by the measurement error correction device in an embodiment of the present invention. [Figure 19] It is an explanatory diagram showing the correction model generation screen displayed by the measurement error correction device in an embodiment of the present invention. [Figure 20] It is an explanatory diagram showing the pedestrian flow measurement screen displayed by the measurement error correction device in an embodiment of the present invention.

Mode for Carrying Out the Invention

[0009] Hereinafter, embodiments of the present invention will be described based on the drawings.

[0010] FIG. 1 is an explanatory diagram showing an overview of measurement error correction in an embodiment of the present invention.

[0011] In this embodiment, a virtual space 111 is generated that reproduces the real space 101 using so-called digital twin technology. The real space 101 is the space where measurements are taken, and typical examples include train stations, commercial facilities, public facilities, roads, intersections, etc. However, the space where measurements are taken is not limited to these, and it may be either an outdoor or indoor space. The object of measurement is, for example, a person or a moving object such as a vehicle. In the following description, the measurement of pedestrian flow in a train station space will be described as an example, but the present invention is not limited to this and can be applied to the measurement of any moving object in any space.

[0012] In the virtual space 111, the measurement simulator 114 performs a measurement simulation. The measurement simulator 114 holds spatial information 113 that constitutes the virtual space and simulation environment factors 112. The spatial information 113 includes 3D data for reproducing the real space of the facility where the measurement is performed, and information about sensors installed in the real space for measurement. The former may include, for example, point cloud data, BIM (Building Information Modeling) data, and lighting data. The latter may include, for example, information about the type and installation location of sensors. The simulation environment factors 112 may include, for example, information about the time of day, weather, lighting, traffic, stores, and demand.

[0013] The measurement simulator 114 performs three-dimensional spatial recognition based on spatial information and conducts agent simulations regarding the moving object to be measured while varying the simulation environment factors 112. For example, if the object to be measured is the number of people, the measurement simulator uses the models of people and moving objects included in the simulation model 115 to generate a flow of people.

[0014] The measurement simulator 114 then performs a measurement simulation to measure the generated pedestrian flow using sensors. At this time, the measurement simulator 114 uses the sensor model included in the simulation model 115 to simulate detection by sensors, and further uses the discrimination and recognition model included in the simulation model 115 to recognize people from the information detected by the sensors and count their numbers.

[0015] The measurement simulator 114 then compares the number of people counted by the measurement simulation with the number of people generated by the agent simulation that should have been counted. If there is a difference between the two, it generates a correction model 105 to correct that difference.

[0016] Meanwhile, in the real space 101, data is acquired from measuring devices 102 such as sensors that are actually installed, people are recognized, and the number of people is counted. Then, by applying the measurement result 104 and the actual environmental information 103 at the time of measurement to a correction model 105 generated by the simulation in the virtual space, the measurement error is corrected, and the corrected measurement result 106 is obtained. The environmental information 103 applied here corresponds to the simulation environmental factors 112 and is information that shows the actual environment at the time of measurement.

[0017] The following describes the details of generating the correction model mentioned above and correcting measurement errors based on the generated correction model.

[0018] Figure 2 is a block diagram showing the configuration of the measurement error correction system in an embodiment of the present invention.

[0019] The measurement error correction device 1, which constitutes the measurement error correction system of this embodiment, includes a central control device 11, an input device 12, an output device 13, a communication device 14, a main memory device 15, and an auxiliary memory device 16. They are connected to each other by a bus.

[0020] The main memory 15 and auxiliary memory 16 store programs and data used by the central control unit 11. The main memory 15 is composed of, for example, semiconductor memory and is mainly used to hold the currently running program and data. The central control unit 11 performs various processes according to the programs stored in the main memory 15. By operating according to the programs, the central control unit 11 realizes functional units such as the measurement simulation unit 21, the correction model generation unit 22, and the measurement error correction unit 23.

[0021] The auxiliary storage device 16 is comprised of a large-capacity storage device, such as a hard disk drive or solid-state drive, and is used to retain programs and data for a long period of time. The auxiliary storage device 16 stores, for example, three-dimensional spatial information 31, equipment and sensor information 32, simulation models 33, simulation conditions 34, simulation results 35, evaluation results of missing data occurrence factors 36, correction model information 37, and measurement results 38. Details of this information will be described later.

[0022] The central control unit 11 can consist of a single processing unit or multiple processing units, and may include one or more arithmetic units or multiple processing cores. The central control unit 11 can be implemented as one or more central processing units, microprocessors, microcomputers, microcontrollers, digital signal processors, state machines, logic circuits, graphics processing units, chip-on systems, and / or any device that manipulates signals based on control instructions.

[0023] The programs and data stored in the auxiliary storage device 16 are loaded into the main storage device 15 at startup or when necessary, and the central control unit 11 executes the program, thereby executing the processing of each functional unit of the measurement error correction device 1. Therefore, in the following description, the processing performed by each functional unit of the measurement error correction device 1 is the processing performed by the central control unit 11 according to the program.

[0024] The input device 12 is a hardware device for the user to input instructions and information to the measurement error correction device 1. The output device 13 is a hardware device that presents various images for input and output, such as a display device or a printing device. The communication device 14 is an interface for connecting to the network 4.

[0025] The measurement error correction device 1 may be connected to an external system 2 and an external server 3 via a network 4. For example, at least a portion of the information held in the auxiliary storage device 16 may be obtained from either the external system 2 or the external server 3. Specifically, for example, the external system 2 may include sensors located in the physical space 101. Also, for example, the external server 3 may include a server that provides weather information and traffic information.

[0026] The measurement error correction device 1 may include two or more central control devices 11. Furthermore, the functions of the measurement error correction device 1 can be implemented on multiple computers. In this case, the multiple computers communicate via a network. For example, some of the functions of the measurement error correction device 1 may be implemented on one computer, while other functions may be implemented on other computers.

[0027] Figure 3 is a flowchart showing the process executed by the measurement simulation unit 21 of the measurement error correction device 1 in an embodiment of the present invention.

[0028] The measurement simulation unit 21 reads 3D spatial information 31, equipment / sensor information 32, and a simulation model 33 (steps 301, 302, and 303, respectively). Next, the measurement simulation unit 21 sets multiple simulation conditions 34 (step 304), executes the measurement simulation (step 305), and saves the result as a simulation result 35 (step 306). Details of these processes will be explained with reference to Figures 4 to 9.

[0029] Figure 4 is an explanatory diagram showing the three-dimensional spatial information 31 held by the measurement error correction device 1 in an embodiment of the present invention.

[0030] Figure 4 shows an example of point cloud data representing walls and objects placed in real space 101 as three-dimensional spatial information 31. Specifically, the three-dimensional spatial information 31 includes coordinates x401, y402, and z403, which represent the x, y, and z coordinates of each point constituting the point cloud, and color 404, which represents the color of each store. In step 301 of Figure 3, three-dimensional spatial information 31, such as that shown in Figure 4, is read, and a virtual space 111 is generated based on this information.

[0031] Figure 5 is an explanatory diagram showing the equipment and sensor information 32 held by the measurement error correction device 1 in an embodiment of the present invention.

[0032] The equipment / sensor information 32 contains information indicating the location and specifications of equipment and sensors placed in the physical space 101. Specifically, the equipment / sensor information 32 includes name 501, type 1_502, type 2_503, specification 1_504, specification 2_505, placement coordinates 506, and placement angle 507. Name 501 is the name of each piece of equipment. Type 502 and type 2_503 indicate the type of each piece of equipment. Specification 1_504 and specification 2_505 indicate the specifications of each piece of equipment. Placement coordinates 506 are the coordinates indicating the location where each piece of equipment is placed. Placement angle 507 is the angle indicating the orientation of each piece of equipment.

[0033] For example, the first line of the equipment / sensor information 32 shown in Figure 5 indicates that the type of lighting equipment named "Lighting A" is a linear LED lighting system, its brightness is 1200 lm, and it is positioned at coordinate (1,2,3) with an angle of (0,0,0).

[0034] Furthermore, the second line of the equipment / sensor information 32 shown in Figure 5 indicates that the sensor named "Sensor A" is a camera that captures images in all directions, its size (i.e., the number of pixels in the captured image) is 1920 x 1080 pixels, and it is positioned at coordinates (10,10,15) with an angle of (180,45,0).

[0035] In step 302 of Figure 3, equipment and sensor information 32, such as that shown in Figure 5, is read. Based on the equipment and sensor information 32, a virtual space 111 is generated such that equipment with the same specifications as the equipment placed in the real space 101 is placed in the same position and angle as in the real space 101.

[0036] The above-mentioned three-dimensional spatial information 31 and equipment / sensor information 32 are included in the spatial information 113 in Figure 1.

[0037] Figure 6 is an explanatory diagram showing the simulation model 33 held by the measurement error correction device 1 in an embodiment of the present invention.

[0038] The simulation model 33 includes coordinate data 600, movement data in a first form 610, movement data in a second form 620, timetable data 630, and a pedestrian model 640.

[0039] The coordinate data 600 is information that defines each point to be dealt with in the simulation, and includes point 601 that identifies each point, and x coordinates 602, y coordinates 603, and z coordinates 604 that indicate the position of each point.

[0040] The first form of movement data 610 is data that defines the location to which a moving object moves and the number of occurrences of the moving object. Specifically, the first form of movement data 610 includes ID 611, moving object 612, moving location 1_613, moving location 2_614, and occurrence count 615.

[0041] ID611 identifies the data registered in the first form of movement data 610. Movement object 612 identifies the moving movement object. Movement point 1_613 and movement point 2_614 indicate the points where the movement object moves, such as the origin and destination of the movement. These points are defined by the coordinate data 600. Occurrence count 615 indicates the number of occurrences (e.g., occurrence frequency) of the movement object that is making the move.

[0042] For example, the first row of the first type of movement data 610 shown in Figure 6 indicates that one person appears every 10 seconds moving from point A to point B.

[0043] The second form of movement data 620 is data that defines the points to which a moving object moves and the schedule of the moving object's movement. Specifically, the second form of movement data 620 includes the moving object 621, point 1_622, point 2_623, point 3_624, and timetable ID 625. In the example in Figure 6, the second form of movement data 620 defines the schedule of movement of a moving object other than the person being measured (e.g., a railway vehicle).

[0044] Mobile object 621 identifies a moving mobile object. Points 1_622, 2_623, and 3_624 indicate points along the mobile object's movement, such as the starting point, intermediate points, and ending point. These points are defined by coordinate data 600. Although only three points are shown in the example in Figure 6, more points may actually be defined for each mobile object. Timetable ID 625 identifies a timetable that defines the schedule for the mobile object's movement.

[0045] For example, the first row of the second form of travel data 620 shown in Figure 6 indicates that a railway vehicle traveling from point X to point Y via point C will operate according to the timetable with timetable ID "1". This timetable is defined by the timetable data 630.

[0046] The timetable data 630 is data that defines the schedule of movement of a moving object. Specifically, the timetable data 630 includes ID 631, departure from point 1 632, arrival at point 2 633, departure from point 2 634, and arrival at point 3 635. ID 631 identifies each defined timetable. Departures from point 1 632 to arrivals at point 3 635 indicate the times when the moving object departs from and arrives at each point defined by the second form of movement data 620.

[0047] The pedestrian model 640 is data that defines a model of the movement of a person, which is the moving object being measured in this embodiment. Arbitrary models are defined as a collision avoidance model and a path selection model to reproduce the movement of a pedestrian. These models are used in the measurement simulation described later. The models defined here may be publicly known models and are not particularly limited. Furthermore, if the moving object being measured is not a pedestrian, a model suitable for that type of moving object is maintained.

[0048] The coordinate data 600 to pedestrian model 640 mentioned above correspond to the models of people and moving objects included in the simulation model 115 shown in Figure 1. In step 303 of Figure 3, for example, a simulation model 33 as shown in Figure 6 is loaded.

[0049] Figure 7 is an explanatory diagram showing the simulation conditions 34 held by the measurement error correction device 1 in an embodiment of the present invention.

[0050] Simulation conditions 34 include simulation condition settings 700, lighting settings 710, store settings 720, and demand settings 730.

[0051] The simulation condition setting 700 is information that defines the settings for each of the multiple simulation conditions. Specifically, the simulation condition setting 700 includes a simulation number 701 that identifies each simulation condition being set, and setting parameters 702 for each simulation condition.

[0052] The setting parameter 702 includes settings related to lighting, traffic, stores, and the generation and concentration of moving objects in the simulation. The lighting settings are defined, for example, by the lighting settings 710 described later. The traffic settings are defined, for example, by the timetable data 630 shown in Figure 6. The store settings are defined, for example, by the store settings 720 described later. The settings related to the generation and concentration of moving objects are defined, for example, by the demand settings 730 described later.

[0053] Lighting setting 710 is information that defines the settings for lighting parameters in the simulation. Specifically, lighting setting 710 includes setting 711, sunlight 712, lighting 1_713, and lighting 2_714.

[0054] Setting 711 identifies each of several settings for lighting parameters. Sunlight 712 includes information defining parameters related to sunlight in the simulation, such as brightness, color temperature, and ray angle. Lighting 1_713 and Lighting 2_714 include information defining the specifications of lighting equipment placed in the virtual space, such as brightness and color temperature. In the example in Figure 6, information about two lighting equipment is defined, but in practice, it is desirable to define information about all lighting equipment placed in the virtual space where the simulation is performed.

[0055] Store settings 720 is information that defines the settings for store parameters in the simulation. Specifically, store settings 720 includes settings 721, store1_722, store2_723, and store3_724.

[0056] Setting 721 identifies each of several settings for store parameters. Store1_722 to Store3_724 each define the state of a store placed in the virtual space, for example, whether the store is open (i.e., open for business) or closed (i.e., closed). In the example in Figure 6, information about three stores is defined, but in practice, it is desirable to define information about all stores placed in the virtual space where the simulation is performed.

[0057] The demand setting 730 is information that defines the parameters related to the movement of the object being measured (a person in this embodiment) within the virtual space where the simulation is performed. Specifically, the demand setting 730 includes setting 731, location 1_732, location 2_733, and location 3_734.

[0058] Setting 731 identifies each of several settings for parameters related to the movement of a moving object. Points 1_732 to 3_734 are information that defines the settings for the movement of a moving object at each point in the virtual space. Here, each point is a location defined by the coordinate data 600 shown in Figure 6, and for each point, movement parameters (e.g., the frequency of appearance of a person moving between points) defined based on the first form of movement data 610 shown in Figure 6 may be set. In the example in Figure 6, information about three points is defined, but in practice, it is desirable to define information about all points set in the virtual space where the simulation is performed.

[0059] The measurement simulation unit 21 either generates simulation conditions 34 as described above in step 304 of Figure 3, or reads simulation conditions 34 that have already been generated, and in step 305 sets the parameters for each simulation number, generates the flow of people in the virtual space by agent simulation, and counts the number of people generated by measurement simulation. By executing the measurement simulation multiple times based on the multiple simulation conditions defined by the simulation condition setting 700, the simulation results for each parameter value are obtained.

[0060] Figures 8A and 8B are a plan view and a front view illustrating an example of a simulation performed by the measurement error correction device 1 in an embodiment of the present invention.

[0061] The space 800 shown in Figures 8A and 8B is at least a part of the virtual space 111 being measured, and is a reproduction of the corresponding portion of the real space 101 based on 3D spatial information 31 and equipment / sensor information 32. In this example, space 800 is a reproduction of the space inside a railway station.

[0062] Store 1_801 is located in space 800. The state of store 1_801 is defined by store settings 720. Figures 8A and 8B show an example of store 1_801 being open. Sensor A802 is also located in space 800. The type, specifications, and location of sensor A802 are defined by equipment / sensor information 32. According to the example in Figure 5, sensor A802 is a camera. Although omitted in Figures 8A, etc., other sensors (e.g., sensor B) may also be located within space 800.

[0063] Based on the definition of the sensor A802's specifications, placement coordinates, and placement angle from the equipment / sensor information 32, and the definition of space 800 from the three-dimensional spatial information 31, the measurement range 803 of sensor A802 can be determined.

[0064] Space 800 also contains lighting fixtures 1_804 and windows 805. These are also defined by the equipment and sensor information 32.

[0065] Points A_806A, B_806B, and C_806C within space 800 are defined by coordinate data 600. The movement of people between points (e.g., the frequency of people moving) is defined by, for example, demand settings 730 and movement data in a first form 610 or a second form 620, etc.

[0066] The measurement simulation unit 21 generates a flow of people 810 based on the corresponding demand setting 730, the first type of movement data 610, and the pedestrian model 640 for each simulation condition set by the simulation condition setting 700. Then, it performs a measurement simulation in which sensor A measures the generated people and recognizes and identifies the results.

[0067] At this time, the lighting conditions within space 800 (i.e., brightness, color temperature, and direction of light rays, etc.) are determined based on the lighting settings 710 and equipment / sensor information 32 for each simulation condition. The measurement results of sensor A_802 (in this example, images captured by the camera in the simulation) are determined based on the flow of people that occurred, the lighting conditions, and the specifications of sensor A_802 stored in the equipment / sensor information 32. Then, by applying the same recognition / discrimination model that is applied to images captured in real space to the images captured by the camera in the simulation, the recognition result of people based on the measured images is obtained.

[0068] Figure 9 is an explanatory diagram showing the simulation results 35 held by the measurement error correction device 1 in an embodiment of the present invention.

[0069] The measurement simulation unit 21 saves the results of the measurement simulation performed in step 305 as simulation result 35 in the auxiliary storage device 16 in step 306. The simulation result 35 shown in Figure 9 includes simulation number 901, the simulation result 902 for sensor A, and the simulation result 903 for sensor B.

[0070] Simulation number 901 is information that identifies the simulation conditions used, and corresponds to simulation number 701 in Figure 7.

[0071] The simulation result 902 for sensor A is the result of a simulation based on measurements of sensor A_802, as shown in Figure 8A, for example, and includes model 902A, measurement result 902B, and error 902C.

[0072] Model 902A shows the model used to recognize and classify images captured by sensor A_802 in the measurement simulation. It is desirable that this model be identical to the model used to recognize and classify images captured by a camera located in the real-world space 101, which sensor A_802 is reproducing. This allows for accurate reproduction of measurements of moving objects in real space in the measurement simulation.

[0073] Measurement result 902B is the number of people in the target space (e.g., space 800) detected by recognizing the image captured by sensor A_802 using a recognition / discrimination model. Error 902C is the difference between the number of people obtained as measurement result 902B and the number of people generated in the target space by simulation (i.e., the correct value).

[0074] For example, if a simulation shows that 12 people exist within space 800, but only 10 people are detected in the image captured by sensor A_802, then the measurement result 902B will be "10 (people)" and the error 902C will be "-2 (people)".

[0075] The simulation result 903 for sensor B is the result of a measurement simulation of another sensor (for example, sensor B which is omitted in Figure 8A, etc.), and includes model 903A, measurement result 903B, and error 903C. These are the same as model 902A, measurement result 902B, and error 902C, so their explanation is omitted.

[0076] In the example shown in Figure 9, the simulation results for sensors A and B are displayed. However, in reality, the simulation results for all sensors placed in space and subjected to measurement simulations are saved as simulation results 35. Furthermore, while Figure 9 shows only the main information as an example, the actual simulation results 35 may include more detailed information about the simulation results (e.g., error rate, values ​​of parameters set in the simulation, etc.).

[0077] Figure 10 is a flowchart showing the process performed by the correction model generation unit 22 of the measurement error correction device 1 in an embodiment of the present invention.

[0078] The correction model generation unit 22 reads the results of the measurement simulation (step 1001). For example, the simulation result 35 shown in Figure 9 is read. Next, the correction model generation unit 22 calculates the importance of each simulation factor for the occurrence of measurement errors based on the simulation result and the simulation factors input into the simulation model to obtain the simulation result (step 1002). Here, the simulation factors are the parameters input into the simulation model based on the simulation conditions 34 when the measurement simulation is executed.

[0079] The importance of each parameter is calculated based on the magnitude of the impact that changes in the parameter values ​​have on the magnitude of the measurement error, with parameters that have a greater impact being given higher importance. For example, the feature importance of a decision tree algorithm can be used when the setting parameters of the measurement simulation are the explanatory variables and the error rate is the dependent variable.

[0080] Next, the correction model generation unit 22 determines the features to be used in the correction model (step 1003). For example, the correction model generation unit 22 may determine one or more items with high importance from among the parameter items used in the simulation as features to be used in the correction model. However, even if an item is a parameter with high importance, if the value of that item (e.g., the measured value of the sensor) cannot be obtained in the real space 101 of the measurement target, that item may be excluded from the features to be used in the correction model. This determination may be performed automatically or manually by the user.

[0081] Next, the correction model generation unit 22 generates a correction model to correct the measurement error based on the features determined in step 1003 and the measurement error obtained from the simulation results (step 1004). This correction model is, for example, a model in which the features determined in step 1003 are explanatory variables and the measurement error is the dependent variable. Any method can be used for generating the correction model, including known methods such as multiple regression models, so a detailed explanation is omitted.

[0082] Here, the correction model generation unit 22 may generate multiple correction models based on data obtained from a measurement simulation based on a single simulation condition. Specifically, for example, the correction model generation unit 22 may generate multiple correction models with different parameters as explanatory variables, or it may generate multiple correction models with different estimation algorithms.

[0083] Furthermore, the correction model generation unit 22 may further generate a model for estimating at least one value of the explanatory variables of the correction model from at least one value of the other explanatory variables. For example, if there is a causal relationship or correlation between parameters obtained from real space 101, a model can be generated for estimating one of them from the other. This reduces the number of parameters that need to be obtained from real space 101 when actually using the correction model to correct measurement errors.

[0084] Specifically, for example, if the explanatory variables of the correction model include sunlight brightness, color temperature, and angle, these are correlated with season, time, and weather, etc. Therefore, a model can be generated that estimates sunlight brightness, color temperature, and angle, etc., using season, time, and weather, etc., as explanatory variables. By using such a model, accurate measurement errors can be corrected using a correction model that uses these parameters as explanatory variables, without having to obtain parameter values ​​related to sunlight from the real space 101.

[0085] Furthermore, when multiple measurement sensors are arranged in the real space 101, the correction model generation unit 22 may generate a correction model that includes the measurement values ​​of other measurement sensors as explanatory variables to correct the measurement error of one of the measurement sensors. For example, when multiple cameras are installed in the real space 101, the measurement result from one camera (e.g., the number of pedestrians) may be included as an explanatory variable in the correction model to correct the error of the measurement result from the other camera. This makes it possible to generate a correction model with high accuracy.

[0086] The correction model generation unit 22 stores the information of the generated correction model (for example, information indicating the parameters, explanatory variables, and type of model that constitute the correction model) in the correction model information 37 (step 1005).

[0087] Details of the above process will be explained with reference to Figures 11 to 13, etc.

[0088] Figure 11 is an explanatory diagram showing the measurement simulation results for a specific sensor included in the simulation results 35 held by the measurement error correction device 1 in an embodiment of the present invention.

[0089] The measurement simulation result 1100 shown in Figure 11 is part of the saved simulation result 35, and as an example, it shows the measurement simulation result for sensor A. For example, the measurement simulation result 1100 includes simulation number 1101, the measurement simulation result for sensor A 1102, and the setting parameter 1103.

[0090] Simulation number 1101 is information that identifies the simulation conditions used, and corresponds to simulation number 701 in Figure 7 and simulation number 901 in Figure 9. The measurement simulation result 1102 for sensor A is the result of a simulation based on the measurement of sensor A_802 shown, for example, in Figure 8A, and includes the measured value 1102A, error 1102B, and error rate 1102C. The measured value 1102A and error 1102B correspond to the measurement result 902B and error 902C shown in Figure 9, respectively. The error rate 1102C is the ratio of error 1102B to the value obtained by correcting the measured value 1102A based on error 1102B.

[0091] The setting parameter 1103 is information indicating the parameter values ​​set for each simulation condition, and includes, for example, the amount of people generated at location A_806A 1103A, the amount of people generated at location C_806C 1103B, the state of store 1_801 1103C, and the brightness of sunlight 1103D. Although omitted in Figure 11, the setting parameter 1103 may also include other parameter values ​​set in the simulation, such as the amount of people generated at location B_806B, the angle of sunlight, the state of lighting equipment, the time, and the weather, or it may not include any of the parameter values ​​shown.

[0092] Figure 12 is an explanatory diagram showing the importance level of measurement error occurrence (1200) held by the measurement error correction device 1 in an embodiment of the present invention.

[0093] The importance level 1200 for measurement error occurrence shown in Figure 12 may be included in the evaluation result 36 of the missing data occurrence factors shown in Figure 1. The importance level 1200 for measurement error occurrence includes, for example, the setting parameter 1201 and importance level 1202. The setting parameter 1201 is a parameter set in the simulation, for example, one of the parameters included in the setting parameter 1103. The importance level 1202 indicates the importance calculated for each parameter in step 1002 of Figure 10. In the example in Figure 12, it is shown that the number of people at point A has a large impact on the magnitude of the error, while the impact of weather on the magnitude of the error is small in comparison.

[0094] Figure 13 is an explanatory diagram showing the correction model information 1300 held by the measurement error correction device 1 in an embodiment of the present invention.

[0095] The correction model information 1300 shown in Figure 13 is included, for example, in the correction model information 37 generated and stored by the correction model generation unit 22. The correction model information 1300 includes a correction model 1301 that identifies each generated correction model, setting parameters 1302 used by each correction model, and the estimation accuracy 1303 of missing values ​​by each correction model.

[0096] The setting parameters 1302 used by each correction model are, for example, information that identifies the parameters corresponding to the features determined in step 1003 when generating each correction model, and may correspond to, for example, the environmental information ID (see Figure 16) described later.

[0097] The estimation accuracy 1303 for missing values ​​is the estimation accuracy of the error by each correction model. For example, it may be calculated based on the difference between the error 1102B obtained by the measurement simulation and the error estimated by inputting the setting parameters 1103 from the measurement simulation into the correction model, such that the smaller the difference, the higher the accuracy.

[0098] Next, an example of how to determine the parameters to be set as simulation conditions will be explained with reference to Figures 14A to 14D.

[0099] Figure 14A is an explanatory diagram showing the actual data acquired by the measurement error correction device 1 in an embodiment of the present invention.

[0100] The actual data 1400 shown in Figure 14A is an example of data acquired in the physical space 101 over a certain period, and may be acquired from, for example, an external system 2 or an external server 3 and stored in the auxiliary storage device 16. The actual data 1400 shown in Figure 14A includes date 1401, weekday / holiday 1402, time period 1403, store 1_1404, weather 1405, number of users of store 1 1406, amount generated at location A 1407, amount generated at location C 1408, and number of train A arrivals 1409.

[0101] Date 1401, Weekday / Holiday 1402, and Time Slot 1403 indicate the date the data was acquired, whether that day was a weekday or a holiday, and the time slot in which the data was acquired, respectively. Store 1_1404 and Store 1 User Count 1406 indicate whether Store 1 (for example, the store in physical space 101 corresponding to Store 1_801 shown in Figure 8A) was open or closed during each time slot in which the data was acquired, and the actual number of users measured at that store. The user count data may be obtained from, for example, a POS (Point of Sale) system. Weather 1405 indicates the weather during each time slot in which the data was acquired.

[0102] The occurrence data for location A (1407) and location C (1408) respectively represent the number of people (number of occurrences) actually measured at location A and location C during each time period for which data was acquired. Locations A and C are, for example, locations within real space 101 corresponding to location A_806A and location C_806C shown in Figure 8A, respectively. For example, real space 101 may be the space within a railway station, with location A_806A being the ticket gate and location C_806C being the platform where trains arrive. The number of train arrivals for train A (1409) is the number of times trains arrived during each time period for which data was acquired. The number of occurrences at each location may be obtained from, for example, ticket gate data, and the number of train arrivals may be obtained from, for example, timetable data.

[0103] Figure 14B is an explanatory diagram showing a first example of pedestrian occurrence and concentration parameters generated by the measurement error correction device 1 in an embodiment of the present invention based on actual data.

[0104] The pedestrian occurrence and concentration parameter 1410 shown in Figure 14B is an example of a parameter for measurement simulation generated from actual data 1400. This parameter lists the points along the movement path of a moving object (in this example, a person) and indicates the number of people appearing at those points. For example, the pedestrian occurrence and concentration parameter 1410 includes the moving object 1411, movement point 1_1412, movement point 2_1413, movement point 3_1414, and the number of people appearing 1415.

[0105] The moving object 1411 indicates the type of moving object (e.g., a person). Movement points 1_1412 to 3_1414 indicate intermediate points along the moving object's path. The number of occurrences 1415 indicates the number of moving objects that travel along each path (e.g., frequency of occurrence).

[0106] For example, the first row of the pedestrian occurrence / concentration parameter 1410 shown in Figure 14B represents a parameter indicating that one person moves from point A to point C every minute. This is an example of a value obtained by aggregating actual data 1400. Since the actual data 1400 is collected every 30 minutes, the actual aggregation would yield a value such as 30 people moving from point A to point C every 30 minutes. By converting this to, for example, the number of occurrences per minute, a parameter indicating that one person moves from point A to point C every minute can be obtained. By using such parameters, it becomes possible to perform measurement simulations with finer temporal granularity.

[0107] The second row of the pedestrian occurrence / concentration parameter 1410 shown in Figure 14B indicates a parameter where one person moves from point A to point C via point B every two minutes. Here, point B may be, for example, a point in real space 101 corresponding to point B_806B shown in Figure 8A, i.e., a point in real space 101 corresponding to store 1_801. Similar to the example in the first row, the parameter of one person occurring every two minutes may be calculated from the aggregated result of actual data 1400 showing that 15 people occur every 30 minutes.

[0108] The third row of the pedestrian occurrence / concentration parameter 1410 shown in Figure 14B indicates a parameter where 10 people move from point C to point A each time train A arrives. For example, if the actual data 1400 is aggregated and train A arrives 4 times in 30 minutes, and 40 people move from point C to point A during those 30 minutes, this is converted to the number of people per train A arrival. By using such parameters, it becomes possible to perform accurate measurement simulations based on actual timetable data.

[0109] Figure 14C is an explanatory diagram showing a second example of pedestrian occurrence and concentration parameters generated by the measurement error correction device 1 in an embodiment of the present invention based on actual data.

[0110] The pedestrian occurrence and concentration parameter 1420 shown in Figure 14C is another example of a parameter for measurement simulation generated from actual data 1400. This parameter indicates the number of occurrences of moving objects (e.g., people) moving between predetermined points in space, and the probability that a moving object passes through a predetermined relay point between those points. For example, the pedestrian occurrence and concentration parameter 1420 includes moving object 1421, moving point 1_1422, moving point 2_1423, relay probability 1424, and number of occurrences 1425.

[0111] The moving object 1421 indicates the type of moving object (e.g., a person). Movement point 1_1422 and movement point 2_1423 indicate the start and end points of the moving object's path. The relay probability 1424 indicates the probability that a moving object traveling between the start and end points passes through a relay point. The number of occurrences 1425 indicates the number of occurrences (e.g., frequency) of moving objects traveling between the start and end points.

[0112] For example, the first row of the pedestrian occurrence / concentration parameter 1420 shown in Figure 14C represents the parameter that one person moves from point A to point C every minute. The second row represents the parameter that 10 people move from point C to point A each time train A arrives. These can be calculated in the same way as the parameters in Figure 14B.

[0113] Furthermore, the example in Figure 14C shows that the probability of passing through point B is 0.2 in both the case of moving from point A to point C and the case of moving from point C to point A. This was calculated based on data from, for example, 18 out of 90 people (30 people moving from point A to point C and 60 people moving from point C to point A in a given 30-minute period) who used point B (i.e., store 1).

[0114] For the measurement simulation, either the parameters in the format shown in Figure 14B or the parameters in the format shown in Figure 14C may be used.

[0115] Figure 14D is an explanatory diagram showing the lighting parameters generated by the measurement error correction device 1 in an embodiment of the present invention based on actual data.

[0116] Figure 14D shows an example of parameters related to sunlight as part of the lighting. For example, the lighting parameter 1430 shown in Figure 14D includes the date 1431, time zone 1432, weather 1433, and sunlight 1434. Weather 1433 and sunlight 1434 indicate the weather for the date indicated by date 1431 and the time zone indicated by time zone 1432, as well as the sunlight irradiation parameters such as the sun's altitude, azimuth, and illuminance at that time. These may be measured in real space 101 for each time zone. Based on such data, appropriate sunlight irradiation parameters can be set according to the settings of the season, time zone, and weather, etc., of the target of the measurement simulation, and a highly accurate measurement simulation can be performed.

[0117] The measurement simulation unit 21 may generate simulation conditions 34 based on parameters generated from actual data 1400, for example, as shown in Figures 14A to 14D. For example, the measurement simulation unit 21 may generate simulation conditions 34 such that the range of parameter values ​​included in the simulation conditions 34 includes the parameter values ​​generated from actual data 1400.

[0118] The parameter setting method for the measurement simulation described with reference to Figures 14A to 14D is just one example and is not mandatory. However, as described above, by setting parameters that can occur in real space 101 based on actual data 1400 in real space 101, it is possible to efficiently perform a measurement simulation to generate a correction model that corrects measurement errors based on the actual environmental conditions of real space 101.

[0119] Figure 15 is a flowchart showing the process performed by the measurement error correction unit 23 of the measurement error correction device 1 in an embodiment of the present invention.

[0120] First, measurements are taken in the physical space 101 (step 1501). This is done, for example, by sensors such as cameras installed in the physical space 101. Next, the measurement error correction unit 23 acquires environmental information during measurement (step 1502), and then acquires the measured data (step 1503). Next, the measurement error correction unit 23 calls up a correction model to correct the measurement error (step 1504), and saves the corrected measurement result 38 (step 1505). Details of the above process will be explained with reference to Figures 16 and 17, etc.

[0121] Figure 16 is an explanatory diagram showing the measurement environment information acquired by the measurement error correction device 1 in an embodiment of the present invention.

[0122] The measurement environment information 1600 shown in Figure 16 includes the date indicating the day the measurement was taken, weekday / holiday 1603 indicating whether the day was a weekday or a holiday, time 1603 indicating the time period during which the measurement was taken, store 1_1604 indicating whether the stores in the space being measured were open during that time period, weather 1605 indicating the weather during that time period, and location A generation amount 1606 and location C generation amount 1607 indicating the amount of people at each point in the space during that time period. These are identified by environment information IDs 1 to 7, respectively. These are just examples, and items that cannot be obtained from the above may be omitted, or other items (e.g., the state of sunlight) may be included if they can be obtained. In step 1502 of Figure 15, the measurement environment information 1600 is acquired for each time period and stored in the auxiliary storage device 16.

[0123] Figure 17 is an explanatory diagram showing the measurement data acquired by the measurement error correction device 1 in an embodiment of the present invention.

[0124] The measurement data 1700 shown in Figure 17 is an image captured by a camera placed in the physical space 101, which corresponds to the sensor A802 shown in Figure 8A, etc. This image includes, for example, multiple people 1701 and a store 1702. Store 1702 is a store in the physical space 101 that corresponds to store 1_801 shown in Figure 8A, etc. In step 1503 of Figure 15, the measurement data 1700 is acquired for each time period and stored in the auxiliary storage device 16.

[0125] In step 1504, the measurement error correction unit 23 selects a correction model to be used for correcting the measurement error from among several correction models, for example, shown in Figure 13. At this time, it is desirable to select a correction model that uses the setting parameters included in the measurement environment information 1600 acquired in step 1502 as explanatory variables. For example, as shown in Figure 16, if environment information IDs 1 to 7 are acquired, it is desirable to select a correction model that uses the corresponding setting parameters as explanatory variables. Alternatively, if there are multiple selectable correction models, the correction model with the highest missing value estimation accuracy 1303 may be selected.

[0126] The measurement error correction unit 23 counts the number of people in the real space 101 being measured by applying a predetermined discrimination / recognition model to the measurement data acquired in step 1503, and further estimates the measurement error by applying the correction model called in step 1504 to the measurement environment information acquired in step 1502, and obtains a measurement result with the error corrected (step 1504).

[0127] In real space, the parameter values ​​input to the correction model may not be specific to a single value, but may be obtained with a certain distribution. In such cases, the measurement error correction unit 23 may input the values ​​with that distribution into the correction model and calculate an interpolated value for the measurement error with that distribution.

[0128] Next, an example of the screen displayed by the measurement error correction device 1 will be explained with reference to Figures 18 to 20.

[0129] Figure 18 is an explanatory diagram of the measurement simulation setting screen displayed by the measurement error correction device 1 in an embodiment of the present invention.

[0130] The measurement simulation settings screen 1800 shown in Figure 18 is an example of a screen displayed by the measurement simulation unit 21 in the process shown in Figure 3. Specifically, the measurement simulation settings screen 1800 shown in Figure 18 includes a 3D spatial information setting unit 1801, an equipment / sensor information setting unit 1802, and a simulation model setting unit 1803.

[0131] In the 3D spatial information setting unit 1801, when the user specifies a file format and file path, the specified file is loaded. For example, if point cloud data is specified as the file format and a file path is specified, the corresponding 3D spatial information 31 is loaded (step 301).

[0132] The equipment / sensor information setting unit 1802 displays information corresponding to the equipment / sensor information 32 read in step 302. The user can add new information to the equipment / sensor information 32 by operating the equipment / sensor information setting unit 1802.

[0133] The simulation model setting unit 1803 displays information corresponding to the simulation model 33 loaded in step 303. The user can add new information to the simulation model 33 by operating the simulation model setting unit 1803.

[0134] Figure 19 is an explanatory diagram of the correction model generation screen displayed by the measurement error correction device 1 in an embodiment of the present invention.

[0135] The correction model generation screen 1900 shown in Figure 19 is an example of a screen displayed by the correction model generation unit 22 in the process shown in Figure 10. Specifically, the correction model generation screen 1900 shown in Figure 19 includes a sensor selection unit 1901, a measurement simulation space display unit 1902, a measurement simulation result display unit 1903, a measurement simulation screen display unit 1904, a feature quantity selection unit 1905, a correction model generation button 1906, and a correction model display unit 1907.

[0136] The user operates the sensor selection unit 1901 to select the sensor for which the correction model will be generated. The measurement simulation space display unit 1902 displays the virtual space that is the measurement target (i.e., the target of the measurement simulation) by the selected sensor. In the example in Figure 19, a plan view of the target virtual space is displayed, similar to that in Figure 8A. The measurement simulation result display unit 1903 displays the results of the executed measurement simulation. Information similar to the measurement simulation result 1100 shown in Figure 11 may be displayed here, for example.

[0137] The measurement simulation screen display unit 1904 displays, for example, the measurement results of the sensor generated in the measurement simulation (for example, images captured by camera A_802). The measured values ​​obtained as a result of this simulation and the correct values ​​may be displayed together.

[0138] The feature selection unit 1905 displays the parameters set in the measurement simulation and their importance. The user can refer to these and select the parameters (i.e., features) to be used as explanatory variables in the correction model. The selected parameters are determined to be used as features in the correction model in step 1003. At this time, the user can generate a correction model corresponding to each combination by operating the correction model generation button 1906 while changing the combination of parameters selected by the user. The correction model display unit 1907 displays information identifying the parameters to be used as explanatory variables and the estimation accuracy of missing values ​​by that correction model for each generated correction model.

[0139] Figure 20 is an explanatory diagram of the pedestrian flow measurement screen displayed by the measurement error correction device 1 in an embodiment of the present invention.

[0140] The pedestrian flow measurement screen 2000 shown in Figure 20 is an example of a screen displayed by the measurement accuracy correction unit 23 in the process shown in Figure 15. Specifically, the pedestrian flow measurement screen 2000 shown in Figure 20 includes a sensor selection unit 2001, a measurement screen display unit 2002, an environmental information display unit 2003, a measurement result display unit 2004, and a correction model change button 2005.

[0141] The user operates the sensor selection unit 2001 to select a sensor placed in the physical space 101 to be measured. The measurement screen display unit 2002 displays the measurement results from the selected sensor (for example, an image taken by a camera). The environmental information display unit 2003 displays environmental information at the time of measurement.

[0142] The measurement result display unit 2004 displays the number of people counted based on the measurement results from the sensor (i.e., the actual measured value) and the error estimated by applying the environmental information at that time to a correction model (i.e., the missing data imputed value). As shown in the example in Figure 20, the changes in the actual measured value and the missing data imputed value over time may also be displayed.

[0143] Furthermore, the user can change the applied correction model by operating the correction model change button 2005. When the correction model is changed, the missing data imputation values ​​based on the changed correction model are newly displayed on the measurement result display unit 2004.

[0144] Furthermore, the system of the embodiment of the present invention may be configured as follows.

[0145] (1) A measurement error correction system comprising: a control device (e.g., a central control device 11) that performs predetermined processing; a storage device (e.g., a main storage device 15 and an auxiliary storage device 16) accessible by the control device, wherein the control device comprises: a measurement simulation unit (e.g., a measurement simulation unit 21) that performs a simulation of measuring a moving object in a virtual space (e.g., a virtual space 111) that reproduces the real space (e.g., real space 101); a correction model generation unit (e.g., a correction model generation unit 22) that generates a correction model for correcting the measurement error of a moving object, with the parameters as explanatory variables, based on the simulation results and parameters set in the simulation; and a measurement error correction unit (e.g., a measurement error correction unit 23) that corrects the measurement results of the moving object in the real space based on the correction model.

[0146] This allows for the completion of measurements of moving objects with higher accuracy by supplementing missing data according to the measurement environment.

[0147] (2) In (1) above, the storage device holds spatial information that reproduces the real space (e.g., 3D spatial information 31) and sensor information that reproduces the position, measurement direction and specifications of measurement sensors placed in the real space (e.g., equipment / sensor information 32), and the measurement simulation unit generates a moving object in a virtual space that reproduces the real space based on the spatial information, based on predetermined parameters, and performs a simulation to measure the moving object based on predetermined parameters and sensor information (e.g., step 305).

[0148] This allows for accurate reproduction of the real world within a virtual space, enabling highly accurate simulations.

[0149] (3) In (2) above, the measurement simulation unit measures the moving object using the same method as the measurement method using measurement sensors placed in the real space (for example, by applying the same model used for recognizing images taken in the real space to the images acquired in the simulation).

[0150] This allows for accurate reproduction of measurements taken in the real world within a virtual space.

[0151] (4) In (2) above, the measurement simulation unit reproduces the movement of a moving object that occurred based on predetermined parameters, based on a movement model (e.g., pedestrian model 640) that reproduces the movement of a moving object in real space.

[0152] This makes it possible to accurately reproduce the movement of objects in the real world within a virtual space.

[0153] (5) In (4) above, the moving object is a pedestrian, the movement model that reproduces the movement of the moving object is a model that reproduces the walking of a pedestrian (e.g., pedestrian model 640), and the parameters set in the simulation include parameters relating to at least one of the pedestrian generation rate, lighting, shops, and means of transportation in the virtual space (e.g., parameters included in simulation condition 33).

[0154] This allows us to generate a model that estimates the measurement error of pedestrians in real space.

[0155] (6) In (2) above, the measurement simulation unit generates multiple combinations of parameter values ​​(for example, multiple combinations of setting parameters, each identified by simulation number 701), and performs multiple simulations based on the multiple combinations of parameter values ​​(for example, step 305).

[0156] This allows for measurement simulations under various environmental conditions.

[0157] (7) In (6) above, the memory device holds actual values ​​of parameters acquired in real space in the past (e.g., actual data 1400), and the measurement simulation unit generates multiple combinations of parameter values ​​based on the actual values ​​of the parameters.

[0158] This allows for measurement simulations under conditions that can occur in real space, and enables the generation of highly accurate correction models in realistic environments.

[0159] (8) In (2) above, the measurement simulation unit performs a simulation to measure the number of moving objects and obtains the difference between the number of moving objects generated and the number of moving objects measured as the measurement error (for example, Figure 3), the correction model generation unit generates a correction model to correct the measurement error of the number of moving objects measured (for example, Figure 10), and the measurement error correction unit corrects the number of moving objects measured in the real space based on the correction model (for example, Figure 15).

[0160] This allows for highly accurate measurement of the number of moving objects.

[0161] (9) In (2) above, the correction model generation unit generates a correction model using machine learning, in which the parameters are explanatory variables and the measurement error is the dependent variable.

[0162] This allows for the generation of a correction model based on the simulation results.

[0163] (10) In (9) above, the correction model generation unit reduces the explanatory variables of the correction model by generating a model that estimates at least one of the multiple parameters based on at least one of the other parameters.

[0164] This reduces the number of parameters that need to be acquired in the real world.

[0165] (11) In (9) above, the sensor information includes information for reproducing multiple measurement sensors placed in real space (for example, sensor type, specifications, placement position and placement angle, etc.), and the correction model generation unit generates a correction model that includes the measurement results of other measurement sensors as explanatory variables, as a correction model for correcting the measurement error of any of the multiple measurement sensors.

[0166] This allows for the generation of highly accurate correction models.

[0167] (12) In (9) above, if information specifying the parameters to be used as explanatory variables is input to the correction model generation unit (for example, selection based on the feature selection unit 1905 in Figure 19 in step 1003), the unit generates a correction model using the specified parameters as explanatory variables.

[0168] This allows the user to select parameters.

[0169] (13) In (9) above, the correction model generation unit calculates importance (for example, step 1002) which indicates the magnitude of the influence that the parameter values ​​have on the magnitude of the measurement error obtained by the simulation, and generates a correction model in which the parameters selected based on importance (for example, the parameters selected in the feature selection unit 1905 in Figure 19) are used as explanatory variables.

[0170] This allows for the selection of appropriate parameters based on the simulation results.

[0171] (14) In (13) above, the correction model generation unit outputs information (for example, the feature selection unit 1905 in Figure 19) that displays the multiple parameters used in the simulation and the importance of each parameter.

[0172] This allows users to select appropriate parameters based on the simulation results.

[0173] (15) In (2) above, the measurement error correction unit corrects the measurement results of a moving object by a measurement sensor in real space by applying the parameter values ​​obtained from real space to the correction model.

[0174] This allows for the completion of missing data based on a correction model, thereby improving the accuracy of measurements of moving objects.

[0175] (16) In (15) above, the moving object is a pedestrian, and the parameters obtained from real space include parameters relating to the number of pedestrians, lighting, shops, and at least one of the means of transport.

[0176] This allows for more accurate measurement of pedestrians based on appropriate environmental factors.

[0177] (17) In (15) above, if the measurement error correction unit obtains a parameter value having a distribution from the real space, it corrects the value to have a distribution by applying the parameter value obtained from the real space to the correction model.

[0178] This allows for correction of measurement errors even when reliable parameter values ​​cannot be obtained.

[0179] (18) In (15) above, the measurement error correction unit corrects the measurement result of the moving object at each time based on the measurement result of the moving object at each time and the parameter values ​​acquired at each time, and outputs information (for example, the measurement result display unit 2004 in Figure 20) that displays the measurement result of the moving object and the change in the correction amount by the correction model.

[0180] This allows us to observe the changes in measurement results and estimated measurement errors.

[0181] It should be noted that the present invention is not limited to the embodiments described above, and various modifications are included. For example, the embodiments described above are explained in detail for a better understanding of the present invention, and are not necessarily limited to those having all of the configurations described. Furthermore, it is possible to replace parts of the configuration of one embodiment with the configuration of another embodiment, and it is possible to add configurations from other embodiments to the configuration of one embodiment. In addition, it is possible to add, delete, or replace parts of the configuration of each embodiment with other configurations.

[0182] Furthermore, each of the above configurations, functions, processing units, and processing means may be implemented in hardware, either partially or entirely, by designing them as integrated circuits, for example. Alternatively, each of the above configurations and functions may be implemented in software by a processor interpreting and executing programs that implement each function. Information such as programs, tables, and files that implement each function can be stored in storage devices such as non-volatile semiconductor memory, hard disk drives, and SSDs (Solid State Drives), or in computer-readable non-temporary data storage media such as IC cards, SD cards, and DVDs.

[0183] Furthermore, the control lines and information lines shown are those deemed necessary for explanation purposes, and do not necessarily represent all control lines and information lines in the actual product. In practice, it can be assumed that almost all components are interconnected. [Explanation of Symbols]

[0184] 1. Measurement error correction device 2 External system 3. External Servers 4 Network 11 Central Control Unit 12 Input devices 13 Output device 14. Communication equipment 15 Main memory 16 Auxiliary storage 21 Measurement and Simulation Department 22 Correction Model Generation Unit 23 Measurement Error Correction Unit 31 3D spatial information 32 Equipment and Sensor Information 33 Simulation Models 34 Simulation Conditions 35 Simulation Results 36. Evaluation results of factors causing missing data 37 Correction Model Information 38 Measurement Results

Claims

1. A measurement error correction system, The system comprises a control device that performs a predetermined process, and a storage device accessible by the control device. The storage device holds spatial information that reproduces the real space, and sensor information that reproduces the position, measurement direction, and specifications of the measurement sensors placed in the real space. The control device includes a measurement simulation unit which performs an agent simulation that generates moving objects in a virtual space that reproduces the real space based on the spatial information, and a measurement simulation that measures the moving objects generated by the agent simulation based on the spatial information and the sensor information, and obtains the difference between the number of moving objects generated by the agent simulation and the number of moving objects measured by the measurement simulation as a measurement error. The control device includes a correction model generation unit that generates a correction model for correcting the measurement error, using the parameters set in the agent simulation and the measurement simulation as explanatory variables, based on the results of the agent simulation and the measurement simulation, and the parameters set in the agent simulation and the measurement simulation. A measurement error correction system characterized in that the control device includes a measurement error correction unit that corrects the number of moving objects measured in the real space based on the correction model.

2. A measurement error correction system according to Claim 1, The measurement simulation unit is characterized in that, in the measurement simulation, it measures the number of moving objects using the same method as the measurement method using the measurement sensors placed in the real space.

3. A measurement error correction system according to Claim 1, The measurement error correction system is characterized in that the measurement simulation unit reproduces the movement of the moving object that occurred in the agent simulation based on a movement model that reproduces the movement of the moving object in the real space.

4. A measurement error correction system according to claim 3, The aforementioned moving body is a pedestrian, The movement model that reproduces the movement of the aforementioned moving object is a model that reproduces the walking of the aforementioned pedestrian, A measurement error correction system characterized in that the parameters set in the agent simulation and the measurement simulation include parameters relating to at least one of the pedestrian population, lighting, shops, and means of transportation in the virtual space.

5. A measurement error correction system according to Claim 1, The measurement simulation unit generates multiple combinations of the parameter values, A measurement error correction system characterized by performing the measurement simulation multiple times based on multiple combinations of the values ​​of the aforementioned parameters.

6. A measurement error correction system according to claim 5, The memory device stores the actual values ​​of the parameters acquired in the real space in the past, The measurement error correction system is characterized in that the measurement simulation unit generates a plurality of combinations of parameter values ​​based on the actual values ​​of the parameters.

7. A measurement error correction system according to Claim 1, The measurement error correction system is characterized in that the correction model generation unit generates a correction model using machine learning, with the parameters as explanatory variables and the measurement error as the dependent variable.

8. A measurement error correction system according to claim 7, The measurement error correction system is characterized in that the correction model generation unit generates a model that estimates at least one of the multiple parameters which are explanatory variables of the correction model based on at least one of the other parameters.

9. A measurement error correction system according to claim 7, The sensor information includes information for reproducing the plurality of measurement sensors arranged in the real space, The measurement error correction system is characterized in that the correction model generation unit generates a correction model that includes the measurement results of other measurement sensors as explanatory variables, as the correction model for correcting the measurement error of any of the plurality of measurement sensors.

10. A measurement error correction system according to claim 7, The measurement error correction system is characterized in that, when information specifying parameters to be used as explanatory variables is input to the correction model generation unit, it generates the correction model using the specified parameters as explanatory variables.

11. A measurement error correction system according to claim 7, The correction model generation unit, The importance of the influence that the parameter values ​​have on the magnitude of the measurement error obtained by the measurement simulation is calculated. A measurement error correction system characterized by generating a correction model in which the parameters selected based on the aforementioned importance are used as explanatory variables.

12. A measurement error correction system according to claim 11, The measurement error correction system is characterized in that the correction model generation unit outputs information displaying a plurality of parameters used in the agent simulation and the measurement simulation, and the importance of each of the parameters.

13. A measurement error correction system according to claim 1, The measurement error correction system is characterized in that the measurement error correction unit corrects the measurement results of the moving object by the measurement sensor in the real space by applying the parameter values ​​obtained from the real space to the correction model.

14. A measurement error correction system according to claim 13, The aforementioned moving object is a pedestrian, A measurement error correction system characterized in that the parameters obtained from the real space include parameters relating to at least one of the pedestrian population, lighting, shops, and means of transportation.

15. A measurement error correction system according to claim 13, The measurement error correction system is characterized in that, when the measurement error correction unit obtains the parameter values ​​having a distribution from the real space, it corrects them to values ​​having a distribution by applying the parameter values ​​obtained from the real space to the correction model.

16. A measurement error correction system according to claim 13, The measurement error correction system is characterized in that the measurement error correction unit corrects the measurement result of the moving object at each time based on the measurement result of the moving object at each time and the value of the parameter acquired at each time, and outputs information that displays the measurement result of the moving object and the change in the correction amount according to the correction model.

17. A measurement error correction method performed by a measurement error correction system, The measurement error correction system comprises a control device that performs a predetermined process and a storage device accessible by the control device. The storage device holds spatial information that reproduces the real space, and sensor information that reproduces the position, measurement direction, and specifications of the measurement sensors placed in the real space. The aforementioned measurement error correction method is: The control device performs an agent simulation that generates moving objects in a virtual space that reproduces the real space based on the spatial information, and a measurement simulation that measures the moving objects generated by the agent simulation based on the spatial information and the sensor information, and obtains the difference between the number of moving objects generated by the agent simulation and the number of moving objects measured by the measurement simulation as a measurement error. The control device provides a correction model generation procedure that generates a correction model for correcting the measurement error, using the parameters set in the agent simulation and the measurement simulation as explanatory variables, based on the results of the agent simulation and the measurement simulation, and the parameters set in the agent simulation and the measurement simulation. A measurement error correction method characterized in that the control device includes a measurement error correction procedure for correcting the number of moving objects measured in the real space based on the correction model.

Images