JP6480319B2 - Road surface condition understanding method, apparatus, and program - Google Patents

Description

  The present invention relates to a road surface situation understanding method, apparatus, and program, and more particularly, to a road surface situation understanding method, apparatus, and program for understanding a road surface situation where a user walks.

  Since the spread of smartphones incorporating a large number of sensor devices, there have been several efforts to understand road conditions using portable sensor devices.

  One approach is to use a smartphone installed in an automobile to detect road steps, road conditions such as paved roads or gravel roads (Non-Patent Document 1, and Non-patent document 2). In these efforts, steps and road surface conditions are estimated based on information from the acceleration sensors of smartphones installed in automobiles, and correlated with positioning information measured at the same time to determine which location is what road surface condition. It is something to try to grasp.

  As another approach, there is an approach for trying to understand a person's behavior (standing, sitting, walking, etc.) based on information such as a smartphone held by the person (see Non-Patent Document 3). Although recognition of standing, sitting, and walking alone does not lead to grasping the road surface condition, it is possible to grasp that there is a stairs at that place by recognizing the action of going up the stairs or going down the stairs. When such behavior recognition is performed, there is a large individual difference in the data trend that can be acquired by the sensor, so recognition is often performed in a supervised learning framework only for personal data, but in order to improve recognition accuracy Similarity is assumed for the purpose of securing a sufficient amount of data.

  In addition, there is an approach to use the user's own data (see Non-Patent Document 4). By using such a framework, it is possible to secure learning data and to understand the actions of users who have not created teacher data.

Tomohiro Nomura et al., “Proposal of Road Surface Estimation Method Using Smartphone for Comfortable Driving Support”, IPSJ 5th National Convention, 5v-6, 2013. Koichi Yagi, "Surface Level Step Observation Method Using Smartphone and Survey Results of Road Level Step Distribution after the Tohoku Earthquake", 9th Japan Road Conference, 2011. Jennifer R. Kwapisz, Gary M. Weiss, Samuel A. Moore, \ Activity Recognition using Cell Phone Accelerometers ", in Proc. Of SensorKDD, 2010. NDLane, Y.Xu, H.Lu, S.Hu, T.Choudhury, ATCambell, and F.Zhao, "Enabling large-scale human activity inference on smartphones using community similarity networks", In Proc. Of UbiComp '11 , pp.355-364,2011.

  However, the above prior art has the following unsolved points.

  Obviously, the means using a smartphone installed in a car cannot collect information on places where the car cannot run, such as sidewalks and indoors.

  The problem of this place can be solved by considering an approach to understand the behavior of people with smartphones, but it seems that there are issues in utilizing the current behavior recognition efforts. One of the conventional behavior recognition methods is to extract sensor output data into a specific time window section as shown in Fig. 1 and grasp the human behavior that continued in time (for example, behavior for 10 seconds). If you try to understand the level difference and road surface condition pinpoint, it is considered that another process is required. In addition, it is thought that it is difficult in the first place to grasp an action that exceeds a step that occurs only at a slight timing by grasping an action that continues in time.

  Approaches to increase teacher data based on similarity between users are generally considered effective for improving recognition accuracy, but grouping based on user attributes and raw sensing data has been effective in previous efforts. It is used for tasks such as typical behavior pattern analysis, and it is considered difficult to function sufficiently with the existing grouping method in the task for determining the state of step by step which is the subject of this time.

  The present invention has been made to solve the above problems, and an object of the present invention is to provide a road surface state understanding method, apparatus, and program capable of accurately recognizing a road surface state for each walking step of a user.

  In order to achieve the above object, in the road surface state understanding method according to the first invention, the walking step recognition unit records the output value of the sensor provided in the portable device carried by the user in time series along with the time stamp. Based on the sensor output data, a step of recognizing a walking step that is a step of an individual foot when the user walks in the sensor output data, and a walking data segmentation unit are recognized by the walking step recognition unit. Segmenting the sensor output data into each of the walking data corresponding to the walking step based on the walking step, and the walking data segmented by the walking data segmentation unit by the data grouping unit A group group in which a plurality of users are grouped in advance, or A step of identifying a group to which the walking data belongs from a group of groups of the walking data of a plurality of users, and a step type recognition unit for each of the walking data recognized by the walking step recognition unit The step of recognizing the walking step state in the walking data based on the group of the walking data specified by the data grouping unit, and the position-specific road surface state understanding unit measured the position where the portable device exists. Based on the positioning data recorded with a time stamp, the positioning data for each of the walking data and the walking step state for each of the walking data recognized by the step type recognition unit, Recognizing the road surface condition of the position, And butterflies.

  Further, in the road surface state understanding method according to the first invention, each group of the group group includes a user attribute representing the attribute of the user belonging to the group or a user attribute of the user of the walking data belonging to the group. And the step of identifying the group by the data grouping unit is based on the user attribute of the user who is walking in advance and the user attribute that is assigned to each group of the group group. A group to which each piece of data belongs may be specified.

  In the road surface condition understanding method according to the first aspect of the present invention, the group group includes a group group in which the plurality of users are grouped by further considering the positioning data, or the plurality of groups further considering the positioning data. The step of specifying the group by the data grouping unit is a group group in which the walking data of the user is grouped, and for each of the walking data segmented by the walking data segmentation unit, A group to which the walking data belongs may be specified from the group group based on positioning data for the data.

  The road surface condition understanding device according to the second invention is based on sensor output data in which the output value of a sensor provided in a portable device carried by a user is recorded in time series together with a time stamp, in the sensor output data, Corresponding to the walking step, the sensor output data corresponds to the walking step recognition unit that recognizes the walking step that is the step of the individual foot when the user walks, and the walking step recognized by the walking step recognition unit. A group of groups of a plurality of users obtained in advance for each of the walking data segmented by the walking data segmentation unit and the walking data segmentation unit that segments into each of the walking data to be grouped, Alternatively, from the group group in which the walking data of the plurality of users are grouped, the walking For each of the walking data recognized by the walking step recognition unit and a data grouping unit that identifies a group to which the data belongs, the walking data is based on the group of the walking data specified by the data grouping unit. The step type recognition unit for recognizing the walking step state and the position-specific road surface state understanding unit measure the position where the portable device is present, based on the positioning data recorded together with the time stamp, for each of the walking data The positioning data and the step of recognizing the road surface condition at each position by aggregating the walking step state for each of the walking data recognized by the step type recognition unit.

  Further, in the road surface condition understanding device according to the second invention, each group of the group group has a user attribute representing the attribute of the user belonging to the group or a user attribute of the user of the walking data belonging to the group. Each of the walking data belongs to the data grouping unit based on a user attribute of the user who is walking and acquired in advance and a user attribute assigned to each group of the group group. A group may be specified.

  In the road surface condition understanding device according to the second aspect of the present invention, the group group includes a group group in which the plurality of users are grouped by further considering the positioning data, or the plurality of groups further considering the positioning data. A group of the user's walking data grouped, wherein the data grouping unit is based on positioning data for the walking data for each of the walking data segmented by the walking data segmentation unit. Then, the group to which the walking data belongs may be specified from the group group.

  A program according to a third invention is a program for causing a computer to execute each step of the road surface condition understanding method according to the first invention.

  According to the road surface state understanding method, apparatus, and program of the present invention, a walking step is recognized, segmented into each of the walking data corresponding to the walking step, and each of the segmented walking data is walked from the group group. By identifying the group to which the data belongs and recognizing any of the walking step states based on the group of walking data, the effect that the road surface condition for each walking step of the user can be accurately recognized can be obtained.

It is a figure which shows an example which cut out sensor output data with the specific time window. It is a block diagram which shows the structure of the road surface condition learning system which concerns on embodiment of this invention. It is a figure which shows an example of the sensor output data stored in a walk data storage part. It is a figure which shows an example of the walking step recognized from the sensor output data. It is a figure which shows an example of the positioning data stored in a positioning data storage part. It is a figure which shows an example of the group group stored in a grouping information storage part. It is a figure which shows an example of the classification data stored in the classification data storage part for learning. It is a flowchart which shows the road surface condition learning process routine in the road surface state learning apparatus of the road surface state learning system which concerns on embodiment of this invention. It is a block diagram which shows the structure of the road surface condition understanding system which concerns on embodiment of this invention. It is a figure which shows an example of the recognition result stored in a walk data analysis result storage part. It is a flowchart which shows the road surface state understanding process routine in the road surface state understanding apparatus of the road surface state understanding system which concerns on embodiment of this invention.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

<Configuration of road surface learning system according to an embodiment of the present invention>

  First, the configuration of the road surface state learning system according to the embodiment of the present invention will be described. As shown in FIG. 2, the road surface state learning system 10 according to the embodiment of the present invention includes a mobile device 100 and a road surface state learning device 300 each carried by a plurality of users.

  The portable device 100 includes a walking data collection unit 110 and a positioning unit 120, acquires an output value and position information of a sensor related to walking together with a time stamp, and stores the acquired value in a storage area 101 as necessary. In addition, the data collected by the walking data collection unit 110 and the positioning unit 120 and the identification information associated with the data collected by the walking data collection unit 110 and the positioning unit 120 on a regular basis or according to an operation from the user, respectively, The data is transmitted to the storage unit 130 and the positioning data storage unit 180 and stored.

  The walking data collection unit 110 is realized as hardware or software incorporated in the mobile device 100. Examples of sensors that realize this function are acceleration sensors, gyro sensors, walking detection sensors, etc., and output values of these sensors are obtained as sensor output data recorded in chronological order together with time stamps. After being stored in the storage area 101, it is transmitted to the walking data storage unit 130 by the function of the portable device 100. When transmitting, you may link | link with identification information, such as terminal ID, as needed.

  The positioning unit 120 is realized as hardware or software incorporated in the mobile device 100. This function is realized by positioning means using satellite radio waves such as GPS, positioning means using Wi-Fi (R) radio waves, and combinations thereof. The position information acquired here is acquired in association with the time stamp, stored in the storage area 101 as necessary, and transmitted to the positioning data storage unit 180 by the function of the portable device 100. When transmitted, it may be transmitted in association with identification information such as a terminal ID as necessary.

  The road surface state learning device 300 can be configured by a computer including a CPU, a RAM, and a ROM that stores a program for executing a road surface state learning process routine described later and various data.

  The road surface condition learning device 300 includes a walking data accumulation unit 130, a walking step recognition unit 140, a walking data segmentation unit 150, a step type learning unit 160, a learning classification data storage unit 163, and a learning model storage unit 165. A positioning data storage unit 180, a data grouping unit 200, and a grouping information storage unit 205.

  The walking data storage unit 130 stores the value of sensor output data received from the portable device 100. An example of sensor output data stored in the walking data storage unit 130 is information on terminal ID, time stamp, acceleration X, Y, and Z, and gyro X, Y, and Z, as shown in FIG.

  The walking step recognizing unit 140 recognizes a walking step, which is an individual foot movement when the user walks, in the sensor output data based on the sensor output data of each user stored in the walking data storage unit 130. Specifically, the sensor output data stored in the walking data storage unit 130 is analyzed, and the walking steps that are the steps of the individual feet when the user walks are recognized. That is, based on the input of the waveform data shown in FIG. 1, processing for recognizing the walking step is performed at time stamp intervals corresponding to the vertical bars shown in FIG. The walking step data to be output is information that is given in the form of a plurality of time stamps, and is obtained by dividing the sensor output data along the waveform of FIG. 1 based on the user's walking step. As an example of this recognition method, a method of specifying a peak of sensor output data, specifying a peak interval time, and dividing a waveform according to the time can be considered. Note that the walking step may be recognized based on information obtained from a step counting sensor provided in the portable device.

  The walking data segmentation unit 150 uses each sensor output data stored in the walking data storage unit 130 based on the walking step recognized by the walking step recognition unit 140 as the walking data corresponding to the walking step. Segment into A simple method is a method in which a section between two vertical bar intervals shown in FIG. 4 is regarded as one segment as step section data. In addition to the step interval data corresponding to the walking step, the step interval data corresponding to the preceding and following walking steps is combined with the idea that the characteristics of one walking step also appear in the preceding and following walking steps. It can also be considered as a segment corresponding to one walking step. In addition, depending on the arrangement location of the mobile device 100, it may be assumed that there is a difference in sensor information between the step of the right foot and the left foot. In that case, it is also conceivable to segment the sensor output data every two steps.

  The positioning data storage unit 180 stores the identification information received from the portable device 100 and the value of the positioning data as positioning data for each of the walking data. An example of the positioning data stored in the positioning data storage unit 180 is shown in FIG. In the example of FIG. 5, the latitude and longitude are recorded as a positioning result on the assumption that it is outdoors. However, when indoor positioning is used, it is also possible to record information indicating the facility name and the location inside the facility. .

  The data grouping unit 200 learns a group group in which walking data of a plurality of users are grouped based on each of the walking data segmented by the walking data segmentation unit 150. In the group group learned here, the user attribute of the user corresponding to the walking data belonging to the group may be given to each group of the group group. Further, the walking data may be grouped in consideration of the positioning data stored in the positioning data storage unit 180.

  Specifically, using the walking data segmented by the walking step unit by the walking data segmentation unit 150, a part of the walking data group linked to the user is grouped based on the similarity, and the characteristics of each group Create a vector. As an example of the grouping similarity calculation method, walking data is expressed as a feature vector based on walking data in units of walking steps by means of average and maximum acceleration, and similarities between vectors such as cosine similarity. It is possible to use the degree. In addition to the above feature vectors, user attributes (physique, sensor location, etc.) that represent previously acquired user attributes can be assigned to groups and used as constraints, It is also possible to incorporate location information into the feature vector and evaluate the similarity.

  As an example of grouping, a feature vector that represents a feature of each group by grouping the walking data group by a clustering algorithm represented by k-means (see Reference 1) based on the walking data group of a plurality of users Create As another method, at the time of learning, a part of walking data associated with each user is regarded as an individual group, and a feature vector expressing the characteristics of each group can be created.

[Reference 1]: S.P.Lloyd, "Least Squares Quantization in PCM", IEEE Trans. On Information Theory Vol. IT-28, No. 2,1982.

  The grouping information (group ID) of each group of the learned group group and the feature vector of each group are stored in the grouping information storage unit 205. FIG. 6 shows an example of the group ID and group feature vector of the stored group group.

  The learning classification data storage unit 163 stores classification data that serves as a teacher label for the step type learning unit 160 to learn the classification model. An example of the stored classification data is shown in FIG. As an example of a method for creating this data, it is conceivable that the input is performed manually simultaneously with the collection of learning sensor output data.

  The step type learning unit 160 is based on each of the walking data recognized by the walking step recognition unit 140, the group group learned by the data grouping unit 200, and the learning classification data storage unit 163. For each group, a classification model for recognizing any of a plurality of walking step states including a walking step on a flat ground, a walking step on a sloping ground, and a walking step on a step as the state of the walking step in the walking data is learned. In the present embodiment, the plurality of walking step states include a walking step on a flat ground, a walking step on an inclined ground, a walking step on an inclined basement, a walking step on a step up, and a walking step on a step down.

  The walking step state indicates the road surface condition such as whether the individual walking data divided into segments by the walking data segmentation unit 150 is walking on a flat ground, walking on an inclined surface, walking on a step, or the like. It is classified in the form corresponding to.

  In order to perform this classification, feature quantities that characterize walking data are acquired, and a classification model is constructed in a supervised learning framework. Examples of the feature amount are as follows. (1) Average direction acceleration, standard deviation, (2) Overall acceleration magnitude, (3) Interval between acceleration peaks before and after.

  In the learning of the classification model, the walking data segmentation unit 150 divides the walking data in units of walking steps, and the classification data stored in the learning classification data storage unit 163 and associated with the walking data; Based on the learning. In actual learning, a classification model is learned for each group based on the group group learned by the data grouping unit 200. One or more classification models are constructed for one group. As an example of the classification model learning algorithm, SVM (see Reference 2) or the like is used. The constructed classification model is managed by the learning model storage unit 165.

[Reference 2]: V.Vapnik, "Statistical Learning Theory", Wiley-Interscience, 1998.

  The learning model storage unit 165 stores the classification model learned by the step type learning unit 160.

<Operation of the road surface state learning system according to the embodiment of the present invention>

  Next, the operation of the road surface state learning device 300 of the road surface state learning system 10 according to the embodiment of the present invention will be described. The road surface state learning device 300 periodically detects that the sensor output data and the positioning data are stored and updated in the walking data storage unit 130 and the positioning data storage unit 180, and the road surface state learning shown in FIG. Execute the processing routine.

  First, in step S <b> 100, based on the sensor output data of each user stored in the walking data storage unit 130, a walking step that is an individual foot movement when the user walks in the sensor output data is recognized.

  Next, in step S102, based on the walking step recognized in step S100, each sensor output data of the user stored in the walking data storage unit 130 is segmented into each of the walking data corresponding to the walking step. .

  In step S <b> 104, based on each of the walking data segmented in step S <b> 102, a group group obtained by grouping a plurality of users' walking data is learned and stored in the grouping information storage unit 205.

  In step S106, based on each of the walking data recognized by the walking step recognition unit 140, the group group learned in step S104, and the classification data stored in the learning classification data storage unit 163, in the group group. For each group, a classification model for recognizing any of a plurality of walking step states is learned as a walking step state in the walking data, and stored in the learning model storage unit 165.

  As described above, according to the road surface state learning system according to the embodiment of the present invention, the mobile device 100 collects data, and the road surface state learning device 300 recognizes the walking step and corresponds to the walking step. To segment into each of the walking data, and to learn a group model that groups the walking data of a plurality of users based on each of the segmented walking data, and to learn a classification model for recognizing the walking step state Thus, it is possible to learn a group group and a classification model for accurately recognizing a road surface situation for each walking step of the user.

<Configuration of road surface condition understanding system according to an embodiment of the present invention>

  Next, the configuration of the road surface condition understanding system according to the embodiment of the present invention will be described. As shown in FIG. 9, the road surface situation understanding system 12 according to the embodiment of the present invention includes a mobile device 102, a road surface situation understanding device 302, and an output unit 50 that are carried by a plurality of users.

  The portable device 102 includes a walking data collection unit 112 and a positioning unit 122, acquires an output value and position information of a sensor related to walking together with a time stamp, and stores it in the storage area 103 as necessary. The configuration of the walking data collection unit 112, the positioning unit 122, and the storage area 103 in the portable device 102 is the same as that of the walking data collection unit 110, the positioning unit 120, and the storage area 101 of the portable device 100.

  The road surface condition understanding device 302 can be configured by a computer including a CPU, a RAM, and a ROM that stores a program and various data for executing a road surface condition understanding process routine described later.

  The road surface condition understanding device 302 includes a walking data accumulation unit 132, a walking step recognition unit 142, a walking data segmentation unit 152, a step type recognition unit 162, a learning model storage unit 166, and a walking data analysis result accumulation unit 172. A positioning data storage unit 182, a position-specific road surface condition understanding unit 192, a data grouping unit 202, and a grouping information storage unit 206.

  The walking data storage unit 132 stores sensor output data values received from the portable device 102. The format of the sensor output data stored is the same as that of the walking data storage unit 130.

  Based on the sensor output data stored in the walking data storage unit 132, the walking step recognition unit 142 recognizes the walking steps that are the steps of individual feet when the user walks in the sensor output data. The specific process is the same as that of the walking step recognition unit 140.

  The walking data segmentation unit 152 segments the sensor output data into each of the walking data corresponding to the walking step based on the walking step recognized by the walking step recognition unit 142. The specific processing is the same as that of the walking data segmentation unit 150.

  The positioning data storage unit 182 stores the identification information received from the mobile device 102 and the value of the positioning data as positioning data for each of the walking data. The format of the positioning data stored is the same as that of the positioning data storage unit 180.

  The grouping information storage unit 206 stores the grouping information (group ID) of each group of the group group learned in advance by the road surface condition learning device 300 and the feature vector of each group. The format of the stored data is the same as that of the grouping information storage unit 205.

  For each of the walking data segmented by the walking data segmentation unit 152, the data grouping unit 202 is stored in the grouping information storage unit 206, and a group group that groups the walking data of a plurality of users. Identify the group to which the walking data belongs. For example, a group of walking data may be specified by assigning to one or more clusters having a high similarity in feature vectors with the walking data among the groups created during learning and outputting the group information. Further, it is conceivable to specify k groups of walking data having a gait similar to walking data by the k-nearest neighbor method, and specify a group. One or more groups may be specified.

  In addition, the data grouping unit 202 identifies the group to which each of the walking data belongs based on the user attribute of the user who is walking in advance and the user attribute assigned to each group of the group group. It may be. For example, when the sensor arrangement is specified in the user attribute of the user who is walking, the group can be specified by narrowing down to the specific sensor arrangement. In addition, the data grouping unit 202 determines the walking data from the group group based on the positioning data for the walking data stored in the positioning data storage unit 182 for each of the walking data segmented by the walking data segmentation unit 152. You may specify the group to which. Thereby, it is possible to specify a group by narrowing down to a group at a close position from the latitude and longitude information incorporated in the feature vector of each group.

  The learning model storage unit 166 stores a classification model learned in advance by the road surface state learning device 300.

  For each of the walking data recognized by the walking step recognizing unit 142, the step type recognizing unit 162 uses the walking data group classification model identified by the data grouping unit 202 as the walking step state in the walking data. Any of a plurality of walking step states including a walking step on a flat ground, a walking step on a sloping ground, and a walking step on a step is recognized. Then, each recognition result of the walking step is stored in the walking data analysis result accumulation unit 172. Here, when the walking step state is recognized, the learning model storage unit 166 is referred to, and the walking step is classified using the classification model learned for the group specified by the data grouping unit 202. Although there may be one classification model used at the time of recognition, it is also conceivable to use a plurality of classification models such as evaluating and classifying a plurality of determination results with weights.

  The walking data analysis result accumulating unit 172 stores a recognition result including a walking step state for each walking step recognized by the step type recognition unit 162. The road surface condition understanding device 302 outputs the recognition result accumulated in the walking data analysis result accumulation unit 172 to the output unit 50 periodically or in response to an update or the like. As shown in FIG. 10, an example of the recognition result stored in the walking data analysis result accumulation unit 172 is that each record corresponds to a walking step, a terminal ID, a time stamp start point, a time stamp end point, and an analysis result (walking Step state). At least walking data for identifying the walking data and a recognition result for the walking data are stored. As an example of recognizing walking data, it is conceivable to record the terminal ID of the terminal device that acquired the data and the start point and end point of the time stamp at the time of data acquisition.

  The position-specific road surface state understanding unit 192 totalizes the positioning data for each of the walking data stored in the positioning data storage unit 182 and the walking step state for each of the walking data stored in the walking data analysis result storage unit 172. The road surface condition at each position in the positioning data is recognized and output to the output unit 50. By recognizing the road surface condition, it is possible to grasp which place is what kind of road surface condition. In addition, using the road surface condition recognition result, for example, it is possible to perform batch processing to generate a map with road surface condition that can grasp the road surface condition, or to acquire the road surface condition in the future direction by online processing Become.

<Operation of the road surface condition understanding system according to the embodiment of the present invention>

  Next, the operation of the road surface state understanding device 302 of the road surface state understanding system 12 according to the embodiment of the present invention will be described. The road surface condition understanding device 302 periodically detects that sensor output data and positioning data are accumulated and updated in the walking data accumulation unit 132 and the positioning data accumulation unit 182, and understands the road surface condition shown in FIG. 11. Execute the processing routine.

  First, in step S200, based on the sensor output data stored in the walking data storage unit 132, a walking step, which is a step of an individual foot when the user walks, in the sensor output data is recognized.

  Next, in step S202, based on the walking step recognized in step S100, the sensor output data is segmented into each of the walking data corresponding to the walking step.

  In step S204, for each of the walking data segmented by the walking data segmenting unit 152, the walking data is stored from the group group in which the walking data of a plurality of users is grouped and stored in the grouping information storage unit 206. Identify the group to which you belong.

  In step S206, for each of the walking data recognized by the walking step recognition unit 142, the state of the walking step in the walking data is recognized based on the classification model of the group of the walking data specified in step S204. The result is stored in the analysis result accumulation unit 172.

  In step S208, the positioning data for each of the walking data stored in the positioning data storage unit 182 and the walking step state for each of the walking data stored in the walking data analysis result storage unit 172 in step S200 are tabulated. Recognize the road surface condition of each position of positioning data.

  In step S210, the recognition result including the walking step state of each walking step recognized in step S206 and the road surface condition of each position recognized in step S208 are output to the output unit 50, and the process ends.

  As described above, according to the road surface condition understanding system according to the embodiment of the present invention, the mobile device 100 collects data, and the road surface state learning apparatus 300 recognizes the walking step and corresponds to the walking step. Segmented into each of the walking data, for each of the segmented walking data, identify the group to which the walking data belongs from the group group, recognize any of the walking step state based on the group of walking data, By counting the positioning data and the walking step state and recognizing the road surface state at each position, the road surface state for each walking step of the user can be accurately recognized.

  The present invention is not limited to the above-described embodiment, and various modifications and applications can be made without departing from the gist of the present invention.

  For example, in the embodiment described above, the road surface situation understanding device 302 includes the position-specific road surface situation understanding unit 192, and the position-specific road surface situation understanding unit 192 is limited to an example of recognizing the road surface situation at each position. However, the present invention is not limited to this, and the road surface condition understanding device 302 may not include the position-specific road surface condition understanding unit 192 and may recognize each walking step state of the walking step.

  Moreover, in the data grouping unit 200 of the road surface state learning device 300 described above, the case of learning a group group in which walking data of a plurality of users is grouped has been described as an example, but the present invention is not limited to this. You may make it learn the group which grouped these users. In this case, based on the user attributes, users having similar physiques or users having the same sensor placement location are grouped to create a feature vector for each group. Further, the walking data segmentation unit 152 of the road surface condition understanding device 300 may specify a group to which walking data belongs from a group group in which a plurality of users are grouped.

DESCRIPTION OF SYMBOLS 10 Road surface condition learning system 12 Road surface condition understanding system 50 Output part 100, 102 Portable device 101, 103 Storage area 110, 112 Walking data collection part 120, 122 Positioning part 130, 132 Walking data storage part 140, 142 Walking step recognition part 150 , 152 Walking data segmentation unit 160, 162 Step type learning unit 163 Learning classification data storage unit 165, 166 Learning model storage unit 172 Walking data analysis result storage unit 180, 182 Positioning data storage unit 192 Positional road surface condition understanding unit 200 202 Data grouping unit 205, 206 Grouping information storage unit 300 Road surface state learning device 302 Road surface state understanding device

Claims (7)

Based on the sensor output data in which the walking step recognition unit records the output value of the sensor provided in the portable device carried by the user along with the time stamp in time series, the sensor output data when the user walks Recognizing walking steps that are individual steps,
A step of segmenting the sensor output data into each of the walking data corresponding to the walking step based on the walking step recognized by the walking step recognition unit;
A group group in which a plurality of users are grouped, or the walking data of the plurality of users is obtained in advance for each of the walking data segmented by the walking data segmentation unit. Identifying a group to which the walking data belongs from the grouped group,
A step type recognizing unit for recognizing a walking step state in the walking data based on the group of the walking data specified by the data grouping unit for each of the walking data recognized by the walking step recognizing unit; When,
The position-specific road surface condition understanding unit is recognized by the positioning data for each of the walking data and the step type recognition unit based on the positioning data obtained by recording the result of measuring the position where the portable device is present together with a time stamp. Counting the walking step state for each of the walking data and recognizing the road surface condition at each position;
How to understand road surface conditions. Each group of the group group is given a user attribute representing the attribute of the user belonging to the group, or a user attribute of the user of the walking data belonging to the group,
The step of identifying the group by the data grouping unit is based on the user attribute of the user who is walking in advance acquired and the user attribute assigned to each group of the group group. The road surface state understanding method according to claim 1, wherein a group to be identified is specified. The group group is a group group in which the plurality of users are grouped in consideration of the positioning data, or a group group in which the walking data of the plurality of users is grouped in further consideration of the positioning data. There,
The step of identifying the group by the data grouping unit includes, for each of the walking data segmented by the walking data segmentation unit, the walking data from the group group based on positioning data for the walking data. The road surface state understanding method according to claim 1, wherein a group to which the user belongs is specified. Based on the sensor output data recorded in time series with the time stamp of the output value of the sensor provided in the portable device carried by the user, the individual output when the user walks in the sensor output data A walking step recognition unit for recognizing a certain walking step;
A walking data segmenting unit that segments the sensor output data into each of the walking data corresponding to the walking step based on the walking step recognized by the walking step recognition unit;
For each of the walking data segmented by the walking data segmentation unit, a group group in which a plurality of users are grouped or a group group in which the walking data of the plurality of users is grouped is obtained in advance. From the data grouping unit for identifying the group to which the walking data belongs,
For each of the walking data recognized by the walking step recognition unit, a step type recognition unit for recognizing the walking step state in the walking data based on the group of the walking data specified by the data grouping unit;
The positioning data for each of the walking data and the walking data recognized by the step type recognition unit based on the positioning data recorded with the time stamp of the result of measuring the position where the portable device exists. By summing up the walking step state, the road surface condition understanding unit by position that recognizes the road surface state of each position,
Road surface condition understanding device. Each group of the group group is given a user attribute representing the attribute of the user belonging to the group, or a user attribute of the user of the walking data belonging to the group,
The data grouping unit specifies a group to which each of the walking data belongs, based on a user attribute of a walking user acquired in advance and a user attribute assigned to each group of the group group. Item 5. Road surface condition understanding device. The group group is a group group in which the plurality of users are grouped in consideration of the positioning data, or a group group in which the walking data of the plurality of users is grouped in further consideration of the positioning data. There,
The data grouping unit identifies, for each of the walking data segmented by the walking data segmentation unit, a group to which the walking data belongs from the group group based on positioning data for the walking data. The road surface condition understanding device according to claim 4.   The program for making a computer perform each step of the road surface condition understanding method of any one of Claims 1-3.