JP6926644B2 - Anomaly estimation device and display device - Google Patents

Description

The present disclosure relates to an abnormality estimation device that estimates an abnormality transition and a display device that displays an estimation result by the abnormality estimation device.

In Patent Document 1, a time range having a travel time pattern similar to the time transition of the travel time up to the present in the road section to be predicted is determined based on the past travel time actual data, and the time range beyond the determined time range is determined. A prediction device that predicts the travel time ahead of the road section to be predicted by referring to the travel time actual data is described.

Further, in Patent Document 1, when a traffic jam occurs, the traffic jam is generated in the road section connected to the upstream. Therefore, not only the road section to be predicted but also the downstream road section connected to the road section to be predicted is used for the prediction. It is stated that it will be selected as a section.

Japanese Patent No. 3792172

However, in the technique described in Patent Document 1, when calculating a detour route for avoiding congestion, a route from the current position to the destination point is comprehensively generated, and congestion occurs for each generated route. There was a problem that it was necessary to judge whether or not it was present.

An object of the present disclosure is to calculate a detour route at high speed.

One aspect of the present disclosure is a collection unit (S10), a feature amount calculation unit (S20, S30), an abnormality determination unit (S110 to S140), a storage unit (S150, S160), and an information creation unit (S210). It is an abnormality estimation device (7) including S265) and an estimation unit (S310 to S330).

The collecting unit is configured to repeatedly collect vehicle data regarding the state of the vehicle for each of a plurality of vehicles.
The feature amount calculation unit is configured to calculate the feature amount from the vehicle data collected by the collection unit, and store the feature amount in association with the location corresponding to the feature amount.

The abnormality determination unit is configured to determine whether or not an abnormality occurrence point where an abnormality has occurred exists at the present time, based on the feature amount calculated by the feature amount calculation unit.
When the abnormality judgment unit determines that the abnormality occurrence point exists at the present time, the storage unit uses the vehicle data at the abnormality occurrence point and the vehicle data at the abnormality peripheral point which is a point around the abnormality occurrence point. Therefore, it is configured to create preset estimation data for estimating the transition of the abnormality from the abnormality occurrence point to the abnormality peripheral point, and to accumulate the created estimation data.

The information creation unit uses the estimation data accumulated by the storage unit to create causal relationship information indicating the causal relationship between the abnormality that occurred at the abnormality occurrence point and the abnormality that occurred at the abnormality peripheral point. It is composed.

When the abnormality judgment unit determines that the abnormality occurrence point exists at the present time, the estimation unit uses the past causal relationship information created by the information creation department to determine that the abnormality occurrence point exists at the present time. It is configured to estimate the transition of the anomaly from the anomaly occurrence point determined by the above to the anomaly peripheral point.

The anomaly estimation device of the present disclosure configured in this way creates and accumulates estimation data created using vehicle data of the abnormality occurrence point and the abnormality peripheral point, and uses the accumulated estimation data for a causal relationship. Creating information. And the causal relationship information shows the causal relationship between the abnormality that occurred at the abnormality occurrence point and the abnormality that occurred at the abnormality peripheral point. Therefore, when it is determined that the abnormality occurrence point exists at the present time, the abnormality estimation device of the present disclosure causes an abnormality to occur at the abnormality occurrence point (that is, the abnormality peripheral point) that transitions from the abnormality occurrence point as the starting point. It can be estimated by extracting past causal relationship information showing a causal relationship with the abnormality that occurred at the point. As a result, the abnormality estimation device of the present disclosure can estimate the transition of the abnormality in a simple method and at high speed at the time of occurrence of the abnormality by accumulating the causal relationship information, and can identify the influence range of the abnormality at high speed. The detour route can be calculated at high speed.

In addition, the reference numerals in parentheses described in this column and the scope of claims indicate the correspondence with the specific means described in the embodiment described later as one embodiment, and the technical scope of the present disclosure is defined. It is not limited.

It is a figure which shows the structure of the abnormality detection system 1. It is a block diagram which shows the structure of an abnormality detection apparatus 7. It is a functional block diagram which shows the outline of the process which a control device 15 executes. It is a figure which shows the vehicle behavior data, a driving scene, and a topic ratio. It is explanatory drawing which shows the division process of a driving scene. It is explanatory drawing which shows the calculation process of a topic ratio. It is a figure explaining the comparison between the present topic ratio and the past topic ratio. It is a figure explaining the calculation method of the degree of difference. It is a figure which shows the periphery of the abnormality occurrence node. It is a figure explaining the data accumulated in the road abnormality accumulation database 42. It is a figure which shows the cumulative abnormality degree histogram. It is a figure explaining the comparison of the anomalous transition graph. It is a figure which shows the abnormal transition map. It is a flowchart which shows the operation state extraction processing. It is a flowchart which shows the abnormality detection processing. It is a flowchart which shows the causal relation extraction process of 1st Embodiment. It is a flowchart which shows the abnormality estimation processing of 1st Embodiment. It is a figure which shows the route guidance which avoided the abnormality occurrence node. It is a flowchart which shows the causal relation extraction process of 2nd Embodiment. It is a flowchart which shows the abnormality estimation process of 2nd Embodiment. It is a flowchart which shows the causal relationship extraction process of 3rd Embodiment. It is a flowchart which shows the abnormality estimation processing of 3rd Embodiment. It is a figure which shows the periphery of the abnormality occurrence node in another embodiment. It is a figure explaining the comparison of the abnormal transition graph in another embodiment. It is a figure which shows the abnormal transition map in another embodiment. It is a figure which shows the vehicle-mounted device 5 and the navigation device 9 mounted on a vehicle. It is a figure which shows the display of the navigation device 9 in 4th Embodiment. It is a figure which shows the display of the navigation device 9 in the modification 8. It is a figure which shows the display of the navigation device 9 in the modification 10. It is a figure which shows the display of the navigation device 9 in the modification 11. It is a figure which shows the display of the navigation device 9 in the modification 12. It is the first figure which shows the display of the navigation device 9 in the modification 13. FIG. 2 is a second diagram showing a display of the navigation device 9 in the modified example 13. It is a figure which shows the display of the navigation device 9 in the modification 14. It is a figure which shows the display of the navigation device 9 in the modification 15.

(First Embodiment)
The first embodiment of the present disclosure will be described below together with the drawings.
As shown in FIG. 1, the abnormality detection system 1 of the present embodiment wirelessly communicates between a plurality of roadside units 3 dispersedly installed near the traveling path of an automobile and the roadside units 3 mounted on the automobile. The vehicle-mounted device 5 is provided with the vehicle-mounted device 5 and the abnormality detection device 7 connected to the roadside device 3 via a wired network NW.

As shown in FIG. 2, the abnormality detection device 7 includes a communication device 11, a data storage device 13, and a control device 15.
The communication device 11 performs data communication with the roadside unit 3 via the wired network NW. The roadside unit 3 has at least the vehicle-mounted device position information indicating the position of the vehicle-mounted device 5, the identification information for identifying the vehicle-mounted device 5, the transmission time information indicating the time when the data is transmitted, and the operation data described later. The behavior data described later and the captured image data described later are transmitted to the abnormality detection device 7. The vehicle-mounted device position information indicates a position detected based on a GPS signal or the like received via a GPS antenna of a vehicle equipped with the vehicle-mounted device 5. GPS is an abbreviation for Global Positioning System.

The data storage device 13 is a device for storing various types of data, and is, for example, a hard disk drive in the present embodiment.
The control device 15 is an electronic control device mainly composed of a well-known microcomputer provided with a CPU 21, a ROM 22, a RAM 23, and the like. Various functions of the microcomputer are realized by the CPU 21 executing a program stored in a non-transitional substantive recording medium. In this example, the ROM 22 corresponds to a non-transitional substantive recording medium in which the program is stored. In addition, by executing this program, the method corresponding to the program is executed. The number of microcomputers constituting the control device 15 may be one or a plurality.

As shown in FIG. 3, the control device 15 has an operation status extraction unit 31, an abnormality detection unit 32, a causal relationship extraction unit 33, and an abnormality estimation as a configuration of functions realized by the CPU 21 executing a program. A unit 34 is provided. The method for realizing these elements constituting the control device 15 is not limited to software, and some or all of the elements may be realized by using one or more hardware. For example, when the above function is realized by an electronic circuit which is hardware, the electronic circuit may be realized by a digital circuit including a large number of logic circuits, an analog circuit, or a combination thereof.

Further, the data storage device 13 includes an operation status storage database 41, a road abnormality storage database 42, and a causal relationship storage database 43. Further, the data storage device 13 stores road map data. In road map data, a road is defined by a node and a link connecting the nodes. The node data is composed of a node ID, which is a node identification number, node coordinates, a node type (for example, a type such as an intersection or a confluence). The link data is composed of the link ID which is the identification number of the link, the link length, the node ID of each node connected to the start point and the end point, the road type (for example, the type of expressway, toll road, general road, etc.). ing.

As shown in FIG. 4, the driving situation extraction unit 31 segments the data showing the behavior of the vehicle in time series, and sets the segmented data as the driving scenes. Further, the driving situation extraction unit 31 calculates the topic ratio for each of the plurality of driving scenes by using, for example, the topic model disclosed in Japanese Patent Application Laid-Open No. 2014-235605.

As shown in FIG. 3, the driving situation extraction unit 31 includes a vehicle behavior data collecting unit 51, a driving scene dividing unit 52, and a topic ratio calculation unit 53.
The vehicle behavior data collecting unit 51 repeatedly collects vehicle-mounted device position information, driving data, behavior data, and captured image data from a plurality of vehicle-mounted devices 5 via the roadside device 3. The driving data is data related to a driving operation in which the driver drives a vehicle equipped with the vehicle-mounted device 5. The behavior data is data relating to the behavior of the vehicle that appears as a result of a driving operation in which the driver drives the vehicle equipped with the vehicle-mounted device 5. The captured image data is data showing an image taken by a front camera attached to the vehicle in order to continuously capture the scenery in front of the vehicle that the driver can see through the windshield.

The vehicle behavior data collection unit 51 generates differential data obtained by differentiating each of the driving data and the behavior data, and generates multidimensional data composed of the driving data, the behavior data, and the differential data as the vehicle behavior data. As the driving data, for example, the operating amount of the accelerator pedal, the operating amount of the brake pedal, the operating amount of the steering wheel, the operating state of the direction indicator, the shift position of the transmission, and the like can be used. Further, as the behavior data, for example, the speed of the vehicle, the yaw rate, and the like can be used.

The driving scene segmentation unit 52 statistically analyzes the vehicle behavior data using a model from the driver's environmental recognition to the operation, and extracts the switching point of the driving scene felt by the driver to obtain a time series of the vehicle behavior data. Is segmented into multiple sub-series, each representing some driving scene.

Specifically, a double segment analyzer that performs segmentation by the unsupervised driving scene division method using the double segment structure is used. As shown in FIG. 5, in the double segment analyzer, first, the clusters representing the states of various vehicles grasped from the vehicle behavior data in the multidimensional space representing the range of the vehicle behavior data, and the transitions between the clusters. The probability is defined in advance. Then, the double segment analyzer statistically processes which cluster the generated vehicle behavior data belongs to by using these defined information, so that the time series of the vehicle behavior data can be classified as a vehicle. (That is, each cluster). However, by associating each cluster with a symbol for identification in advance, the time series of vehicle behavior data is converted into a symbol string indicating which cluster it belongs to. For the generation of this symbol string, for example, a hierarchical Dirichlet process hidden Markov model (hereinafter, HDP-HMM), which is one of the models expressed by the hidden state and the stochastic transition between the hidden states, can be used.

Next, the double segment analyzer uses the Nested Pitman-Yor Language Model (hereinafter, NPYLM), which is an example of an unsupervised chunking method for discrete character strings using statistical information, to convert the generated symbol string into NPYLM. Segment into sub-series that mean a given driving scene. At this time, the dictionary size (that is, the number of subseries) is made as small as possible, and the probability of generating the entire symbol string consisting of the sequence of subseries is maximized. This makes it possible to segment the vehicle behavior data into driving scenes. However, as the transition probability between the sub-series and the generation probability of the sub-series, those generated in advance by learning are used.

Regarding the double segment analyzer to which HDP-HMM and NPYLM are applied, the non-patent document separately published by the applicant of the present application, T. Taniguchi et al, "Semiotic Prediction of Driving Behavior using Unsupervised Double Articulation Analyzer" IEEE Intelligent Vehicles Symposium, 2012 and K. Takenaka et al, "Contextual Scene Segmentation of Driving Behavior based on Double Articulation Analyzer" IEEE / RSJ International Conference on Intelligent Robots and Systems, 2012, etc. Is omitted. However, the method used for symbol generation and symbol segmentation is not limited to HDP-HMM and NPYLM, and other methods may be used.

The topic ratio calculation unit 53 generates a vehicle behavior histogram corresponding to the feature amount distribution for each driving scene generated by the driving scene dividing unit 52, using the vehicle behavior data belonging to the partial series corresponding to the driving scene.

Then, the topic ratio calculation unit 53 calculates and calculates the topic ratio obtained by expressing the generated vehicle behavior histogram by the weighted sum of a plurality of characteristic distributions (that is, driving topics) prepared in advance. The topic ratio is stored in the operation status accumulation database 41.

The vehicle behavior histogram represents the feature amount of the vehicle behavior data belonging to the partial series corresponding to the driving scene of interest as a distribution. For example, when multidimensional data consisting of accelerator pedal operation amount, brake pedal operation amount, steering operation amount, vehicle speed, and differential data of each of these is used as a feature amount, a feature space representing the value range of the multidimensional data is used. The index is divided into a plurality of indexes, and a histogram expressing the frequency of data appearance is generated for each index. As shown in FIG. 6, a histogram is generated for each individual data, and the generated histogram is concatenated to be called a vehicle behavior histogram.

The driving topic consists of K (eg, 100) basis feature distributions Topic 1 to Topic K used when the generated vehicle behavior histogram is represented by a mixture of multiple distributions (ie, histograms).

The topic ratio is a mixing ratio obtained assuming that the vehicle behavior histogram is expressed by mixing the basal feature distributions, and represents the content ratio of each driving topic to the vehicle behavior histogram.

To generate the topic ratio, each sub-series corresponding to one of the driving scenes is regarded as "one document", and the observed feature distribution (here, vehicle behavior histogram) is regarded as "one word". , Generated using the topic estimation method used in the field of natural language processing. Here, the latent Dirichlet allocation method (hereinafter, LDA) is used. LDA is an abbreviation for Latent Dirichlet Allocation. When the vehicle behavior histogram to be processed is composed of a plurality of types of histograms, a multimodal LDA, which is an extension of the LDA, is used. Specifically, using approximation methods such as variational Bayes and Gibbs sampling that realize LDA (or multimodal LDA), only E step is executed among the E step and M step processes performed in these processes. It can be obtained by.

For more information on these techniques, see, for example, D. Blei et al, "Latent Dirichlet Allocation" Journal of Machine Learning Research, 2003, and T. Griffiths & M. Steyvers, "Finding Scientific Topics," Proceedings of the National Academy. Since it is described in of Sciences, 2004, etc., the description is omitted here. Further, since the method for generating the basal feature distribution is a known method described in, for example, Japanese Patent Application Laid-Open No. 2014-235605, the description thereof will be omitted.

As shown in FIG. 7, the abnormality detection unit 32 extracts the topic ratio from the operation status accumulation database 41, and the degree of difference between the current topic ratio and the past topic ratio in the same node as the current topic ratio. Is calculated. As will be described later, the topic ratio is associated with the node ID of the corresponding node based on the on-board unit position information. Therefore, the abnormality detection unit 32 extracts the topic ratio associated with the node ID of the current topic ratio and the same node ID as the "past topic ratio in the node having the same current topic ratio".

Specifically, as shown in FIG. 8, the abnormality detection unit 32 first calculates the distance between the current topic ratio and the past topic ratio (for example, the Euclidean distance) by using the driving topic of the topic ratio as a feature amount. do. Then, the abnormality detection unit 32 calculates the average value of the calculated one or a plurality of distances, and uses this average value as the degree of difference. Further, when the calculated degree of difference is larger than the preset first abnormality occurrence determination value, the abnormality detection unit 32 determines that an abnormality has occurred in the same node as the current topic ratio. Further, the abnormality detection unit 32 assigns an abnormality identification number for distinguishing other abnormalities to this abnormality.

Then, the abnormality detection unit 32 associates the received data associated with the node determined to have caused the abnormality (hereinafter referred to as the abnormality occurrence node) with the nodes around the abnormality occurrence node until it is determined that the abnormality has disappeared. Based on the received data received, the accumulated data described later is created. Whether or not it is associated with the abnormality occurrence node and whether or not it is associated with the nodes around the abnormality occurrence node are determined by the position of the node and the position indicated by the on-board unit position information included in the received data. Judgment is based on. Specifically, among the plurality of nodes, the node closest to the position indicated by the vehicle-mounted device position information is associated with the received data including the vehicle-mounted device position information.

The abnormality detection unit 32 stores the created accumulated data in the road abnormality accumulation database 42. The abnormality detection unit 32 stores the accumulated data in a state associated with the corresponding node ID. As shown in FIG. 9, the abnormality detection unit 32 determines that the inside of the peripheral setting circle Ca having the peripheral setting radius Ra set in advance about the abnormality occurrence node Pt is “around the abnormality occurrence node”. do.

The abnormality detection unit 32 determines that the abnormality has disappeared when the calculated degree of difference is equal to or less than the first abnormality occurrence determination value. Further, the abnormality detection unit 32 receives the accumulated data accumulated from the time when it is determined that an abnormality has occurred (hereinafter, when the abnormality has occurred) to the time when it is determined that the abnormality has disappeared (hereinafter, when the abnormality disappears). And give the same abnormality identification number.

As shown in FIG. 10, the accumulated data accumulated in the road abnormality accumulation database 42 by the abnormality detection unit 32 includes node ID, date and time, latitude, longitude, and node-specific information (for example, in front of an intersection, in front of a station, in front of an event venue, etc.). ), Abnormality, topic ratio. As the degree of abnormality, for example, the cumulative degree of difference between the most recent plurality of vehicles (for example, 5 vehicles) can be mentioned. It should be noted that these accumulated data may be the average of a plurality of vehicles, representative data, or data of all of the plurality of vehicles.

As shown in FIG. 11, the causal relationship extraction unit 33 uses the data accumulated from the time of determining the occurrence of the abnormality to the time of determining the disappearance of the abnormality to determine the cumulative abnormality degree by a preset abnormality degree calculation time (for example, 5). It is calculated for each minute), a cumulative abnormality degree histogram is created, and it is stored in the road abnormality accumulation database 42. The abnormality degree calculation timing for calculating the cumulative abnormality degree comes every time the abnormality degree calculation time elapses with the time when the abnormality occurrence is determined as the first abnormality degree calculation timing and the first abnormality degree calculation timing as the starting point. For the cumulative abnormality degree, the operation data and behavior data of the required number (for example, 5) transmitted from the roadside machine 3 immediately before the abnormality degree calculation timing are used at each of the one or multiple abnormality degree calculation timings. It is the total value of the degree of difference calculated by. The cumulative anomaly degree histogram is a histogram in which the vertical axis is the cumulative anomaly degree and the horizontal axis is the time. The causal relationship extraction unit 33 creates a cumulative anomaly degree histogram for each of the anomaly occurrence node and one or a plurality of nodes around the anomaly occurrence node.

Then, the causal relationship extraction unit 33 determines the abnormality occurrence time and the abnormality disappearance time for each of the cumulative abnormality degree histograms created by the plurality of nodes. The abnormality occurrence time is a time when the cumulative abnormality degree becomes equal to or higher than the preset second abnormality occurrence determination value. The abnormality disappearance time is a time when the cumulative abnormality degree becomes less than the second abnormality occurrence determination value.

As shown in FIG. 12, the causal relationship extraction unit 33 creates an anomaly transition graph using a cumulative anomaly degree histogram for each of the anomaly occurrence node and its surrounding nodes. The anomaly transition graph is a graph in which the heights of a plurality of bins constituting the cumulative anomalyness histogram are plotted and the plotted points are connected by a straight line. In FIG. 12, as shown by the arrow MG1, an abnormality transition graph is created from the cumulative abnormality degree histogram in the abnormality occurrence node, and as shown by the arrow MG2, the abnormality transition graph is created from the cumulative abnormality degree histogram in the node closest to the abnormality occurrence node. Indicates that you are creating.

Then, the causal relationship extraction unit 33 analyzes the causal relationship between the change in the abnormality shown by the abnormality transition graph in the abnormality occurrence node and the change in the abnormality shown in the abnormality transition graph in the nodes around the abnormality occurrence node. Evaluate the effect of the anomaly that occurred in on the nodes around the anomaly occurrence node. In the present embodiment, the causal relationship extraction unit 33 uses, for example, a Granger causal model to perform a causal relationship analysis between the anomaly-occurring node and each of a plurality of nodes around the anomaly-occurring node. In FIG. 12, as shown by arrows CG1 and arrow CG2, it is shown that the causal relationship analysis is performed by comparing the abnormality transition graph in the abnormality occurrence node with the abnormality transition graph in the node closest to the abnormality occurrence node.

Further, the causal relationship extraction unit 33 uses the abnormal transition graph of the abnormal occurrence node and the abnormal transition graph of a plurality of nodes around the abnormal occurrence node to show how the abnormality is transitioning with the passage of time. Create the time transition information shown.

Then, when the causal relationship extraction unit 33 determines that there is a causal relationship between the abnormal occurrence node and the nodes around the abnormal occurrence node as a result of performing the causal relationship analysis, the causal relationship information shown below is used as the causal relationship. Stored in the relationship accumulation database 43.

The causal relationship information includes anomaly node information, occurrence time information, disappearance time information, transition time information, anomaly degree information, and feature amount information.
The abnormal node information includes information for specifying the position of the node in which the abnormality has occurred and information for specifying the position of the node to which the abnormality has transitioned (hereinafter referred to as the abnormal transition node).

The occurrence time information includes information indicating the time when the abnormality occurred at the abnormality occurrence node and information indicating the time when the abnormality occurred at the abnormality transition node.
The transition time information includes information indicating the time required for the abnormality to transition from the abnormality occurrence node to the abnormality transition node.

The anomaly degree information includes information indicating the degree of abnormality in the abnormality occurrence node and information indicating the degree of abnormality in the abnormality transition node.
The feature amount information is the information of the feature amount (topic ratio in this embodiment) used to calculate the degree of abnormality in the abnormality occurrence node and the feature amount used to calculate the degree of abnormality in the abnormality transition node. Including information.

As a result of the causal relationship analysis, the causal relationship information of the abnormality determined to have a causal relationship with the abnormality generated in the abnormality occurrence node is the same abnormality identification number as the abnormality identification number stored in the road abnormality accumulation database 42. It is stored in the causal relationship accumulation database 43 in a numbered state.

When the abnormality detection unit 32 detects an abnormality, the abnormality estimation unit 34 provides information indicating an abnormality of the abnormality transition node having a causal relationship with the abnormality occurrence node corresponding to the abnormality detected by the abnormality detection unit 32 in the causal relationship accumulation database 43. Extract from. Then, the abnormality estimation unit 34 maps the transition of the abnormality as shown in FIG. 13 by using the number of extracted abnormalities having a causal relationship. FIG. 13 shows an abnormality transition map when an abnormality occurs in a node having a node ID of 50.

In FIG. 13, nodes having node IDs of 51, 53, and 55 are arranged in the vicinity of the node having the node ID of 50, respectively. The arrow TR1 starting from the node with the node ID of 50 and ending at the node with the node ID of 51 indicates an abnormal transition from the node with the node ID of 50. The number "80" written near the arrow TR1 indicates the number of cases extracted as an abnormality having a causal relationship with an abnormality occurring in a node having a node ID of 50 in a node having a node ID of 51.

Similarly, the arrow TR2 indicates an abnormal transition from a node with a node ID of 50 to a node with a node ID of 53. The number "21" written near the arrow TR2 indicates the number of cases extracted as an abnormality having a causal relationship with an abnormality occurring in a node having a node ID of 50 in a node having a node ID of 53.

Further, the abnormality estimation unit 34 creates an abnormality transition map so that the corresponding arrow becomes thicker as the number of extraction cases increases. In FIG. 13, the arrow TR1 is described so as to be thicker than the arrow TR2. When the two nodes are not connected by an arrow, it indicates that there is no abnormal transition between these two nodes.

Next, the procedure of the operation status extraction process executed by the control device 15 will be described. The operation status extraction process corresponds to the operation status extraction unit 31, and is a process that is repeatedly executed during the operation of the control device 15.

When this operation status extraction process is executed, the CPU 21 of the control device 15 first uses the operation data and the behavior data received from the roadside unit 3 in the communication device 11 in S10 to behave as the vehicle, as shown in FIG. Generate data. The processing of S10 corresponds to the vehicle behavior data collecting unit 51.

Then, in S20, the vehicle behavior data is statistically analyzed, and the time series of the vehicle behavior data is segmented into a plurality of driving scenes. The processing of S20 corresponds to the driving scene dividing unit 52. Further, in S30, the topic ratio is calculated for each driving scene, and the calculated topic ratio is stored in the driving status accumulation database 41 in a state of being associated with the node ID of the node corresponding to the received in-vehicle device position information. The operation status extraction process is temporarily terminated. The "node corresponding to the vehicle-mounted device position information" is the node closest to the position indicated by the vehicle-mounted device position information, and one node is associated with one vehicle-mounted device position information. The process of S30 corresponds to the topic ratio calculation unit 53.

Next, the procedure of the abnormality detection process executed by the control device 15 will be described. The abnormality detection process corresponds to the abnormality detection unit 32 and is a process that is repeatedly executed during the operation of the control device 15.
When this abnormality detection process is executed, as shown in FIG. 15, the CPU 21 of the control device 15 first determines in S110 from the topic ratios stored in the operation status accumulation database 41 that an abnormality has been detected. Extract the most recent topic percentages that are not used in.

Then, in S120, the degree of difference between the extracted topic ratio and the past topic ratio in the same node is calculated for the latest topic ratio extracted in S110.
Further, in S130, the presence or absence of abnormality is determined. Specifically, for each of the calculated degrees of difference, it is determined whether or not the calculated degree of difference is larger than the preset first abnormality occurrence determination value. Here, when the degree of difference is larger than the first abnormality occurrence determination value, it is determined that an abnormality has occurred in the node corresponding to this degree of difference. On the other hand, when the degree of difference is equal to or less than the first abnormality occurrence determination value, it is determined that no abnormality has occurred in the node corresponding to this degree of difference.

Then, in S140, based on the determination result in S130, it is determined whether or not there is a node in which an abnormality has occurred (hereinafter, an abnormality occurrence node). Here, if there is no abnormality occurrence node, the abnormality detection process is temporarily terminated. On the other hand, when there is an abnormality-occurring node, in S150, the above-mentioned accumulated data is used by using the received data associated with the abnormality-occurring node and the received data associated with the nodes around the abnormality-occurring node. Is created, and the created accumulated data is stored in the road abnormality accumulation database 42. Since the details of the accumulated data have already been described, the description here will be omitted.

Then, in S160, it is determined whether or not the abnormality has disappeared at the abnormality occurrence node. Specifically, first, from the topic ratio stored in the operation status accumulation database 41, the latest topic ratio that has not yet been used for determining the abnormality detection in the abnormality occurrence node is extracted. In addition, the degree of difference is calculated for the extracted topic ratio, and it is determined whether or not the calculated degree of difference is larger than the first abnormality occurrence determination value. Here, when the degree of difference is larger than the first abnormality occurrence determination value, it is determined that the abnormality continues at the abnormality occurrence node. On the other hand, when the degree of difference is equal to or less than the first abnormality occurrence determination value, it is determined that the abnormality has disappeared at the abnormality occurrence node.

If it is determined in S160 that the abnormality does not disappear in the abnormality occurrence node, the process proceeds to S150. On the other hand, when it is determined that the abnormality has disappeared in the abnormality occurrence node, the abnormality detection process is temporarily terminated.

Next, the procedure of the causal relationship extraction process executed by the control device 15 will be described. The causal relationship extraction process corresponds to the causal relationship extraction unit 33, and is a process that is repeatedly executed during the operation of the control device 15.

When this causal relationship extraction process is executed, as shown in FIG. 16, the CPU 21 of the control device 15 first performs causal relationship analysis from the data stored in the road abnormality accumulation database 42 in S210. Determine if there is any recent data that has not been used. Here, if there is no latest data that has not been used for the causal relationship analysis, the causal relationship extraction process is temporarily terminated.

On the other hand, when there is the latest data that has not been used for the causal relationship analysis, first, in S210, the data to which the same abnormality identification number is assigned is extracted from the road abnormality accumulation database 42. Then, using the extracted data, a cumulative anomaly degree histogram is created for each of the anomaly occurrence node and the nodes around it.

Further, in S230, an abnormality transition graph corresponding to each of the cumulative abnormality degree histograms created in S220 is created. Then, in S240, a causal relationship analysis is performed between the anomaly-occurring node and the nodes around the anomaly-occurring node using the anomaly transition graph created in S230. Then, when it is determined that there is a causal relationship between the abnormal occurrence node and the nodes around the abnormal occurrence node, the above-mentioned causal relationship information is stored in the causal relationship accumulation database 43, and the causal relationship extraction process is once performed. finish.

Next, the procedure of the abnormality estimation process executed by the control device 15 will be described. The abnormality estimation process corresponds to the abnormality estimation unit 34, and is a process that is repeatedly executed during the operation of the control device 15.
When this abnormality estimation process is executed, the CPU 21 of the control device 15 first determines in S310 whether or not there is an abnormality occurrence node based on the determination result of S140 in the abnormality detection process, as shown in FIG. to decide. Here, if there is no abnormality occurrence node, the abnormality estimation process is temporarily terminated.

On the other hand, when there is an abnormal occurrence node, in S320, causal relationship information indicating an abnormality of the abnormal transition node having a causal relationship with the abnormal occurrence node is extracted from the causal relationship accumulation database 43, and the abnormal transition node having a causal relationship is extracted. The number of abnormalities extracted is counted for each case.

Then, in S330, an anomaly transition map is created by mapping the anomaly transitions using the number of extracted anomalies having a causal relationship. Further, in S340, the created abnormality transition map is transmitted to the roadside machine 3 installed near the abnormality occurrence node and the roadside machine 3 installed near the node around the abnormality occurrence node, and the abnormality estimation process is performed. Is closed once.

Then, the roadside machine 3 that has received the abnormal transition map transmits the received abnormal transition map to the on-board unit 5 mounted on the vehicle traveling in the vicinity of the roadside machine 3. As a result, in the vehicle equipped with the on-board unit 5 that has received the abnormality transition map, as shown in FIG. 18, the navigation device mounted on the vehicle can provide route guidance avoiding the abnormality occurrence node. .. In FIG. 18, an abnormality is currently occurring in the vicinity of the node having the node ID of 50, the abnormality transitions in the vicinity of the node having the node ID 51 after about 15 minutes, and the vicinity of the node having the node ID 53. It shows that the abnormality changes after about 10 minutes. Further, FIG. 18 shows that the guide routes Rg1, Rg2, Rg3, and Rg4 from the current location to the destination are set while avoiding the vicinity of the nodes whose node IDs are 50, 51, and 53.

The abnormality detection device 7 configured in this way includes an operation status extraction unit 31, an abnormality detection unit 32, a causal relationship extraction unit 33, and an abnormality estimation unit 34.
The vehicle behavior data collection unit 51 of the driving situation extraction unit 31 repeatedly collects driving data, behavior data, and captured image data for each of a plurality of vehicles.

The driving scene division unit 52 and the topic ratio calculation unit 53 of the driving status extraction unit 31 calculate the topic ratio from the collected driving data and behavior data, and store the topic ratio and the node corresponding to the topic ratio in association with each other. ..

The abnormality detection unit 32 determines whether or not the abnormality occurrence node in which the abnormality has occurred exists at the present time, based on the degree of difference calculated using the calculated topic ratio. Since the node ID is associated with the topic ratio, the abnormality detection unit 32 compares the topic ratio calculated at the present time with the topic ratio calculated in the past for the same node as this topic ratio. Determine if the anomaly node currently exists.

When the abnormality detection unit 32 determines that the abnormality occurrence node exists at the present time, the abnormality detection unit 32 is associated with the received data associated with the abnormality occurrence node and the periphery of the abnormality occurrence node (hereinafter referred to as the abnormality occurrence node). Accumulated data is created using the received data, and the created accumulated data is accumulated.

The causal relationship extraction unit 33 uses the accumulated accumulated data to create causal relationship information indicating a causal relationship between the abnormality generated in the abnormality occurrence node and the abnormality generated in the nodes around the abnormality occurrence node. ..

The anomaly estimation unit 34 exists at the present time using the past causal relationship information created by the causal relationship extraction unit 33 when the anomaly detection unit 32 determines that the abnormality occurrence node exists at the present time. If there is, the abnormality transition from the abnormality occurrence node determined by the abnormality detection unit 32 to the periphery of the abnormality occurrence node is estimated.

As described above, the abnormality detection device 7 accumulates the accumulated data created by using the operation data and the behavior data of the abnormality occurrence node and the abnormality peripheral node, and creates the causal relationship information by using the accumulated accumulated data. And the causal relationship information shows the causal relationship between the abnormality that occurred in the abnormality occurrence node and the abnormality that occurred in the abnormality peripheral node. Therefore, when the abnormality detection device 7 determines that the abnormality occurrence node exists at the present time, the abnormality occurrence node (that is, the abnormality occurrence node) that transitions from the abnormality occurrence node as a starting point is set as the abnormality occurrence node. It can be estimated by extracting past causal relationship information showing a causal relationship with the abnormalities that have occurred. As a result, the abnormality detection device 7 can estimate the transition of the abnormality in a simple method and at high speed by accumulating the causal relationship information, and can identify the range of influence of the abnormality at high speed. The detour route can be calculated.

Further, the abnormality detection unit 32 determines that the node inside the peripheral setting circle Ca set in advance to include the abnormality occurrence node is the periphery of the abnormality occurrence node. That is, the abnormality detection device 7 determines that the abnormality generated in the abnormality occurrence node may transition even for the node that is not directly connected to the abnormality occurrence node by the road, and estimates the abnormality transition. As a result, the abnormality detection device 7 can estimate the transition of the abnormality in relation to the structure of the road.

Further, the abnormality estimation unit 34 extracts past causal relationship information of the abnormal transition node having a causal relationship with the abnormality occurrence node, and counts the number of extracted abnormalities extracted for each abnormal transition node having a causal relationship. As a result, the abnormality detection device 7 can estimate the transition of the abnormality so that the greater the number of extractions, the higher the possibility that the abnormality will transition from the abnormality occurrence node to the abnormality transition node. Then, since the high or low possibility of abnormal transition can be determined by a simple method of counting the number of extracted cases, the abnormality detecting device 7 can reduce the processing load of estimating the abnormal transition.

In the above-described embodiment, the abnormality detection device 7 corresponds to the abnormality estimation device, S10 corresponds to the processing as the collection unit, S20 and S30 correspond to the processing as the feature amount calculation unit, and S110 to S140 correspond to the abnormality. Corresponds to processing as a judgment unit.

Further, S150 and S160 correspond to the processing as the storage unit, S210 to S240 correspond to the processing as the information creation unit, and S310 to S330 correspond to the processing as the estimation unit.
Further, the driving data and the behavior data correspond to the vehicle data, the abnormality occurrence node corresponds to the abnormality occurrence point, the topic ratio corresponds to the feature amount, and the accumulated data corresponds to the estimation data.

Further, the current topic ratio corresponds to the current feature amount, the past topic ratio corresponds to the past feature amount, the peripheral setting circle Ca corresponds to the peripheral determination area, and the cumulative abnormal degree corresponds to the abnormal degree.
(Second Embodiment)
The second embodiment of the present disclosure will be described below together with the drawings. In the second embodiment, a part different from the first embodiment will be described. The same reference numerals are given to common configurations.

The abnormality detection system 1 of the second embodiment is different from the first embodiment in that the causal relationship extraction process and the abnormality estimation process are changed.
As shown in FIG. 19, the causal relationship extraction process of the second embodiment is different from the first embodiment in that the processes of S250 and S260 are added.

That is, when the processing of S240 is completed, the event information associated with the abnormal occurrence node of the abnormality for which the causal relationship analysis was performed in S240 and the abnormality occurrence date and time is acquired in S250 via the Internet or the like. After that, in S260, the event information acquired in S250 is stored in the causal relationship accumulation database 43 in a state of being associated with the causal relationship information stored in S240, and the causal relationship extraction process is temporarily terminated.

As shown in FIG. 20, the abnormality estimation process of the second embodiment is different from the first embodiment in that the process of S320 is omitted and the process of S312 and S322 is added.
That is, when there is an abnormality occurrence node in S310, the event information associated with the abnormality occurrence node determined in S310 and the abnormality occurrence date and time is acquired in S312 via the Internet or the like. Next, in S322, the abnormality of the abnormal transition node having a causal relationship with the abnormality occurrence node determined in S310 is shown, and the causal relationship information associated with the same event information as the event information acquired in S312 is accumulated. It is extracted from the database 43, and the number of abnormal extractions is counted for each abnormal transition node having a causal relationship. When the processing of S322 is completed, the process proceeds to S330.

In the abnormality detection device 7 configured in this way, the causal relationship extraction unit 33 acquires event information indicating the cause of the abnormality that has occurred in the abnormality occurrence node, and associates the acquired event information with the causal relationship information. Then, when the abnormality detection unit 32 determines that the abnormality occurrence node exists, the abnormality estimation unit 34 acquires the event information of the abnormality generated in the abnormality occurrence node, and obtains the same event information as the acquired event information. Anomalous transitions are estimated by extracting the associated past causal information.

As a result, the anomaly detection device 7 can extract causal relationship information by excluding anomalies that have not occurred due to the same event, and can improve the accuracy of estimating anomaly transitions.

In the embodiment described above, S210 to S260 correspond to the processing as the information creation unit, S310 to S330 correspond to the processing as the estimation unit, and the event information corresponds to the abnormality cause information.

(Third Embodiment)
The third embodiment of the present disclosure will be described below together with the drawings. In the third embodiment, a part different from the first embodiment will be described. The same reference numerals are given to common configurations.

The abnormality detection system 1 of the third embodiment is different from the first embodiment in that the causal relationship extraction process and the abnormality estimation process are changed.
As shown in FIG. 21, the causal relationship extraction process of the third embodiment is different from the first embodiment in that the processes of S255 and S265 are added.

That is, when the processing of S240 is completed, the type of abnormality in the abnormality occurrence node (for example, using the captured image data acquired at the abnormality occurrence date and time in the abnormality occurrence node of which the causal relationship analysis was performed in S240 in S240). Identify traffic jams or accidents). After that, in S265, the anomaly type information indicating the type of anomaly identified in S255 is stored in the causal relationship accumulation database 43 in a state of being associated with the causal relationship information stored in S240, and the causal relationship extraction process is temporarily terminated. ..

As shown in FIG. 22, the abnormality estimation process of the third embodiment is different from the first embodiment in that the process of S320 is omitted and the processes of S314 and S324 are added.
That is, when there is an abnormality occurrence node in S310, the type of abnormality in the abnormality occurrence node is identified in S314 by using the captured image data acquired at the abnormality occurrence date and time in the abnormality occurrence node determined in S310. Next, in S324, the causal relationship information associated with the abnormality type information indicating the abnormality of the abnormal transition node having a causal relationship with the abnormality occurrence node and the type of the abnormality identified in S314 is obtained from the causal relationship accumulation database 43. Extract and count the number of abnormal extractions for each abnormal transition node that has a causal relationship. When the processing of S324 is completed, the process proceeds to S330.

In the abnormality detection device 7 configured in this way, the causal relationship extraction unit 33 identifies and identifies the type of abnormality that has occurred in the abnormality occurrence node by using the captured image data acquired at the abnormality occurrence node at the abnormality occurrence date and time. The anomaly type information indicating the type of anomaly that has occurred is associated with the causal relationship information. Then, when the abnormality detection unit 32 determines that the abnormality occurrence node exists, the abnormality estimation unit 34 uses the captured image data to identify the type of abnormality that has occurred in the abnormality occurrence node, and the identified abnormality. The transition of the anomaly is estimated by extracting the past causal relationship information associated with the anomaly type information of the same type.

As a result, the anomaly detection device 7 can extract causal relationship information by excluding anomalies that have not occurred due to anomalies of the same type, and can improve the accuracy of estimating anomaly transitions. ..

In the embodiment described above, S210 to S265 correspond to the processing as the information creation unit, and S310 to S330 correspond to the processing as the estimation unit.
(Fourth Embodiment)
The fourth embodiment of the present disclosure will be described below together with the drawings. In the fourth embodiment, a part different from the first embodiment will be described. The same reference numerals are given to common configurations.

In the abnormality detection system 1 of the fourth embodiment, as shown in FIG. 26, the automobile includes a navigation device 9 connected so that data can be acquired from the vehicle-mounted device 5.
The navigation device 9 acquires map data from a map storage medium that records road map data and various information, and detects the current position of the own vehicle based on a GPS signal or the like received via a GPS antenna (not shown).

The navigation device 9 executes control for displaying the current location of the own vehicle on the display screen, control for guiding the route from the current location to the destination, and the like.
Then, when the navigation device 9 acquires the abnormality transition map created by the abnormality detection device 7 from the vehicle-mounted device 5, it displays "current abnormality" in the vicinity of the abnormality occurrence node as shown in FIG. 27, and displays the vicinity of the abnormality transition node. "Prediction error" is displayed. As a result, the navigation device 9 notifies the driver that an abnormality is predicted to occur at the abnormality transition node.

Further, when the navigation device 9 acquires the abnormality transition map created by the abnormality detection device 7 from the vehicle-mounted device 5, the navigation device 9 avoids the abnormality occurrence node and the abnormality transition node as shown in FIG. 27 when the route is being guided. Display the avoidance route. FIG. 27 shows that the guide routes Rg11, Rg12, Rg13, and Rg14 from the current location to the destination are set while avoiding the vicinity of the abnormality occurrence node and the abnormality transition node.

The navigation device 9 configured in this way is mounted on the vehicle, acquires an abnormality transition map showing the estimation result by the abnormality detection device 7, and displays information that can identify the position of the abnormality occurrence node and the abnormality transition node. ..

In this way, the navigation device 9 can make the driver recognize that the abnormality may occur at the abnormality transition node by displaying the point where the abnormality is predicted to occur to the driver in advance. Therefore, the navigation device 9 suppresses the driver from encountering an accident or encountering a state one step before the accident although it does not lead to the accident due to the abnormality at the abnormality transition node. be able to.

Further, the navigation device 9 displays an avoidance route for avoiding the abnormality occurrence node and the abnormality transition node. As a result, the navigation device 9 can prevent the driver from encountering an accident due to an abnormality at the abnormality occurrence node and the abnormality transition node. Further, the navigation device 9 can show the driver the basis for setting the route for avoiding the abnormal transition node by displaying the abnormality occurrence node, the abnormal transition node, and the avoidance route.

In the embodiment described above, the navigation device 9 corresponds to the display device, the abnormality transition map corresponds to the estimation information, the abnormality occurrence node corresponds to the current abnormality occurrence point, and the abnormality transition node corresponds to the current abnormality transition point. ..

Further, the position of the abnormal occurrence node and the position of the abnormal transition node shown in FIG. 27 correspond to the abnormal position identification information, the position of the abnormal occurrence node shown in FIG. 27 corresponds to the occurrence position information, and the abnormal transition node shown in FIG. 27. The position of corresponds to the transition position information.

Although one embodiment of the present disclosure has been described above, the present disclosure is not limited to the above embodiment, and can be implemented in various modifications.
[Modification 1]
For example, in the above embodiment, the topic ratio is used as the feature amount to be calculated to determine the occurrence of an abnormality, but the topic ratio is not limited, and the occurrence of an abnormality can be determined. All you need is.

[Modification 2]
In the above embodiment, as shown in FIG. 9, the inside of the peripheral setting circle Ca having the peripheral setting radius Ra centered on the abnormal occurrence node Pt is determined to be "the periphery of the abnormal occurrence node". However, as shown in FIG. 23, the map may be divided into rectangular grids, and the inside of the grid Gt including the anomaly occurrence node Pt may be determined to be "around the anomaly occurrence node".

[Modification 3]
In the above embodiment, a causal relationship analysis is performed using a Granger causal model. As a method for more easily determining the causal relationship, as shown in FIG. 24, the abnormality occurrence time calculated using the abnormality transition graph of the abnormality occurrence node and the abnormality transition of one or more nodes around the abnormality occurrence node. The causal relationship may be determined by comparing with the abnormality occurrence time calculated using the graph. That is, if an abnormality occurs in the vicinity of the abnormality occurrence node after the abnormality occurs in the abnormality occurrence node, it is determined that there is a causal relationship. FIG. 24 shows a state in which the abnormality occurrence time calculated using the abnormality transition graph Gr1 of the abnormality occurrence node and the abnormality occurrence time calculated using the abnormality transition graph Gr2 near the abnormality occurrence node are compared. ing.

Further, in order to determine the causal relationship in more detail, after determining the presence or absence of the causal relationship by the causal relationship analysis, the comparison based on the time of occurrence of the abnormality may be executed. That is, when it is determined by the causal relationship analysis that there is a causal relationship, and it is determined that an abnormality has occurred in the vicinity of the abnormal occurrence node after the abnormality has occurred in the abnormal occurrence node, the above-mentioned causal relationship information is used as the causal relationship. It may be stored in the storage database 43.

[Modification example 4]
In the above embodiment, the causal relationship information of the abnormality determined to have a causal relationship with the abnormality generated in the abnormality occurrence node is given the same abnormality identification number as the abnormality identification number stored in the road abnormality accumulation database 42. The state stored in the causal relationship accumulation database 43 is shown. However, the identification number for determining the similarity abnormality may be assigned by calculating the similarity degree of the abnormality based on the topic ratio. As a result, in the abnormality estimation, the time required to extract the information indicating the similar abnormality from the causal relationship accumulation database 43 can be shortened, and the transition of the abnormality can be estimated by handling only the similar abnormality. Therefore, the estimation accuracy can be improved.

[Modification 5]
In the above embodiment, data having a causal relationship with the abnormality generated in the abnormality occurrence node is extracted, and the transition relationship of the abnormality is mapped according to the magnitude of the number of extractions. In this case, in addition to the anomaly occurrence node, the degree of similarity with at least one of the anomaly degree (for example, cumulative anomaly degree) and the anomaly feature amount (for example, topic ratio) is included, and the data having a causal relationship is extracted. You may. As a result, it is possible to exclude and extract anomalies that are not similar to the anomalies that have occurred in the anomaly-occurring node, and it is possible to improve the accuracy of estimating the anomaly transition. In addition, restrictions may be set such as setting an extraction restriction depending on the time zone or preventing extraction of data that is too old. Further, since the abnormality transition is managed by the abnormality identification number, if at least one causally related abnormality can be extracted, the related abnormality can be easily extracted by the abnormality identification number.

[Modification 6]
In the above embodiment, as shown in FIG. 13, an abnormality transition map is created so that the corresponding arrow becomes thicker as the number of extraction cases increases. However, as shown in FIG. 25, instead of the number of extraction cases, the abnormal transition time required for the abnormality to transition from the abnormal occurrence node to the abnormal transition node is shown, and the shorter the abnormal transition time, the thicker the corresponding arrow. Anomalous transition maps may be created in. In FIG. 25, the arrow TR3 starting from the node with the node ID of 50 and ending at the node with the node ID of 51 indicates an abnormal transition from the node with the node ID of 50. Then, "20 min" described near the arrow TR3 indicates the abnormality transition time required for the abnormality to transition from the node with the node ID of 50 to the node with the node ID of 51.

Similarly, the arrow TR4 indicates an abnormal transition from a node with a node ID of 50 to a node with a node ID of 53. Then, "10 min" described near the arrow TR4 indicates the abnormality transition time required for the abnormality to transition from the node with the node ID of 50 to the node with the node ID of 53.

In addition, the degree of causal relationship may be shown instead of the number of extraction cases and the abnormal transition time. The degree of causal relationship indicates the degree of causal relationship between the abnormality of the abnormal occurrence node and the abnormality of the abnormal transition node, and is calculated by the causal relationship analysis. The abnormal transition time and the degree of causal relationship may be the average of the data of a plurality of vehicles, the representative data, or the data of all the plurality of vehicles.

Further, instead of the abnormal transition time and the degree of causality, a weighted average of the abnormal transition time and the degree of causality may be shown.
[Modification 7]
In the above embodiment, as shown in FIG. 18, information for predicting the time when the abnormality occurs, such as "after about 10 minutes" and "after about 15 minutes", is displayed, but the abnormality disappears. Information for predicting the time may be displayed.

[Modification 8]
In the fourth embodiment, as shown in FIG. 27, the navigation device 9 displays a road represented by a node and a link. However, as shown in FIG. 28, the navigation device 9 may represent the road only by the link, and may display only the abnormality occurrence node and the abnormality transition node as the node. Alternatively, the navigation device 9 may superimpose and display the abnormality occurrence node and the abnormality transition node on a general road map.

[Modification 9]
In the fourth embodiment, as shown in FIG. 27, the navigation device 9 displays the abnormality occurrence node and the abnormality transition node, but the time may be displayed with respect to the abnormality transition node. Examples of the time to be displayed include an abnormality occurrence predicted time that predicts the time when the abnormality occurs, an abnormality occurrence time zone that predicts the time when the abnormality occurs, an abnormal end predicted time that predicts the time when the abnormality ends, and the like.

In this way, the navigation device 9 can further make the driver recognize the time when the abnormality occurs by displaying the occurrence time identification information that can specify the time when the abnormality occurs in the abnormality transition node. This allows the driver to make a route plan to avoid the abnormal transition node with a margin.

[Modification 10]
In the fourth embodiment, as shown in FIG. 27, the navigation device 9 displays the abnormality occurrence node and the abnormality transition node. However, as shown in FIG. 29, the navigation device 9 may display the arrow as moving from the abnormality occurrence node to the abnormality transition node. In FIG. 29, the navigation device 9 first displays the arrow A1 in order to indicate the movement to the abnormal transition node located at the lower left of the abnormality occurrence node. Next, the navigation device 9 displays the arrow A2 after turning off the display of the arrow A1. Further, the navigation device 9 displays the arrow A3 after turning off the display of the arrow A2. In this way, the navigation device 9 expresses the transition of the abnormality by sequentially displaying the arrows A1, the arrows A2, and the arrows A3. Then, the navigation device 9 expresses the speed at which the abnormality transitions by the transition time from the display of the arrow A1 to the display of the arrow A3. For example, when the transition time is short, it indicates that the time when the abnormality occurs in the abnormal transition node arrives earlier, and when the transition time is long, the time when the abnormality occurs in the abnormal transition node arrives later. It is good to show that. In FIG. 29, the navigation device 9 displays the abnormal transition node and the abnormal occurrence node in a distinguishable manner. For example, the navigation device 9 displays the abnormal transition node by blinking or the like. Further, in FIG. 29, the navigation device 9 first displays an arrow A4 in order to indicate the movement to the abnormal transition node located at the lower right of the abnormality occurrence node. Next, the navigation device 9 displays the arrow A5 after turning off the display of the arrow A4.

In this way, the navigation device 9 allows the driver to intuitively grasp the time when the abnormality occurs at the abnormality transition node by displaying the arrow as if it moves by animation.

[Modification 11]
In the fourth embodiment, as shown in FIG. 27, the navigation device 9 displays the abnormality occurrence node and the abnormality transition node. However, as shown in FIG. 30, the navigation device 9 may display the area including the abnormality occurrence node and the abnormality transition node as an abnormality. In FIG. 30, the navigation device 9 displays the circular region Rt1 including the abnormality occurrence node and the abnormality transition node as an abnormality.

In this way, the navigation device 9 displays the abnormal region information indicating the region Rt1 including the position of the abnormal occurrence node and the position of the abnormal transition node. As a result, the navigation device 9 can broadly indicate the abnormality occurrence point instead of pinpointing it. Therefore, it is possible to prevent the driver from approaching abnormally. Further, by displaying in this way, the navigation device 9 can make the driver consider the possibility of abnormality even at a point that has not been determined to be abnormal in the past. For example, when flooding occurs at the abnormal occurrence node and the abnormal transition node due to heavy rain, it is desirable to keep away from the abnormal occurrence node and the abnormal transition node as much as possible instead of avoiding the flooding pinpointly. By displaying the navigation device 9 as shown in FIG. 30, the driver can avoid the abnormality with a margin. The region Rt1 shown in FIG. 30 corresponds to the abnormal position identification information and the abnormal region information.

The area including the abnormality occurrence node and the abnormality transition node is not limited to a circle, and may be a shape including the abnormality occurrence node and the abnormality transition node. For example, it may have a quadrangular shape, or may have a shape that surrounds the anomaly occurrence node and the anomaly transition node along the road.

[Modification 12]
In the above modification 11, the navigation device 9 displays the circular region Rt1 including the abnormality occurrence node and the abnormality transition node as an abnormality. However, as shown in FIG. 31, the navigation device 9 may display the area including only the abnormality occurrence node among the abnormality occurrence node and the abnormality transition node so as to be distinguishable from the area including the abnormality transition node. In FIG. 31, the navigation device 9 displays the circular area Rt1 including the abnormality occurrence node and the abnormality transition node as caution information, and displays the circular area Rt2 including the abnormality occurrence node as an alarm. In FIG. 31, the region Rt1 and the region Rt2 are displayed in different colors. In FIG. 31, the navigation device 9 is painted in two colors. However, in the region Rt1, the periphery of the point where the probability of occurrence of an abnormality is high may be painted with a gradation like a contour line, for example, by making the color darker.

In this way, the navigation device 9 can distinguish between the area Rt2 including the position of the abnormality occurrence node and the area Rt1 including the position of the abnormality transition node. As a result, the navigation device 9 can make the driver recognize the area including the position of the abnormality occurrence node and the area including the position of the abnormality transition node separately. Therefore, the driver can make a route plan to avoid the abnormal transition node in consideration of the position of the abnormal transition node and the position of the abnormal transition node. The region Rt1 and the region Rt2 shown in FIG. 31 correspond to the abnormal position identification information. Further, the region Rt2 shown in FIG. 31 corresponds to the abnormality occurrence region, and the region Rt2 corresponds to the abnormality transition region.

[Modification 13]
In the fourth embodiment, as shown in FIG. 27, the navigation device 9 displays the abnormality occurrence node and the abnormality transition node. However, as shown in FIG. 32, the navigation device 9 may change the size of the icon indicating the abnormal transition node according to the abnormal transition probability. In FIG. 32, the navigation device 9 displays such that the higher the probability of abnormal transition, the larger the icon indicating the abnormal transition node. The abnormal transition probability is calculated so that the larger the number of extractions, the larger the number of extractions, based on the number of extractions in which the abnormal transition node is extracted as an abnormality having a causal relationship with the node in which the abnormality occurs. The icon indicating the abnormal transition node is not limited to a circle.

Further, as shown in FIG. 33, the navigation device 9 may change the color of the icon indicating the abnormal transition node according to the abnormal transition probability. In FIG. 33, the navigation device 9 displays the icon of the abnormal transition node having a high abnormal transition probability in red, and the icon of the abnormal transition node having a low abnormal transition probability in blue.

In this way, the navigation device 9 further highlights the abnormal transition node having a high abnormal transition probability. As a result, the navigation device 9 can make the driver recognize the abnormal transition node having a high probability of abnormal transition. Therefore, the driver can make a route plan to avoid the abnormal transition node in consideration of the abnormal transition probability.

[Modification 14]
In the fourth embodiment, as shown in FIG. 27, the navigation device 9 displays the abnormality occurrence node and the abnormality transition node. However, as shown in FIG. 34, the navigation device 9 displays only one or more abnormal transition nodes that the vehicle may reach based on the current position of the vehicle. May be good. For example, the navigation device 9 may not display the abnormal transition node that is predicted to have an abnormal transition after 30 minutes if it cannot be reached within 30 minutes from the current position.

In this way, the navigation device 9 further displays only one or more abnormal transition nodes that the vehicle equipped with the navigation device 9 may reach. As a result, the navigation device 9 can prevent the driver from unnecessarily displaying unnecessary information.

[Modification 15]
In the fourth embodiment, as shown in FIG. 27, the navigation device 9 expresses and displays the node as a branch point of the road, but the node is set to the branch point of the road. Not limited. As shown in FIG. 35, nodes may be finely arranged in the link at intervals of, for example, 1 m, and the abnormality detection device 7 may determine an abnormality occurrence and an abnormality transition for each of these plurality of nodes. As a result, the abnormality detection device 7 can more accurately determine the occurrence of an abnormality and the abnormality transition on the road.

[Modification 16]
In the fourth embodiment, as shown in FIG. 27, the navigation device 9 displays the abnormality occurrence node and the abnormality transition node. However, the display device that displays the abnormality occurrence node and the abnormality transition node is not limited to the navigation device, and may be, for example, a smartphone on which a map application or the like is installed. The map application does not have to have a navigation function.

Further, the function of one component in the above embodiment may be shared by a plurality of components, or the function of the plurality of components may be exerted by one component. Further, a part of the configuration of the above embodiment may be omitted. Further, at least a part of the configuration of the above embodiment may be added or replaced with the configuration of the other embodiment. It should be noted that all aspects included in the technical idea specified from the wording described in the claims are embodiments of the present disclosure.

In addition to the above-mentioned abnormality detection device 7, various forms such as a system including the abnormality detection device 7 as a component, a program for operating a computer as the abnormality detection device 7, a medium on which this program is recorded, an abnormality estimation method, and the like. This disclosure can also be realized.

7 ... Abnormality detection device, 31 ... Driving situation extraction unit, 32 ... Abnormality detection unit, 33 ... Causality extraction unit, 34 ... Abnormality estimation unit, 51 ... Vehicle behavior data collection unit, 52 ... Driving scene division unit, 53 ... Topic Ratio calculation unit

Claims (14)

A collection unit (S10) configured to repeatedly collect vehicle data related to the state of the vehicle for each of the plurality of vehicles.
A feature amount calculation unit (S20, S30) configured to calculate a feature amount from the vehicle data collected by the collection unit and store the feature amount in association with a location corresponding to the feature amount. When,
An abnormality determination unit (S110 to S140) configured to determine whether or not an abnormality occurrence point where an abnormality has occurred currently exists based on the feature amount calculated by the feature amount calculation unit. )When,
When the abnormality determination unit determines that the abnormality occurrence point exists at the present time, the vehicle data at the abnormality occurrence point and the vehicle data at the abnormality peripheral point which is a point around the abnormality occurrence point. Is used to create preset estimation data for estimating the transition of the abnormality from the abnormality occurrence point to the abnormality peripheral point, and a storage unit configured to accumulate the created estimation data. (S150, S160) and
Using the estimation data accumulated by the storage unit, it is configured to create causal relationship information showing a causal relationship between the abnormality generated at the abnormality occurrence point and the abnormality generated at the abnormality peripheral point. With the information creation unit (S210-S265)
When the abnormality determination unit determines that the abnormality occurrence point exists at the present time, the abnormality determination is determined to exist at the present time by using the past causal relationship information created by the information creation unit. Bei example configured estimator as part estimates the transition abnormal to the abnormal peripheral point from the abnormality generation point is determined and (S310~S330),
The abnormality determination unit uses the feature amount currently calculated by the feature amount calculation unit as the current feature amount, and the feature amount calculated in the past as compared with the current feature amount as the past feature amount, and uses the current feature amount as the past feature amount. An abnormality estimation device (7) for determining whether or not the abnormality occurrence point currently exists by comparing the amount with the past feature amount. A collection unit (S10) configured to repeatedly collect vehicle data related to the state of the vehicle for each of the plurality of vehicles.
A feature amount calculation unit (S20, S30) configured to calculate a feature amount from the vehicle data collected by the collection unit and store the feature amount in association with a location corresponding to the feature amount. When,
An abnormality determination unit (S110 to S140) configured to determine whether or not an abnormality occurrence point where an abnormality has occurred currently exists based on the feature amount calculated by the feature amount calculation unit. )When,
When the abnormality determination unit determines that the abnormality occurrence point exists at the present time, the vehicle data at the abnormality occurrence point and the vehicle data at the abnormality peripheral point which is a point around the abnormality occurrence point. Is used to create preset estimation data for estimating the transition of the abnormality from the abnormality occurrence point to the abnormality peripheral point, and a storage unit configured to accumulate the created estimation data. (S150, S160) and
Using the estimation data accumulated by the storage unit, it is configured to create causal relationship information showing a causal relationship between the abnormality generated at the abnormality occurrence point and the abnormality generated at the abnormality peripheral point. With the information creation unit (S210-S265)
When the abnormality determination unit determines that the abnormality occurrence point exists at the present time, the abnormality determination is determined to exist at the present time by using the past causal relationship information created by the information creation unit. With the estimation unit (S310-S330) configured to estimate the transition of the abnormality from the abnormality occurrence point to the abnormality peripheral point determined by the unit.
With
The estimation unit extracts past causal relationship information of the abnormality peripheral point having a causal relationship with the abnormality generated at the abnormality occurrence point, and acquires information indicating the number of extracted abnormalities. .. A collection unit (S10) configured to repeatedly collect vehicle data related to the state of the vehicle for each of the plurality of vehicles.
A feature amount calculation unit (S20, S30) configured to calculate a feature amount from the vehicle data collected by the collection unit and store the feature amount in association with a location corresponding to the feature amount. When,
An abnormality determination unit (S110 to S140) configured to determine whether or not an abnormality occurrence point where an abnormality has occurred currently exists based on the feature amount calculated by the feature amount calculation unit. )When,
When the abnormality determination unit determines that the abnormality occurrence point exists at the present time, the vehicle data at the abnormality occurrence point and the vehicle data at the abnormality peripheral point which is a point around the abnormality occurrence point. Is used to create preset estimation data for estimating the transition of the abnormality from the abnormality occurrence point to the abnormality peripheral point, and a storage unit configured to accumulate the created estimation data. (S150, S160) and
Using the estimation data accumulated by the storage unit, it is configured to create causal relationship information showing a causal relationship between the abnormality generated at the abnormality occurrence point and the abnormality generated at the abnormality peripheral point. With the information creation unit (S210-S265)
When the abnormality determination unit determines that the abnormality occurrence point exists at the present time, the abnormality determination is determined to exist at the present time by using the past causal relationship information created by the information creation unit. With the estimation unit (S310-S330) configured to estimate the transition of the abnormality from the abnormality occurrence point to the abnormality peripheral point determined by the unit.
With
The information creation unit creates the causal relationship information so as to include at least one of the degree of abnormality indicating the degree of abnormality occurring at the abnormality occurrence point and the feature amount when the abnormality occurs at the abnormality occurrence point. death,
The estimation unit uses at least one of the abnormality degree and the feature amount included in the past causal relationship information to determine that the abnormality occurrence point is abnormal that the abnormality determination unit determines that it exists at the present time. An abnormality estimation device that estimates the transition of an abnormality from the abnormality occurrence point to the abnormality peripheral point by extracting the causal relationship information of similar abnormalities. A collection unit (S10) configured to repeatedly collect vehicle data related to the state of the vehicle for each of the plurality of vehicles.
A feature amount calculation unit (S20, S30) configured to calculate a feature amount from the vehicle data collected by the collection unit and store the feature amount in association with a location corresponding to the feature amount. When,
An abnormality determination unit (S110 to S140) configured to determine whether or not an abnormality occurrence point where an abnormality has occurred currently exists based on the feature amount calculated by the feature amount calculation unit. )When,
When the abnormality determination unit determines that the abnormality occurrence point exists at the present time, the vehicle data at the abnormality occurrence point and the vehicle data at the abnormality peripheral point which is a point around the abnormality occurrence point. Is used to create preset estimation data for estimating the transition of the abnormality from the abnormality occurrence point to the abnormality peripheral point, and a storage unit configured to accumulate the created estimation data. (S150, S160) and
Using the estimation data accumulated by the storage unit, it is configured to create causal relationship information showing a causal relationship between the abnormality generated at the abnormality occurrence point and the abnormality generated at the abnormality peripheral point. With the information creation unit (S210-S265)
When the abnormality determination unit determines that the abnormality occurrence point exists at the present time, the abnormality determination is determined to exist at the present time by using the past causal relationship information created by the information creation unit. With the estimation unit (S310-S330) configured to estimate the transition of the abnormality from the abnormality occurrence point to the abnormality peripheral point determined by the unit.
With
The information creation unit (S210 to S260) is configured to acquire abnormality cause information indicating the cause of the abnormality that occurred at the abnormality occurrence point, and to associate the acquired abnormality cause information with the created causal relationship information. Being done
When the abnormality determination unit determines that the abnormality occurrence point exists, the estimation unit acquires the abnormality cause information of the abnormality generated at the abnormality occurrence point, and is the same as the acquired abnormality cause information. An abnormality estimation device that estimates the transition of an abnormality from the abnormality occurrence point to the abnormality peripheral point by extracting the past causal relationship information associated with the abnormality cause information. The abnormality estimation device according to any one of claims 1 to 4.
The accumulation unit is an abnormality estimation device that determines that a point in a peripheral determination area preset to include the abnormality occurrence point is around the abnormality occurrence point. The abnormality estimation device according to any one of claims 1 to 5.
The data collected by the collecting unit includes photographed image data obtained by a camera mounted on the vehicle taking a picture of the periphery of the vehicle.
The information creation unit (S210 to S265) uses the captured image data to identify the type of abnormality that has occurred at the abnormality occurrence point, and the abnormality type information indicating the type of the identified abnormality and the created causal relationship. Configured to associate with information
When the abnormality determination unit determines that the abnormality occurrence point exists, the estimation unit identifies the type of abnormality that has occurred at the abnormality occurrence point using the captured image data, and the identified abnormality. An abnormality estimation device that estimates the transition of an abnormality from the abnormality occurrence point to the abnormality peripheral point by extracting the past causal relationship information associated with the abnormality type information of the same type. A display device (9) mounted on a vehicle to acquire estimation information indicating an estimation result by the estimation unit of the abnormality estimation device according to any one of claims 1 to 6.
The abnormality occurrence point determined by the abnormality determination unit to exist at the present time is set as the current abnormality occurrence point, and the abnormality peripheral point estimated by the estimation unit as the abnormality transition from the current abnormality occurrence point is the current abnormality transition point. A display device that displays abnormal position identification information capable of specifying the positions of the current abnormality occurrence point and the current abnormality transition point. The display device according to claim 7.
The abnormality position identification information is a display device which is generation position information indicating the position of the current abnormality occurrence point and transition position information indicating the position of the current abnormality transition point. The display device according to claim 8.
Further, a display device that displays occurrence time identification information capable of identifying the time when an abnormality occurs at the current abnormality transition point. The display device according to claim 8 or 9.
Further, a display device that highlights the current abnormal transition point having a high probability of abnormal transition. The display device according to any one of claims 8 to 10.
Further, a display device that displays the current abnormality occurrence point and the avoidance route for avoiding the current abnormality transition point. The display device according to any one of claims 8 to 11.
Further, among the one or more current abnormal transition points, a display device that displays only the current abnormal transition points that the vehicle equipped with the display device may reach. The display device according to claim 7.
The abnormal position identification information is a display device which is abnormal area information indicating an abnormal area including the position of the current abnormality occurrence point and the position of the current abnormal transition point. The display device according to claim 13.
The abnormal region includes an abnormal occurrence region including the position of the current abnormal occurrence point and an abnormal transition region including the position of the current abnormal transition point.
A display device that distinguishably displays the abnormality occurrence area and the abnormality transition area.

Images