JP6410670B2 - Traffic difficulty estimation device, method and program - Google Patents

Description

The present invention, traffic difficulty estimation TeiSo location to estimate the difficulty at the time of passing the outdoor and indoor with mobile equipment such as a person is walking or wheelchair, a method and a program.

  2. Description of the Related Art In recent years, many navigation services have been provided with the advance of GPS (Global Positioning System) technology and the spread of mobile terminals such as smartphones. One of them provides information on route ease. For example, the barrier-free route search of the Ministry of Land, Infrastructure, Transport and Tourism provides information such as the presence or absence of steps in the route, slopes of slopes, and the like. Information such as the presence or absence of steps and slopes of slopes is often collected by service providers. However, since collecting information by visiting the site is a heavy burden on the business operator, information collection is insufficient in many areas.

  In view of this situation, recently, attempts have been made to automatically estimate a road surface condition and a movement route when passing a wheelchair using sensor data obtained by an acceleration sensor or the like (for example, Non-Patent Document 1 or Non-Patent Document 2). reference).

  In addition, attempts have been made to acquire and analyze data measured by an inertial sensor such as an acceleration sensor when a wheelchair or a pedestrian moves, and map information such as steps, inclinations, movement trajectories, etc. on a map ( For example, see Non-Patent Document 3 or Non-Patent Document 4.)

  Each of these attempts collects data measured by an inertial sensor when a wheelchair user or a user who uses a stroller or suitcase moves indoors or outdoors, and uses SVM (support vector machine) or regression By analyzing the measurement data using an estimator realized by the above, road surface information such as the presence or absence of a step or slope of a slope is collected. Therefore, by using these technologies, the business operator can grasp the situation of each place on the traffic route at a low cost.

Yusuke Iwasawa, Atsuko Yairi, “Road surface state estimation system based on spatio-temporal analysis of wheelchair travel life log”, 2013 Annual Conference of Japanese Society for Artificial Intelligence, JSAI2013, 1D3-5 (2013) Murao Makoto, Nishino Takanori, Naruse Osamu, “A Wheelchair Moving Path Estimation Method Using 3-Axis Accelerometer”, IEICE Technical Report, WIT2012-40, pp.41-46 (2012) “Promoting R & D for Realizing Diversity Navigation”, NTT Holding Company News Release, January 15, 2015, Internet <URL: http://www.ntt.co.jp/news2015/1501/ 150115a.html> “Navigation that evolves with social information”, NTT R & D Forum 2015 materials, February 18, 2015, Internet <URL: https://labevent.ecl.ntt.co.jp/forum2015/elements/pdf_jpn/01/C- 9_j.pdf>

  However, since the techniques described in Non-Patent Documents 1-4 use an estimator realized by SVM (support vector machine) or regression, a large amount of sensor data with correct labels must be prepared in advance. For example, when constructing an estimator that estimates the presence or absence of steps or slopes using SVM, a large amount of machine learning measurement data with correct labels such as `` steps '' and `` slopes '', for example, at least a few It is necessary to prepare more than 100 samples.

  However, since it is extremely difficult to acquire a large amount of measurement data for machine learning in advance for obstacles such as “steps” and “tilts”, the accuracy of estimation of unknown phenomena tends to be low. There is. For example, even if one “step” is taken, there is a large single step and a small continuous step. For this reason, it is not realistic to collect machine data and perform machine learning when passing through all kinds of steps.

The present invention has been made paying attention to the above circumstances, and the object of the present invention is to estimate the degree of difficulty of travel, which can estimate the degree of difficulty of travel with high accuracy without collecting a large amount of machine learning data in advance. and to provide a TeiSo location, method and program.

The first aspect of the present invention in order to achieve the above object, we provided corresponding to each of the plurality of users is, during the movement of the mobile device to which the user or the user uses, the user Or the degree of difficulty of traffic that enables communication with multiple data measuring devices that measure the movement and position of mobile devices and send measurement data that includes 3-axis acceleration data, 3-axis angular velocity data, and position information In the estimation apparatus, means for receiving measurement data transmitted from the plurality of data measurement devices during the movement of the plurality of users, and the received plurality of measurement data are included in the measurement data means for aggregating at predetermined geographical areas on the basis of the information representing the position, for each measurement data the aggregated, the three-axis acceleration data and the three-axis angular velocity data which are included in those of the regimen measurement data It means for calculating a variation of the axes of the record, to normalize the variation of each axis which is calculated above, and means for determining the variation score for each axis, the passage difficulty in that geographical area from the variation score for each axis Means for calculating the information to be expressed is provided.

A second aspect of the invention, the measurement data includes information representing a moving direction, means for the aggregate, the multiple measurement data aggregated for each movement direction for each of the geographical area, the The traffic difficulty level is calculated for each moving direction in the geographical area .

  According to the first aspect of the present invention, the following operational effects can be obtained. That is, in general, it is more likely that a difference in the manner of traffic of each user occurs in places where it is difficult to travel. For example, if it is a flat road, many passers-by can pass smoothly without any difficulty, so that variations in movement are unlikely to occur. On the other hand, when there are steps or slopes, the user's movements vary widely. For example, a pedestrian may travel across a level difference or may catch a foot. In addition, the electric wheelchair user may move over the step and the manual wheelchair user may move while stepping left and right.

  Focusing on the above-mentioned tendency, in the first aspect of the present invention, for each predetermined geographical area, the variation is based on information representing the movements of a plurality of users or their mobile devices passing through the area. A score representing the degree of the above is calculated, and the calculated score is output as information representing the degree of difficulty in passing the geographical area. That is, the degree of difficulty of passing for each position is estimated based on the degree of variation in the manner of passage of a plurality of passers-by. For this reason, it is possible to accurately estimate the degree of difficulty in passing for each geographical area with a small amount of measurement data without acquiring a large amount of measurement data for machine learning in advance.

According to a first aspect of the present invention, when calculating a score representing the degree of variations in traffic, 3 axis included in each meter measurement data for a plurality of measurement data aggregated for each geographic area calculated variation of each of the respective acceleration data and 3-axis angular velocity data axis, variation score of each axis is calculated by normalizing the variation in each axis the calculated. For this reason, it is possible to accurately calculate the degree of variation in the movement of the user using the triaxial acceleration data and the triaxial angular velocity data.

According to the second aspect of the present invention, the calculation data includes information indicating the moving direction, and when calculating a score indicating the degree of variation during travel , for each geographical region and within the geographical region. The variation of the user's movement is calculated for each moving direction. For this reason, it is possible to further estimate not only the traffic difficulty level for each geographical area but also the traffic difficulty level for each movement direction of the user in the geographical area.

That is, according to the present invention, traffic difficulty estimation TeiSo location you can estimate the traffic difficulty with high precision without pre collect a large amount of machine learning data, it is possible to provide a method and a program.

The block diagram which shows the function structure of the traffic difficulty estimation system which concerns on 1st Embodiment of this invention. The flowchart which shows the process sequence and processing content of the server apparatus provided as a traffic difficulty estimation apparatus in the system shown in FIG. The figure which shows an example of the sensor data collected by the sensor data collection process part shown in FIG. The figure which shows an example of the sensor data group collected from the some user by the sensor data collection process part shown in FIG. The figure which shows an example of an area mesh. The figure which shows an example of the user's movement path | route on an area mesh. The figure which shows the outline | summary of the sensor data measured with each area mesh. The figure which shows an example of the sensor data divided | segmented for every area mesh. The figure which shows an example of the acceleration and angular velocity data aggregated for every area mesh. The figure which shows an example of the dispersion value (variation data) of the acceleration and angular velocity data calculated for every area mesh by the dispersion | distribution calculation part shown in FIG. The figure which shows an example of the score data of the dispersion value of the acceleration and angular velocity data calculated for every area mesh by the normalization processing part shown in FIG. The figure which shows an example of the traffic difficulty calculated for every area mesh by the traffic difficulty calculation part shown in FIG. The figure which shows the example of a display of a traffic difficulty level. The flowchart which shows the process sequence and process content of the server apparatus provided as a traffic difficulty estimation apparatus in 2nd Embodiment of this invention. The figure which shows an example of the sensor data with a moving direction collected by the sensor data collection process shown in FIG. The figure which shows an example of the sensor data group with a moving direction collected from the some user by the sensor data collection process shown in FIG. The figure which shows an example of the sensor data with a moving direction divided | segmented for every area mesh. The figure which shows an example of the moving direction, acceleration, and angular velocity data aggregated for every area mesh. The figure which shows an example of the moving direction, acceleration, and angular velocity data which were collected according to the moving direction for every area mesh. The figure which shows an example of the dispersion value (variation data) of the acceleration and angular velocity data calculated for every moving direction for every area mesh by the dispersion | distribution calculation process shown in FIG. The figure which shows an example of the score data of the acceleration and angular velocity data calculated for every moving direction for every area mesh by the normalization process shown in FIG. The figure which shows an example of the traffic difficulty according to the moving direction calculated for every area mesh by the traffic difficulty calculation process shown in FIG. The figure which shows the example of a display of a traffic difficulty level.

Embodiments according to the present invention will be described below with reference to the drawings.
[First Embodiment]
(Constitution)
FIG. 1 is a block diagram showing a functional configuration of a traffic difficulty level estimation system according to the first embodiment of the present invention.
The traffic difficulty level estimation system includes, for example, a server device SV that a local government or a service provider operates as a traffic difficulty level estimation device, a wheelchair MD that a user uses as a mobile device, and a data measurement device that is fixed to the wheelchair MD. Smartphone TM and a map system MP. The wheelchair MD and the smartphone TM are actually plural because a plurality of users individually use them, but the illustration is omitted here for the sake of simplicity.

  The smartphone TM can communicate with the server device SV via the communication network NW. As the communication network NW, for example, a mobile communication system such as 3G or 4G, a wireless LAN (Local Area Network), or a wireless communication network adopting a low-power wireless data communication standard such as Bluetooth (registered trademark) is used.

  The wheelchair MD is a commercially available wheelchair and may be a manual type or an electric type. The mobile device is not limited to the wheelchair MD, and may be a baby stroller, a robot suit or cane that assists the movement of the user, a shopping carrier case with wheels, a travel suitcase, or the like.

  The smartphone TM incorporates an inertial sensor including an acceleration sensor and an angular velocity sensor and a GPS (Global Positioning System) sensor as a position sensor. And the sensor data containing the measured value obtained by these sensors are transmitted to the said server apparatus SV via the communication network NW. In addition, although this embodiment demonstrates taking the case where the smart phone TM is fixed to the wheelchair MD as an example, the smart phone MD may be hold | gripped by a user or may be accommodated in a pocket, a bag, etc.

  Further, it is assumed that the axial directions of the acceleration sensor and the angular velocity sensor of the smartphone TM are known, and here, for simplicity, the front direction of the wheelchair MD is the positive direction of the x axis, and the left direction of the wheelchair MD is the positive direction of the y axis. The upward direction of the wheelchair MD is defined as the positive direction of the z axis. Note that the axial directions of the acceleration sensor and the angular velocity sensor may be unknown, and in this case, the axial direction is estimated from the state of changes in sensor data. For example, since most of the time during which the wheelchair MD is moving should move in the forward direction, it can be estimated that the axis where the acceleration increases or decreases in many time zones is the forward axis. In addition, a general acceleration sensor detects about 9.8m / s ^ 2 (gravity acceleration) upwards in a stationary state, so an acceleration of about 9.8m / s ^ 2 is generated upwards for most of the time. It can be estimated that the existing axis is the upward axis. If the forward and upward axes are obtained, the left direction is uniquely determined.

  The map system MP is composed of a general-purpose map display system that stores map data, and is capable of communicating with the smartphone TM and the terminal of the local government or business operator via the server device SV. Then, when the display instruction information of the traffic difficulty is received from the server device SV, display data is generated by superimposing a mark representing a position with a high traffic difficulty on the map data according to the instruction information, and the generated display data Is transmitted to the smart phone TM, the local government, or the terminal of the operator via the server device SV. The general-purpose map display system is described in detail in, for example, wheelmap.org (URL: http://wheelmap.org).

  The server device SV includes a control unit 1, a storage unit 2, and a communication interface unit 3. The communication interface unit 3 performs communication with the smartphone TM for collecting sensor data via the communication network NW.

  The storage unit 2 uses a nonvolatile memory that can be written and read as needed, such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive) as a storage medium, and is a storage necessary for implementing this embodiment. The area includes a collected data storage unit 21, a divided data storage unit 22, an aggregated data storage unit 23, a variation data storage unit 24, a score data storage unit 25, and a traffic difficulty data storage unit 26.

  The collected data storage unit 21 is used to store sensor data collected from the plurality of smartphones TM. The divided data storage unit 22 is used to store the collected sensor data in association with any of a plurality of geographical areas (area meshes) divided into meshes.

  The variation data storage unit 23 is used to store data representing the degree of variation of the sensor data calculated for each area mesh. The score data storage unit 24 is used to store variation score data calculated based on data representing the degree of variation. The traffic difficulty level data storage unit 25 is used to store the traffic difficulty level data calculated based on the variation score data.

  The control unit 1 includes a CPU (Central Processing Unit). As processing functions necessary for carrying out the present embodiment, a sensor data collection processing unit 11, a sensor data division processing unit 12, a distributed calculation unit 13, A normalization processing unit 14, a traffic difficulty level calculation unit 15, and a traffic difficulty level data output unit 16 are provided. These processing functions are realized by causing the CPU to execute a program stored in the program memory.

  The sensor data collection processing unit 11 receives the sensor data transmitted from the smartphone TM according to the movements of a plurality of users, and performs processing for storing the received sensor data in the collected data storage unit 21.

  The sensor data division processing unit 12 defines a plurality of area meshes by dividing the service area into a predetermined size in a mesh shape. Then, each sensor data stored in the collected data storage unit 21 is associated with any one of the plurality of area meshes based on the position data included in the sensor data, and the sensor data of the same user is 1 A process of collecting the data as one unit and storing it in the divided data storage unit 22 is performed.

  The variance calculation unit 13 reads out sensor data of the same user for each area mesh from the divided data storage unit 22, merges acceleration data and angular velocity data included in the sensor data for each axis, and then calculates the variance value. Calculate Then, the dispersion data storage unit 23 stores the calculated dispersion values for each axis of the acceleration data and the angular velocity data as data representing the degree of dispersion.

  The normalization processing unit 14 normalizes the variance values for each axis of acceleration data and angular velocity data stored in the variation data storage unit 23. As a normalization method, for example, a method is used in which normalization is performed so that the variance value of each axis of acceleration data and angular velocity data is 0 when the value is minimum and 1 when the value is maximum. And the process which memorize | stores the dispersion value after the said normalization in the score data memory | storage part 24 as dispersion | variation score data is performed.

  The traffic difficulty level calculation unit 15 calculates an average score value for each area mesh based on the variation score data stored in the score data storage unit 24, and calculates the calculated average score value for the area mesh. A process of storing in the traffic difficulty level data storage unit 25 as data representing the level of difficulty of travel is performed.

  The traffic difficulty level data output unit 16 determines whether any of the traffic difficulty level data for each area mesh stored in the traffic difficulty level data storage unit 25 exceeds a reference value. And when the traffic difficulty level data exceeding a reference value is found, the display instruction information including the position of the area mesh corresponding to the traffic difficulty level data is transmitted to the map system MP.

(Operation)
Next, the operation of the traffic difficulty level estimation system configured as described above will be described. FIG. 2 is a flowchart showing a processing procedure and processing contents of the server apparatus SV which is a main part of the system.

(1) Sensor data collection processing It is assumed that the user has moved on the wheelchair MD, for example, in the city. During this movement, the smartphone TM fixed to the wheelchair MD measures acceleration and angular velocity at regular intervals (eg, every 20 milliseconds) and accumulates sensor data in an internal storage area. At the same time, the smartphone TM measures and accumulates latitude / longitude data representing the current position by the GPS sensor at the same timing as the measurement timing of the acceleration and angular velocity. Then, the measured latitude / longitude data is wirelessly transmitted to the server device SV as sensor data at regular time intervals (for example, every second) together with the acceleration and angular velocity measurement data.

  On the other hand, in the standby state, the server device SV monitors the arrival of sensor data from the smartphone TM in step S11 under the control of the sensor data collection processing unit 11. Each time sensor data arrives, the sensor data is received in step S12 and stored in the collected data storage unit 21. FIG. 3 shows an example of sensor data collected from one user. Each sensor data includes measurement date and time, latitude / longitude data indicating a position, and acceleration x-axis, y-axis, and z-axis measurements. And the measured values of the angular velocity in the x-axis, y-axis, and z-axis.

  The sensor data collection processing unit 11 repeatedly executes the sensor data collection processing in steps S11 to S13 until it is confirmed in step S13 that a certain amount of sensor data has been collected. FIG. 4 shows an example of a storage result of sensor data collected from a plurality of users by this iterative process.

(2) Sensor data division processing When it is confirmed in step S13 that a certain amount of sensor data has been collected, the server device SV controls the sensor data for each measurement position under the control of the sensor data division processing unit 12. The dividing process is executed as follows.

  That is, first, in step S14, each sensor data stored in the collected data storage unit 21 is associated with one of the predefined area meshes. In other words, sensor data measured at the same position is collected. Specifically, as shown in FIG. 5, an area mesh having one side of latitude 0.0001 and longitude 0.0001 is defined. In step S15, the sensor data continuously measured by the movement of the same user in the same mesh are collected as one unit based on the position data included in the sensor data.

  As a result, for example, if the users A, B, and C move as shown in FIG. 6, the division result of the sensor data for each area mesh is as shown in FIG. 7, and the actual divided sensor in this case The data is as shown in FIG. The divided sensor data is stored in the divided data storage unit 22.

(3) Calculation of variance of sensor data Next, the server device SV uses each sensor data stored in the divided data storage unit 22 for each area mesh in step S16 under the control of the variance calculation unit 13. Sensor data generated by a person's traffic is merged, and the variation of the sensor data is calculated for each area mesh.

  For example, with respect to the divided sensor data shown in FIG. 8, the measured values of the respective axes of acceleration and angular velocity obtained by the passage of each user are merged for each area mesh. FIG. 9 shows the values of the acceleration and angular velocity axes after merging. Then, a variance value of the values of the acceleration and angular velocity axes after merging is calculated for each area mesh, and the calculation result is stored in the variation data storage unit 23 as data indicating variations in the movements of a plurality of users. FIG. 10 shows variation data calculated from the values of the respective axes of acceleration and angular velocity after merging shown in FIG.

  When calculating variations in sensor data, as described above, the measurement values of the acceleration and angular velocity axes of the sensor data may be calculated as targets, or the acceleration and angular velocity of the sensor data may be calculated. The measurement value may be calculated by subjecting the measurement value to smoothing using a low-pass filter or the like. Further, for example, a feature amount of sensor data may be extracted by obtaining a correlation value between acceleration x-axis and y-axis, and the extracted sensor data feature amount may be calculated as a target. All the differences between the measured value and the i + 1th measured value may be obtained, and the difference group may be calculated. In addition, as a variation index, a variance value may be used as in the present embodiment, or another index that can express variations such as a standard deviation value may be used.

(4) Calculation of Variation Score Data The server device SV continues to control the variation based on the variation sensor data stored in the variation data storage unit 23 in step S17 under the control of the normalization processing unit 14. A process for calculating the score is executed.

For example, normalization is used as a method for calculating the variation score. That is, for each variation sensor data stored in the variation data storage unit 23, the variance value for each axis of acceleration and angular velocity is “0” for the minimum value and “1” for the maximum value. Normalize as follows. Specifically, when v _i is a value before normalization, v _i ′ is a value after normalization, v _max is a maximum value of each axis, and vmin is a minimum value of each axis,
_{v i '= (v i -vmin} ) / (vmax -vmin)
Calculate according to Then, the value after the normalization process is stored in the score data storage unit 24 as variation score data. FIG. 11 shows an example of variation score data stored in the score data storage unit 24.

(5) Calculation of traffic difficulty The server device SV subsequently controls the traffic difficulty based on the variation score data stored in the score data storage unit 24 in step S18 under the control of the traffic difficulty calculation unit 15. The degree is calculated as follows:

For example, an average value calculation method is used as the calculation method of the traffic difficulty level. Specifically, when vax, vay, and vaz are scores of acceleration x-axis, y-axis, and z-axis, respectively, and vgx, vgy, and vgz are scores of angular velocity x-axis, y-axis, and z-axis, respectively, the average value v _mean is obtained.
v _mean = (vax + vay + vaz + vgx + vgy + vgz) / 6
Calculate according to Then, the calculated traffic difficulty level data is stored in the traffic difficulty level data storage unit 25. FIG. 12 shows an example of the traffic difficulty level data stored in the traffic difficulty level data storage unit 25.

(6) Output of difficulty level of traffic The server device SV finally displays an area mesh having a high degree of difficulty of traffic based on the calculated difficulty level of traffic under the control of the difficulty level data output unit 16. The process of generating and outputting the display instruction information is executed as follows.

  That is, first, in step S19, the traffic difficulty level data stored in the traffic difficulty level data storage unit 25 is compared with a threshold value as a reference value, and traffic difficulty level data equal to or higher than the threshold value is extracted. In step S20, the latitude / longitude data of the area mesh from which the extracted traffic difficulty level is obtained is read from the traffic difficulty data storage unit 25, and display instruction information including the latitude / longitude data of the area mesh is generated. And output to the map system MP.

  For example, if the reference value is now “0.5”, “0.876” in the traffic difficulty list stored in FIG. 12 exceeds this, so the latitude / longitude data (35.2254) of the corresponding area mesh , 139.6637) is generated and output to the map system MP.

  As a result, in the map system MP, in accordance with the display instruction information output from the server device SV, a position corresponding to the latitude / longitude data included in the display instruction information on the map data, for example, as shown in FIG. A black point representing and character data “notice to pass” are superimposed and displayed.

  Thus, for example, a user who is about to move can browse the map data with his / her smartphone to confirm in advance where to pay attention to traffic.

(effect)
As described above in detail, in the first embodiment, sensor data including acceleration and angular velocity measurement data and latitude / longitude data using GPS is collected from a plurality of users' smartphones TM, and the collected plurality of sensors. Data is aggregated in association with any of a plurality of area meshes. Then, for each area mesh, after calculating the variance of measurement values for each axis of acceleration and angular velocity included in the aggregated sensor data, the variance value is normalized to calculate score data representing the degree of variation. The average value of the score data is defined as the traffic difficulty level in the area mesh. Then, the traffic difficulty level higher than the threshold is extracted from the traffic difficulty levels of each area mesh, and the display instruction information including the latitude / longitude data of the area mesh corresponding to the extracted traffic difficulty level is displayed on the map system MP. Is output to the position corresponding to the area mesh on the map data.

  Therefore, the degree of difficulty in passing for each area mesh is estimated based on the degree of variation in movement of a plurality of users during passage. For this reason, it becomes possible to accurately estimate the degree of difficulty in passing for each area mesh with a small amount of sensor data without acquiring a large amount of machine learning measurement data in advance. Further, since the three-axis acceleration data and the three-axis angular velocity data included in the sensor data are used when calculating the degree of variation in traffic, the degree of variation in user movement can be accurately calculated. .

[Second Embodiment]
In the second embodiment of the present invention, the sensor data transmitted from the smartphone TM includes data representing the moving direction of the user, and the server device SV has a variation score of the movement of the user for each moving direction for each area mesh. And the degree of difficulty in passing for each moving direction is obtained based on the movement variation score for each moving direction.

  FIG. 14 is a flowchart showing a processing procedure and processing contents by the server apparatus SV according to the second embodiment of the present invention. The configuration of the traffic difficulty level estimation system is basically the same as that of the first embodiment described above, and will be described with reference to FIG.

  The smartphone TM includes an orientation sensor in addition to an acceleration sensor, an angular velocity sensor, and a GPS sensor. Azimuth sensors work in conjunction with GPS sensors to provide north (N), northeast (NE), east (E), southeast (SE), south (S), southwest (SW), west (W), and northwest (NW ) The direction of movement is sensed with a resolution of 8 directions. Note that the resolution is not limited to eight directions, but may be four directions or more than eight directions.

(1) Sensor data collection process with moving direction It is assumed that a user has moved in a wheelchair MD, for example, in a city. During this movement, the smartphone TM fixed to the wheelchair MD measures acceleration and angular velocity at regular intervals (eg, every 20 milliseconds) and accumulates sensor data in an internal storage area. At the same time, the smartphone TM measures and accumulates the latitude / longitude data representing the current position by the GPS sensor at the same timing as the acceleration and angular velocity measurement timing, and detects the moving direction by the GPS sensor and the direction sensor. accumulate. Then, the measured latitude / longitude data and azimuth data are wirelessly transmitted as sensor data to the server device SV at regular time intervals (eg, every second) together with the acceleration and angular velocity measurement data.

  On the other hand, in the standby state, the server device SV monitors the arrival of sensor data from the smartphone TM in step S21 under the control of the sensor data collection processing unit 11. Each time sensor data arrives, the sensor data is received in step S22 and stored in the collected data storage unit 21. FIG. 15 shows an example of sensor data with a moving direction collected from a single user. Each sensor data includes measurement date and time, data indicating a moving direction, latitude / longitude data indicating a position, and x of acceleration. Each measurement value of the axis, y-axis, and z-axis and each measurement value of the angular velocity, x-axis, y-axis, and z-axis.

  The sensor data collection processing unit 11 repeatedly executes the sensor data collection processing in steps S21 to S23 until it is confirmed in step S23 that a certain amount of sensor data has been collected. FIG. 16 shows an example of a storage result of sensor data with a moving direction collected from a plurality of users by this iterative process.

(2) Dividing process of sensor data with moving direction When it is confirmed that a certain amount of sensor data with moving direction has been collected in step S13, the server device SV moves under the control of the sensor data dividing processing unit 12 A process of dividing the sensor data with direction for each measurement position is executed as follows.

  That is, first, in step S24, the sensor data with the moving direction stored in the collected data storage unit 21 is collected for each area mesh shown in FIG. FIG. 17 shows an example of sensor data with a moving direction aggregated for each area mesh.

  Next, in step S25, with respect to the sensor data with a moving direction aggregated for each area mesh, the moving direction is aggregated for each sensor data continuously measured by the movement of the same user within the same mesh. Specifically, majority processing is performed on a group of movement directions continuously measured by the movement of the same user within the same area mesh. In step S26, the moving direction with the largest number of the majority processing results is set as the aggregation result.

  For example, in FIG. 16, within the area mesh with “35.2252, 139.6637” at the northwestern end, the same user's during the period “2015-01-15_11-01-01.122 to 2015-01-15_11-01-12.242” If the number of movement directions measured by movement is 500 for N, 45 for NE, and 11 for E, the aggregation result is N.

  As described above, a set of sensor data aggregated by movement direction for each area mesh is stored in the divided data storage unit 22. FIG. 17 shows an example of the storage result.

(3) Calculation of variance of sensor data with moving direction The server SV next reads the sensor data aggregated by the moving direction from the divided data storage unit 22 in step S27 under the control of the variance calculating unit 13, The sensor data passing in the same movement direction for each area mesh is merged, and the variation of the sensor data for each movement direction is calculated for each area mesh.

  For example, with respect to the sensor data aggregated for each moving direction shown in FIG. 17, the measured values of the axes of acceleration and angular velocity obtained in the process in which each user passes in the same direction are merged for each area mesh. FIG. 19 shows the values of the acceleration and angular velocity axes after merging. Then, for each area mesh, the variance value of the value of each axis of the acceleration and the angular velocity after the merging is calculated for each moving direction, and the calculation result is stored in the variation data storage unit 23 as data indicating the variation of the movements of a plurality of users. Remember. FIG. 20 shows variation data calculated from the values of the acceleration and angular velocity axes after merging shown in FIG.

  In this case as well, when calculating the variation, as described above, the measurement values of the acceleration and angular velocity axes of the sensor data may be calculated as targets, or the acceleration and angular velocity of the sensor data may be calculated. The measurement value may be calculated by subjecting the measurement value to smoothing using a low-pass filter or the like. Further, for example, a feature amount of sensor data may be extracted by obtaining a correlation value between acceleration x-axis and y-axis, and the extracted sensor data feature amount may be calculated as a target. All the differences between the measured value and the i + 1th measured value may be obtained, and the difference group may be calculated. In addition, as a variation index, a variance value may be used as in the present embodiment, or another index that can express variations such as a standard deviation value may be used.

(4) Calculation of Variation Score Data The server device SV subsequently obtains sensor data representing variation in each moving direction stored in the variation data storage unit 23 in step S28 under the control of the normalization processing unit 14. Based on this, a process for calculating a variation score for each moving direction is executed.

  As a method for calculating the variation score, normalization is used as in the first embodiment. That is, with respect to the variation sensor data for each moving direction stored in the variation data storage unit 23, the variance value for each axis of acceleration and angular velocity is “0” for the minimum value and “1” for the maximum value. ”To normalize. And the value after the said normalization process is memorize | stored in the score data memory | storage part 24 as dispersion | variation score data according to a moving direction. FIG. 21 shows an example of variation score data for each moving direction stored in the score data storage unit 24.

(5) Calculation of traffic difficulty level The server device SV is subsequently based on the variation score data for each moving direction stored in the score data storage unit 24 in step S29 under the control of the traffic difficulty level calculation unit 15. In addition, the degree of travel difficulty for each moving direction is calculated for each area mesh.

  As a calculation method of the traffic difficulty level, an average value calculation method is used as in the first embodiment. Then, the calculated traffic difficulty level data is stored in the traffic difficulty level data storage unit 25. FIG. 22 shows an example of the travel difficulty level data for each area mesh stored in the travel difficulty level data storage unit 25.

(6) Output of difficulty level of traffic The server device SV finally displays an area mesh having a high degree of difficulty of traffic based on the calculated difficulty level of traffic under the control of the difficulty level data output unit 16. The process of generating and outputting the display instruction information is executed as follows.

  That is, first, in step S30, the travel difficulty data for each moving direction for each area mesh stored in the travel difficulty data storage unit 25 is compared with a threshold value as a reference value, and the traffic difficulty exceeding the threshold value. Extract degree data. In step S31, the latitude / longitude data of the area mesh from which the extracted traffic difficulty level is obtained is read from the traffic difficulty data storage unit 25, and display instruction information including the latitude / longitude data of the area mesh is generated. And output to the map system MP.

  For example, when the reference value is “0.5” as in the first embodiment, “0.876” exceeds the traffic difficulty level list stored in FIG. Display instruction information including the latitude / longitude data (35.2254, 139.6637) and the moving direction data is generated and output to the map system MP.

  As a result, according to the display instruction information output from the server device SV, the map system MP moves to a position corresponding to the latitude / longitude data included in the display instruction information on the map data as shown in FIG. The arrow pattern indicating the direction and the character data “passing attention” are superimposed and displayed.

  Thus, for example, a user who is about to move can view the map data with his / her smartphone and confirm in advance where to pay attention to the passage and its moving direction.

(effect)
As described above in detail, in the second embodiment, sensor data including the moving direction is transmitted from the smartphone TM, the server device SV collects the sensor data, and moves the collected sensor data for each area mesh. Aggregate by direction. Then, after calculating the variance of measured values for each axis of acceleration and angular velocity included in the sensor data aggregated for each moving direction for each area mesh, the variance value is normalized to express the degree of variation for each moving direction. The score data is calculated, and the average value of the score data is set as the degree of difficulty of travel for each moving direction in the area mesh. Then, for each area mesh, a traffic difficulty level higher than the threshold is extracted from the traffic difficulty levels according to the movement direction, and the latitude / longitude data and the movement direction of the area mesh corresponding to the extracted traffic difficulty level are included. By outputting the display instruction information to the map system MP, a → mark indicating the moving direction and a text message “passing attention” are displayed at a position corresponding to the area mesh on the map data.

  Therefore, the degree of difficulty in passing for each moving direction for each area mesh is estimated based on the degree of variation in movement when a plurality of users pass. For this reason, it becomes possible to estimate the degree of difficulty in passing for each area mesh and for each moving direction with a small amount of sensor data without acquiring a large amount of measurement data for machine learning in advance. In addition, since the 3-axis acceleration data and 3-axis angular velocity data included in the sensor data are used when calculating the degree of variation in traffic, the degree of variation in user movement according to movement direction is accurately calculated. can do.

[Other Embodiments]
In the second embodiment, the movement direction is determined in the smartphone TM, and the data is included in the sensor data and transmitted. However, in the server device SV, the moving direction may be determined from the transition of position data included in a plurality of sensor data received in time series from the same smartphone.

  In addition, the type of the data measuring device, the type and configuration of the traffic difficulty estimating device, the processing procedure, the processing content, and the like can be variously modified and implemented without departing from the gist of the present invention.

  In short, the present invention is not limited to the above-described embodiments as they are, and can be embodied by modifying the components without departing from the scope of the invention in the implementation stage. Moreover, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the above embodiments. For example, some components may be deleted from all the components shown in each embodiment. Furthermore, you may combine suitably the component covering different embodiment.

  SV ... server device as traffic difficulty estimation device, TM ... smart phone, MD ... wheelchair, MP ... map system, 1 ... control unit, 2 ... storage unit, 3 ... communication interface unit, 11 ... sensor data collection processing unit, 12 ... sensor data division processing unit, 13 ... dispersion calculation unit, 14 ... normalization processing unit, 15 ... traffic difficulty level calculation unit, 16 ... traffic difficulty level data output unit, 21 ... collected data storage unit, 22 ... divided data storage unit , 23... Variation data storage unit, 24... Score data storage unit, 25.

Claims (4)

3-axis acceleration data provided corresponding to each of a plurality of users, measuring the movement and position of the user or the mobile device while the user or the mobile device used by the user is moving , A traffic difficulty estimation device capable of communicating with a plurality of data measuring devices that transmit measurement data including triaxial angular velocity data and information representing position,
Means for receiving measurement data transmitted from the plurality of data measurement devices, respectively, during movement of the plurality of users;
Means for aggregating the plurality of received measurement data for each predetermined geographical area based on information representing a position included in the measurement data;
Wherein each aggregated measurement data, means for calculating a variation of each axis of the three respective axis acceleration data and the three-axis angular velocity data included in those the regimen measurement data,
The variation of each axis which is the calculated normalized means for determining the variation score for each axis,
Means for calculating information representing the degree of difficulty in passing in the geographical area from the variation score of each axis . The measurement data includes information indicating a moving direction,
The means for aggregating a plurality of measurement data for each geographical area for each movement direction,
The traffic difficulty is calculated for each moving direction in the geographical area.
The traffic difficulty estimating device according to claim 1, wherein 3-axis acceleration data provided corresponding to each of a plurality of users, measuring the movement and position of the user or the mobile device while the user or the mobile device used by the user is moving , A traffic difficulty level estimation method executed by a traffic difficulty level estimation device capable of communication with a plurality of data measurement devices that transmit measurement data including triaxial angular velocity data and information representing position,
A process of receiving the measurement data transmitted from the plurality of data measurement devices during the movement of the plurality of users by the traffic difficulty level estimation device;
The process of estimating the degree of difficulty of passing the traffic, the process of aggregating the plurality of received measurement data for each predetermined geographical area based on information representing a position included in the measurement data;
And process the traffic difficulty estimating apparatus, the per aggregated measurement data, calculating a variance of each axis of the three respective axis acceleration data and the three-axis angular velocity data included in those the regimen measurement data,
The traffic difficulty estimating apparatus normalizes the variance of each axis is the calculated, a process of obtaining a variation score for each axis,
The traffic difficulty level estimation device comprises a step of calculating information indicating the traffic difficulty level in the geographical area from the variation score of each axis . The program which makes the processor with which the said traffic difficulty estimation apparatus performs the process by said each means with which the traffic difficulty estimation apparatus of Claim 1 or 2 is provided is performed.