Path estimation system

Abstract

This technology provides an easy way to reduce the discrepancies in coordinate position and time between multiple sensor data. [Solution] The path estimation system comprises: an acquisition unit that acquires multiple detection data related to the movement path of a subject from each of multiple sensors; a path estimation unit that generates multiple movement path data representing the movement path of the subject according to each of the multiple detection data; an extraction unit that extracts multiple specific data indicating that the subject moved along a predetermined specific path from each of the multiple movement path data; and a correction unit that compares the multiple specific data and performs a correction process to reduce the discrepancy between coordinate positions on the coordinate axis and the discrepancy between times on the time axis in the multiple movement path data.

Description

[Technical Field]

[0001] This disclosure relates to the technology of a path estimation system. [Background technology]

[0002] Conventionally, there is a known technology that uses sensor data acquired by multiple sensors to recognize the actions being taken by an operator (for example, Patent Document 1). [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2018-147153 [Overview of the project] [Problems that the invention aims to solve]

[0004] When generating multiple travel path data sets using detection data from multiple sensors, discrepancies in coordinate positions on the coordinate axis and time discrepancies on the time axis may occur between them. In this case, it may become difficult to perform various analyses using multiple travel path data sets with high accuracy. Therefore, there is a need for a technology that can easily reduce the discrepancies in coordinate positions and time between multiple travel path data sets. [Means for solving the problem]

[0005] This disclosure can be implemented in the following forms:

[0006] (1) According to one embodiment of the present disclosure, a path estimation system is provided. The path estimation system comprises: an acquisition unit that acquires a plurality of detection data relating to the movement path of a subject from each of a plurality of sensors; a path estimation unit that generates a plurality of movement path data representing the movement path of the subject according to each of the plurality of detection data, wherein each of the plurality of movement path data is defined by a coordinate axis for representing the position of each point in the movement path and a time axis for representing the time at each point; an extraction unit that extracts a plurality of specific data from each of the plurality of movement path data indicating that the subject has moved along a predetermined specific path; and a correction unit that compares the plurality of specific data and performs a correction process to reduce the discrepancy between coordinate positions on the coordinate axis and the discrepancy between times on the time axis in the plurality of movement path data. According to the above configuration, correction processing can be easily performed by comparing multiple specific data points that indicate movement along a specific route. (2) In the above embodiment, the correction unit may, in the correction process, correct the other movement path data such that the difference between the coordinate position on the coordinate axis and the time on the time axis in the other movement path data and the specific movement path data generated using the detection data of a specific sensor among the multiple sensors is reduced. According to the above configuration, other movement path data can be easily corrected based on specific movement path data generated using a specific sensor. (3) In the above configuration, the specific route may consist of a first straight route and a second straight route which includes the intersection point with the first straight route as its starting point and bends relative to the first straight route. According to the above configuration, since the specific route is a simple route that includes a first straight-line route and a second straight-line route, it is possible to easily compare multiple specific data points. (4) In the above aspect, in the correction process, the correction unit may correct the time on at least one of the time axes of the plurality of movement route data such that the difference between the times indicated by at least one piece of time data among the start time data at the movement start point of the specific route, the intersection time data at the intersection point, and the end time data at the movement end point of the specific route in the plurality of specific data is minimized. At the movement start point, intersection point, and movement end point, the movement of the worker temporarily stops. Therefore, according to the above aspect, it is possible to easily grasp the correspondence relationship of the times indicated by the start time data, the intersection time data, and the end time data between the plurality of specific data, so that the time correction can be easily performed. (5) In the above aspect, in the correction process, the correction unit may compare the plurality of specific data, and correct the coordinate position on at least one of the plurality of coordinate axes in the plurality of movement route data such that the difference between the coordinate positions indicating the respective points possessed by the plurality of specific data is minimized. According to the above aspect, the coordinate position can be easily corrected by comparing the plurality of specific data. Note that the present disclosure can be implemented in forms such as a route estimation method in addition to a route estimation system. [Brief Description of the Drawings]

[0007] [Figure 1] An explanatory diagram showing the schematic configuration of a route estimation system. [Figure 2] A diagram for explaining a method of generating first movement route data using first detection data. [Figure 3] A diagram for explaining a specific route. [Figure 4] A flowchart of correction processing executed by the route estimation system. [Figure 5] A diagram for explaining a plurality of movement route data. [Figure 6] A diagram for explaining specific data. [Figure 7] A diagram for explaining step S40 and step S60. [Figure 8] A diagram showing a specific route of the first other embodiment. [Mode for Carrying Out the Invention]

[0008] A. Embodiment: Figure 1 is an explanatory diagram showing a schematic configuration of a route estimation system 10. The route estimation system 10 estimates the movement route RT of a worker WK. The worker WK moves around a factory FC to perform various types of work, such as work related to vehicle assembly, work related to component installation, and work related to vehicle inspection. The route estimation system 10 may provide the worker with instructions related to various types of work according to each point on the estimated movement route RT.

[0009] The route estimation system 10 includes a plurality of sensors 20, 52, a route estimation device 100, an input device 200, and a display device 300.

[0010] The plurality of sensors 20, 52 are each used to detect the movement route RT of the worker WK who is the subject. The plurality of sensors 20, 52 are of different types. The plurality of sensors 20, 52 are a camera 20 and an inertial measurement unit 52 (IMU: Inertial Measurement Unit).

[0011] The camera 20 is a monocular camera. The position of the camera 20 is fixed. The camera 20 captures an image of the work area where the worker WK is present from above, and generates an image of the work area including the worker WK as first detection data. The camera 20 captures images of the work area at regular intervals to generate the first detection data. The camera 20 has a timer function for measuring time. The first detection data includes first time data indicating the time when the image was captured. The camera 20 transmits the first detection data to the route estimation device 100 at predetermined timings via wireless communication.

[0012] The inertial measurement device 52 includes an acceleration sensor that detects acceleration in three axes and an angular velocity sensor that detects angular velocity around three axes. In other words, in this embodiment, the inertial measurement device 52 generates acceleration and angular velocity as second detection data, which are physical quantities related to the motion state of the worker WK. The inertial measurement device 52 also has a timer function for measuring time. The inertial measurement device 52 detects acceleration and angular velocity at regular intervals and generates second detection data. The second detection data includes second time data indicating the time when acceleration and angular velocity were detected. The inertial measurement device 52 is fixed to the torso of the work clothes worn by the worker WK. The inertial measurement device 52 transmits the second detection data to the path estimation device 100 at predetermined intervals via wireless communication.

[0013] The path estimation device 100 acquires first detection data and second detection data, which are multiple detection data related to the worker WK's movement path RT, from each of the multiple sensors 20 and 52, and estimates the worker WK's movement path RT.

[0014] The path estimation device 100 is a computer comprising a memory 101, an input / output interface 102, a processor 103, an internal bus 104, and a communication device 105. The memory 101, the input / output interface 102, and the processor 103 are connected via the internal bus 104 to enable bidirectional communication.

[0015] Memory 101 stores a program 160 used for various processes performed by the path estimation device 100. The input / output interface 102 is connected to a communication device 105, an input device 200, and a display device 300. The communication device 105 can communicate with sensors 20 and 52 via wired or wireless communication. The input device 200 is, for example, a keyboard or mouse. The display device 300 is, for example, a liquid crystal display.

[0016] The processor 103 of the path estimation device 100 has the following functional units by executing the program 160: acquisition unit 110, path estimation unit 120, extraction unit 125, correction unit 130, and output unit 140.

[0017] The acquisition unit 110 acquires multiple detection data related to the movement path RT of worker WK from each of the multiple sensors 20 and 52. In this embodiment, the acquisition unit 110 acquires first detection data from the camera 20 at predetermined intervals and second detection data from the inertial measuring device 52 at predetermined intervals.

[0018] The path estimation unit 120 generates multiple movement path data representing the movement path RT of worker WK, corresponding to each of the multiple detection data. In this embodiment, the path estimation unit 120 generates first movement path data representing the estimated movement path RT using multiple first detection data acquired from the camera 20 at predetermined timings. The path estimation unit 120 also generates second movement path data representing the estimated movement path RT using multiple second detection data acquired from the inertial measuring device 52 at predetermined timings. Each of the first and second movement path data is defined by a coordinate axis for representing the position of each point (detection point) along the movement path and a time axis for representing the time at each point. The first and second movement path data are data that represent the position of worker WK in time series. That is, each of the first and second movement path data has the coordinate position of worker WK on a two-dimensional plane and the time (duration) when worker WK is located at the coordinate position.

[0019] Figure 2 illustrates a method for generating first movement path data using first detection data acquired from camera 20. The image data acquired from camera 20 represents a view of the work area from diagonally above. The path estimation unit 120 uses a skeleton detection model MD to detect the skeleton BN of worker WK from the first detection data. The skeleton BN includes multiple key points KP that indicate the positions of facial features and joint points. In this embodiment, the multiple key points KP include key points KP that indicate the end positions of the left and right legs, more specifically, the positions of the left and right heels. The path estimation unit 120 identifies the heel midpoint PM, which is the center position of the left and right heels. However, if the key point KP indicating the end position of the leg is a key point KP indicating the ankle instead of the heel, the path estimation unit 120 may identify the center positions of the left and right ankles instead of the center positions of the left and right heels. The path estimation unit 120 converts the coordinate position of the heel midpoint PM in the image coordinate system to the coordinate position of the heel midpoint PM in the first work area coordinate system. The image coordinate system is a two-dimensional orthogonal coordinate system having an H axis parallel to the horizontal direction of the image and a V axis parallel to the vertical direction of the image, and the first work area coordinate system is a two-dimensional orthogonal coordinate system having an X axis and a Y axis parallel to the floor surface of the work area WR. The path estimation unit 120 can convert the coordinate position of the heel midpoint PM in the image system to the coordinate position P(X,Y) of the heel midpoint PM in the first work area coordinate system by performing an affine transformation based on the correspondence between each position on the floor surface of the actual work area and each of the above positions in the image, thereby converting an image of looking down on the work area from diagonally above to an image of looking down on the work area from directly above. The path estimation unit 120 associates the time T detected in the first detection data with the coordinate position P(X,Y) in the first work area coordinate system. The path estimation unit 120 generates first movement path data by arranging the coordinate positions P(X,Y) of worker WK in the first work area coordinate system in a time series. However, the method of generating first movement path data using first detection data acquired from camera 20 is not limited to the above. For example, the path estimation unit 120 may estimate the coordinate position of worker WK based on the positional relationship between worker WK and a reference object whose position in the work area is known, using the image data contained in the first detection data.

[0020] The path estimation unit 120 shown in Figure 1 generates second movement path data using second detection data acquired from the inertial measurement device 52 in the following manner. The second detection data includes the acceleration of the three axes and the angular velocity around the three axes in the sensor coordinate system. The path estimation unit 120 calculates the acceleration of worker WK in the second work area coordinate system using the second detection data, and calculates the coordinate position p(x,y) of worker WK in the second work area coordinate system based on the acceleration of worker WK in the second work area coordinate system. The second work area coordinate system is a two-dimensional Cartesian coordinate system having x and y axes parallel to the floor surface of the work area WR. As is well known, velocity can be obtained by integrating acceleration over time, and the amount of movement can be obtained by integrating velocity over time. If the coordinate position before movement is known, the coordinate position after movement can be obtained from the coordinate position before movement and the amount of movement. The path estimation unit 120 associates the time t detected in the second detection data with the coordinate position P(x,y) in the second work area coordinate system. The path estimation unit 120 generates second movement path data by arranging the coordinate positions p(x, y) of worker WK in the second work area coordinate system in a time series.

[0021] The extraction unit 125 extracts multiple specific data from each of the multiple movement path data to indicate that worker WK moved along a predetermined specific path CR. Figure 3 is a diagram illustrating the specific path CR. The specific path CR consists of a first straight path SR1 and a second straight path SR2. The first straight path SR1 and the second straight path SR2 intersect at a right angle. The starting point of the first straight path SR1 is position SP1, and its ending point is position SP2. The starting point of the second straight path SR2 is position SP2, and its ending point is position SP3. In other words, position SP2 is the point where the specific path CR bends midway, and is the intersection point where the first straight path SR1 and the second straight path SR2 intersect. In this embodiment, position SP1 is the starting point of movement along the specific path CR. Position SP3 is the ending point of movement along the specific path CR. For example, the lengths of the first straight path SR1 and the second straight path SR2 are the same, each being 1m. In this embodiment, the specific route CR may be set as the first route of the travel route RT on which worker WK is scheduled to move, or it may be generated as an intermediate route, or it may be set as the last route. The extraction unit 125 extracts first specific data indicating the specific route CR from the first travel route data. The extraction unit 125 also extracts second specific data indicating the specific route CR from the second travel route data. The extraction unit 125 identifies the first specific data and the second specific data, for example, by identifying a route that turns at a right angle from the estimated travel route RT.

[0022] The correction unit 130 shown in Figure 1 compares multiple specific data and performs a correction process to reduce the discrepancy between coordinate positions on the coordinate axes and the discrepancy between times on the time axis in multiple movement path data. In this embodiment, the correction unit 130 compares the first specific data and the second specific data and performs a correction process to reduce the discrepancy in the coordinate axes between the first movement path data and the second movement path data, and the discrepancy in the time axis between the first movement path data and the second movement path data. Details of the correction process will be described later.

[0023] The output unit 140 displays various information on the display device 300. For example, the output unit 140 displays the movement path data after correction processing on the display device 300.

[0024] Figure 4 is a flowchart of the correction process performed by the path estimation system 10. Figure 5 is a diagram illustrating multiple travel path data RDGs. The correction process is initiated when the path estimation device 100 receives an instruction to perform the correction process from an external source, such as worker WK. Once the correction process has started, it is determined that worker WK will travel along a specific path CR during a predetermined data collection period.

[0025] First, in step S10, the acquisition unit 110 acquires detection data from multiple sensors 20 and 52 at predetermined intervals. In this embodiment, in step S10, the acquisition unit 110 acquires first detection data from the camera 20 and second detection data from the inertial measuring device 52 during a predetermined data collection period. The acquired first and second detection data are stored in the memory 101.

[0026] Next, in step S20, the path estimation unit 120 generates multiple movement path data RDGs according to each of the multiple sensors 20 and 52. In this embodiment, as described above, the path estimation unit 120 generates a first movement path data RD1 and a second movement path data RD2. The generated multiple movement path data RDGs are stored in memory 101. As shown in Figure 5, each point P1 to Px on the movement path of the first movement path data RD1 is defined by the coordinate axes X and Y of the first work area coordinate system and the time axis T representing the dimension of time. For example, point P1 in the first movement path data RD1 that corresponds to a certain point on the actual movement path RT has a coordinate position of (X1, Y1) and a time of time T1. Each point P1 to Px in the first movement path data RD1 is stored in memory 101 as Pz(X, Y, T), which is represented by the position on the X axis, the position on the Y axis, and the position on the time axis T. The subscript "z" is an integer of 1 or more and is a number used to identify each point. Each point p1 to px in the second movement path data RD2 is defined by the coordinate axes x and y of the second work area coordinate system and the time axis t, which represents the dimension of time. For example, point p1 in the second movement path data RD2 that corresponds to a certain point in the actual movement path RT has coordinate position (x1, y1) and time t1. Each point p1 to px in the second movement path data RD2 is stored in memory 101 as pz(x, y, t), which is represented by the position on the x axis, the position on the y axis, and the position on the time axis t. The subscript "z" is an integer of 1 or more and is a number used to identify each point.

[0027] As shown in Figure 4, in step S30, the extraction unit 125 extracts multiple specific data RS1 and RS2 from each of the multiple movement path data RD1 and RD2, indicating that the specific path CR was traveled. Figure 6 is a diagram illustrating the specific data RS1 and RS2. In step S30, the extraction unit 125 arranges the coordinate positions of each point P1 to Px in the first work area coordinate system in chronological order and extracts points that form right angles by connecting adjacent coordinate positions with straight lines. In Figure 6, the extraction unit 125 identifies that point Pn corresponds to position SP2 of the specific path CR shown in Figure 5. Then, in the first movement path data RD1, point Pn A point P located up to 1 m before in chronological order n-1 and the point P n A point P located up to 1 m ahead in chronological order n+1 are specified. In addition, in FIG. 6, the first specified data RS1 is a set of three points P n-1 , P n , P n+1 , but may be configured as a set of more than three points.

[0028] In step S30, the extraction unit 125 arranges the coordinate positions of the points p1 to P in the second work area coordinate system in chronological order and connects adjacent coordinate positions with straight lines, thereby extracting points forming a right angle. In FIG. 6, the extraction unit 125 specifies that the point p x corresponds to the position SP2 of the specific route CR shown in FIG. 5. Then, in the second movement route data RD2, the extraction unit 125 n specifies that the point p corresponds to the position SP2 of the specific route CR shown in FIG. 5. Then, the extraction unit 125, in the second movement route data RD2, n identifies a point p located up to 1 m before in chronological order n-1 and the point p n a point p located up to 1 m ahead in chronological order n+1 . In addition, in FIG. 6, the second specified data RS2 is a set of three points p n-1 , p n , p n+1 , but may be configured as a set of more than three points.

[0029] In addition, when it is defined that the worker WK first moves along the specific route CR simultaneously with the start of a predetermined data collection period, the extraction unit 125 can specify the first points P1, p1 in chronological order in each of the first movement route data RD1 and the second movement route data RD2 as the point SP1 of the specific route CR. Then, the extraction unit 125 may extract the specified data RS1, RS2 based on the first points P1, p1 in chronological order.

[0030] As shown in Figure 4, in steps S40 to S70, the correction unit 130 compares a plurality of specific data RS1 and RS2 and performs coordinate position and time correction processing. The correction unit 130 corrects the second movement path data RD2, which is other movement path data generated using the second detection data of another sensor, the inertial measuring device 52, based on the first movement path data RD1, which is specific movement path data generated using the first detection data of a specific sensor, the camera 20. In this embodiment, the correction unit 130 corrects the second movement path data RD2 so that the difference between the coordinate position p(x,y) of each point p (each data) on the coordinate axis (x,y) in the second movement path data RD2 and the coordinate position P(X,Y) of each point P (each data) on the coordinate axis (X,Y) in the first movement path data RD1 is reduced. Furthermore, in this embodiment, the correction unit 130 corrects the second travel path data RD2 so that the discrepancy between the time p(t) of each point P (each data) on the time axis T in the second travel path data RD2 and the time p(T) of each point P (each data) on the time axis T in the first travel path data RD1 is reduced.

[0031] In step S40, the correction unit 130 calculates a first correction value (coordinate transformation formula) for transforming the coordinate positions p(x,y) of the second movement path data RD2 so that the difference between the coordinate positions p(x,y) of each point p (each data) in the first movement path data RD1 and the coordinate positions p(x,y) of each point p in the second movement path data RD2 is minimized. This correction value may be expressed as a matrix equation for performing the coordinate transformation. In this embodiment, in step S40, the correction unit 130 calculates a first correction value for transforming the coordinate positions p(x,y) of the second specific data RS2 so that the difference between the coordinate positions p(x,y) of each point p in the second specific data RS2 and the first specific data RS1 is minimized. Specifically, the correction unit 130 calculates a first correction value (coordinate transformation formula) that minimizes the discrepancy between the coordinate position p(x,y) of each point p in the second specific data RS2 and the coordinate position P(X,Y) of each point P in the first specific data RS1, which is part of the first specific travel path data RD1, as specific travel path data, on the coordinate axis (X,Y), i.e., the degree of agreement is highest.

[0032] Figure 7 is a diagram illustrating steps S40 and S60. As shown in Figure 7, the correction unit 130 compares the first specific data RS1 and the second specific data RS2 and determines the location P of the first specific data RS1. n-1 , P n , P n+1 The coordinate positions indicated by and each point P in the second specific data RS2 n-1 , P n , P n+1 Each point p corresponds to n-1 , p n , p n+1 The correction unit 130 calculates a correction value (coordinate transformation formula) for transforming the coordinate axes according to the second detection data so that the difference with the coordinate position indicated by the first detection data is minimized. In other words, the correction unit 130 calculates a first correction value for rotating or translating each coordinate position of the coordinate axes (x, y) of the second specific data RS2 to match each coordinate position of the coordinate axes (X, Y) of the first specific data RS1. The calculated first correction value is stored in the memory 101.

[0033] As shown in Figure 4, in step S50, the correction unit 130 corrects the coordinate positions of each point in the second movement path data RD2 using the first correction value calculated in step S40. That is, the correction unit 130 performs a coordinate transformation to bring the coordinate axes (x,y) of the second movement path data RD2 closer to the coordinate axes (X,Y) of the first movement path data RD1.

[0034] Next, in step S60, the correction unit 130 calculates the amount of discrepancy between the time of the first specific data RS1 and the time of the second specific data RS2 at a specific timing of the specific route CR, and calculates a second correction value that minimizes the amount of discrepancy. As shown in Figure 7, in this embodiment, the specific timing is the time Tn and tn indicated by the intersection time data, which is the time data of points Pn and pn representing the intersection point SP2. The correction unit 130 calculates a second correction value that corrects the time on the time axis t of the second movement route data RD2 so that the difference between the times Tn and tn indicated by the intersection time data is minimized, i.e., zero. The calculated second correction value is stored in the memory 101. For example, if time tn is 0.1 seconds earlier than time Tn, the second correction value is "+0.1". In other embodiments, the specific timing is the time T indicated by the start time data at the movement start point SP1 of the specific route CR. n-1 , t n-1 And the time T shown by the intersection time data at intersection point SP2. n ,t n And the time T indicated by the accommodation time data at the end point of movement SP3. n+1 , t n+1 A second correction value may be calculated such that the difference between the times indicated by at least one time data point is minimized.

[0035] Next, in step S70, the correction unit 130 corrects the time at each point in the second travel path data RD2 using the second correction value calculated in step S60. That is, the correction unit 130 performs a correction to bring the time axis t of the second travel path data RD2 closer to the time axis T of the first travel path data RD1.

[0036] Steps S60 and S70 may be executed before steps S40 and S50, or steps S50 and S70 may be executed after the first and second correction values ​​have been calculated by steps S40 and S60. Furthermore, the first and second correction values ​​calculated by the correction process shown in Figure 4 may be used to correct the second movement path data RD2 generated using the second detection data even after the flowchart in Figure 4 has finished.

[0037] According to the above embodiment, by comparing multiple specific data RS1, RS1 indicating movement along a specific path CR, correction processing can be easily performed without comparing all point data of the first movement path data RD1 and the second movement path data RD2. Furthermore, according to the above embodiment, the second movement path data RD2, which is other movement path data, can be easily corrected using the first movement path data RD1, which is specific movement path data generated using the camera 20 as a specific sensor, as a reference. Here, it is generally preferable to use data that is expected to have a higher accuracy in path estimation as the specific movement path data used as a reference. In this embodiment, the second movement path data RD2 is estimated by calculating the amount of movement by integrating the acceleration of the second detection data of the inertial measurement device 52 twice with respect to time. Therefore, since the detection error tends to be gradually added in the second movement path data RD2, the estimation error of the movement path is generally larger than that of the first movement path data RD1. Therefore, in this embodiment, it is preferable that the specific movement path data used as a reference is the first movement path data RD1.

[0038] Furthermore, according to the above embodiment, as shown in Figure 3, since the specific path CR is a simple path that includes the first straight path SR1 and the second straight path SR2, it is possible to easily compare multiple specific data RS1 and RS2. In other words, the first and second correction values ​​can be calculated with less computation.

[0039] As shown in Figure 5, the worker WK's movement (walking) temporarily stops at the starting point SP1, the intersection point SP2, and the ending point SP3. In other words, unlike points along a straight path, at the start and end points of a straight path, and at intersections of straight paths, the worker WK temporarily stops moving in order to change direction. Therefore, the correspondence between the start time data, intersection time data, and end time data between multiple specific data points RS1 and RS2 can be easily grasped, making it easy to correct the time (time axis). Furthermore, according to the above embodiment, the coordinate position (coordinate axis) can be easily corrected by comparing multiple specific data points RS1 and RS2. In addition, according to the above embodiment, since the first correction value related to the coordinate position and the second correction value related to time are calculated independently, the amount of computation can be reduced compared to calculating correction values ​​to reduce the discrepancy between the coordinate position and time simultaneously.

[0040] B. Other embodiments: B-1. Other Embodiments 1: In the above embodiment, the specific path CR was a path that included a point SP2 where it turned at a right angle, as shown in Figure 3, but it is not limited to this. The specific path CR can be any combination of a turning point and a straight path. Figure 8 shows specific paths CRa, Rb, and Rc of another embodiment 1. In the specific path CRa, the angle α1 at which the first straight path SR1 and the second straight path SR2 intersect is less than 90°. For example, the angle α1 is in the range of 70° to 85°. In the specific path CRb, the angle α2 at which the first straight path SR1 and the second straight path SR2 intersect is greater than 90°. For example, the angle α2 is in the range of 115° to 125°. In the specific path CRc, the path consists of the first straight path SR1, the second straight path SR2, and the third straight path SR3. The first straight path SR1 and the second straight path SR2 intersect at a right angle, and the second straight path SR2 and the third straight path SR3 intersect at a right angle. Worker WK moves along the first straight path SR1 to the third straight path SR3 in the order of points SP1, SP2, SP3, and SP4.

[0041] B-2. Other Embodiments 2: In the above embodiment, the second movement path data RD2, generated using the second detection data from the inertial measurement device 52, was corrected based on the first movement path data RD1, generated using the first detection data from the camera 20. However, the embodiment is not limited to this. For example, the first movement path data RD1 may be corrected based on the second movement path data RD2. Also, in the above embodiment, the multiple sensors 20 and 52 were the camera 20 and the inertial measurement device 52, but the embodiment is not limited to these and may include LiDAR or other sensors.

[0042] This disclosure is not limited to the embodiments described above, and can be implemented in various configurations without departing from its spirit. For example, the technical features of the embodiments corresponding to the technical features in each form described in the summary of the invention can be replaced or combined as appropriate in order to solve some or all of the above-described problems, or to achieve some or all of the above-described effects. Furthermore, if a technical feature is not described as essential in this specification, it can be deleted as appropriate. [Explanation of Symbols]

[0043] 10…Path estimation system, 20…Camera, 52…Inertial measurement device, 100…Path estimation device, 101…Memory, 102…Input / output interface, 103…Processor, 104…Internal bus, 105…Communication device, 110…Acquisition unit, 120…Path estimation unit, 125…Extraction unit, 130…Correction unit, 140…Output unit, 160…Program, 200…Input device, 300…Display device, CR…Specific path, FC…Factory, KP…Key point, MD…Skeleton detection model, RD1…First movement path data, RD2…Second movement path data, RS1…First specific data, RS2…Second specific data, SP1…Movement start point, SP2…Intersection point, SP3…Movement end point, WK…Worker

Claims

1. A path estimation system, An acquisition unit that acquires multiple detection data related to the movement path of a subject from each of multiple sensors, A path estimation unit that generates a plurality of movement path data representing the movement path of the subject according to each of the plurality of detection data, wherein each of the plurality of movement path data is defined by a coordinate axis for representing the position of each point in the movement path and a time axis for representing the time at each point, An extraction unit extracts from each of the aforementioned multiple travel path data sets a plurality of specific data indicating that the subject traveled along a predetermined specific route, A path estimation system comprising: a correction unit that compares the aforementioned plurality of specific data and performs a correction process to reduce the discrepancies between coordinate positions on the coordinate axes and between time points on the time axis in the plurality of travel path data.

2. A path estimation system according to claim 1, A path estimation system in which the correction unit corrects other movement path data in the correction process such that the discrepancy between the coordinate position on the coordinate axis and the time on the time axis in other movement path data and the specific movement path data generated using the detection data of a specific sensor among the multiple sensors is reduced.

3. A path estimation system according to claim 1, The aforementioned specific route is a route estimation system comprising a first straight route and a second straight route that includes an intersection point with the first straight route as its starting point and bends relative to the first straight route.

4. A path estimation system according to claim 3, A path estimation system in which the correction unit corrects the time on at least one of the time axes of the plurality of travel path data in the correction process such that the difference between the times indicated by at least one time data of the plurality of specific data is minimized, which is the start time data at the starting point of the movement of the specific path, the intersection time data at the intersection point, and the end time data at the end point of the movement of the specific path.

5. A path estimation system according to claim 4, A path estimation system in which the correction unit compares the plurality of specific data in the correction process and corrects the coordinate position of at least one of the plurality of coordinate axes in the plurality of travel path data so that the difference between the coordinate positions indicating each point in the plurality of specific data is minimized.

Images