JP2014238654A - Operation assist device and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide operation assistance so that the type of danger can be estimated without the need for a danger information input person to designate the type of danger and a danger indicated by the type of danger can be avoided.SOLUTION: An operation assist device 10 includes: a danger type dictionary 16 in which each of values indicating the degree of applicability to which each word applies to each danger type are stored in correlation of each of a plurality of words with each of a plurality of danger types; a danger information input unit 12 for accepting, as its input, danger information including position information and explanatory information; a danger type estimation unit 18 for extracting a word from the explanatory information included in the danger information, and estimating, on the basis of a value indicating the degree of applicability stored in the danger type dictionary 16 in correlation of a word equivalent to the extracted word and stored in the danger type dictionary 16 with each of the plurality of danger types, a danger type at a place indicated by the position information included in the danger information; and an assist unit 28 for providing a driver with operation assistance so that the danger indicated by the estimated danger type at the place where the danger type is estimated can be avoided.

Description

  The present invention relates to a driving support device and a program, and more particularly to a driving support device and a program that provide driving support based on danger information.

  There are various devices that provide driving assistance.

  For example, a traffic risk diagnosis system for diagnosing a traffic risk by specifying a traffic accident occurrence day in various conditions and situations to provide to a company or an individual is known (for example, Patent Document 1).

  There is also known an apparatus that automatically collects danger information related to road traffic, provides effective information to drivers and pedestrians using the information, and prevents accidents from occurring (for example, Patent Document 2). . For example, risk information (near hat information and accident information) related to road traffic is collected and stored from the communication device, and the risk information is distributed in response to a request from the communication device. When the danger information is received, an alarm can be issued or the danger area can be avoided when passing through the danger area. The near-miss information includes "Danger types (steep steering wheel, sudden braking, sudden acceleration, slip, overspeed, etc.)", "Dangerous position" and "Danger time", etc. Specified by the provider.

  In addition, the traffic information detection means installed on the road side detects and collects driving state information and accident occurrence information including traffic conditions of vehicles traveling on the road, and the collected information and preset road shape, etc. Based on the above, we will provide information services to vehicles traveling on the road, obtain information on the occurrence of accidents on the road, analyze the cause of the accident, and implement traffic control such as the construction of new signs and the addition of a new service for preventing unforeseen risks A road traffic control system that determines and outputs the content of improvement proposals for equipment is also known (for example, Patent Document 3).

JP 2002-298280 A JP 2003-123185 A JP 2008-135070 A

  With the technology described in Patent Document 1, it can be recognized that the risk of a traffic accident increases on a specific day, but what type of risk increases, that is, the risk type cannot be recognized.

  In the technique described in Patent Document 2, although the type of danger can be recognized, the danger type must be specified by the person who entered the danger information. Therefore, the type of danger can be selected according to the input person. There may be a difference.

  Further, in the technique described in Patent Document 3, it is necessary to provide traffic information detection means on the road side, and infrastructure maintenance is required, and there are limits to the causes of accidents that can be estimated.

  The present invention has been made to solve the above problem, and it is possible for a person who inputs danger information to estimate the risk type without specifying the risk type, and to assist the driving so that the risk indicated by the estimated risk type is avoided. An object of the present invention is to provide a driving support device and a program that can be used.

  In order to achieve the above object, the driving support device of the present invention relates to each of a plurality of types of words and each of a plurality of types of danger types, and a value indicating the degree to which each word corresponds to each risk type. A risk type dictionary each storing a risk information, input means for inputting risk information including location information and explanation information regarding the danger of the point indicated by the location information, and extracting words from the explanation information included in the danger information The risk is based on a value indicating the corresponding degree stored in the risk type dictionary in association with the word stored in the risk type dictionary corresponding to the selected word and each of the plurality of risk types. The estimation means for estimating the risk type of the point indicated by the position information included in the information, and the driver's driving support so that the risk indicated by the estimated risk type is avoided at the point where the risk type is estimated. Including, and support means for performing.

  In the case where there is one word extracted from the explanation information, the estimation means is a value indicating the degree of correspondence stored in association with the word stored in the risk type dictionary corresponding to the extracted word. Is estimated as the risk type of the point indicated by the position information, and when there are a plurality of words extracted from the explanation information, they are stored in the risk type dictionary corresponding to each of the extracted words. The risk type having the maximum sum for each risk type of the value indicating the degree of correspondence stored in association with each of the words may be estimated as the risk type of the point indicated by the position information.

  In addition, the driving support device extracts a document including a specific keyword predetermined for each risk type from a document set relating to traffic danger as a document corresponding to the risk type for which the specific keyword is determined, For each word included in the extracted document, a construction means for constructing the risk type dictionary by calculating a value indicating a degree corresponding to the risk type may be further included.

  In addition, the driving support device may be configured such that a part of documents in the document set relating to traffic danger is classified into the plurality of types of dangers in advance, and the already classified document is removed from the document set. , Obtaining a similarity with the already classified document, classifying each of the remaining documents into any of the plurality of risk types based on the obtained similarity, and each word included in the classified document Further, a construction means for constructing the risk type dictionary by calculating a value indicating a degree corresponding to the risk type to which the document is classified may be further included.

  Note that the value indicating the corresponding degree in the risk type dictionary may be a tfidf value.

  In addition, the estimation unit estimates the risk type for each risk information input by the input unit, and the support unit determines the risk type or the number of estimations for each point is equal to or greater than a predetermined threshold. The driving assistance of the driver may be performed so that the danger indicated by the danger type with the largest estimated number of times is avoided.

  In addition, the information processing apparatus further includes a determination unit that determines whether or not the danger information whose risk type is estimated is reliable information, and the support unit includes the risk type estimated based on the risk information determined to be reliable by the determination unit. The driver may be provided with driving assistance so as to avoid the danger indicated by.

  Moreover, the program of this invention is a program for functioning a computer as an estimation means and a support means in the said driving assistance apparatus. Moreover, it is good also as a program for functioning a computer as an estimation means in the said driving assistance apparatus, a support means, and a construction means. Moreover, it is good also as a program for functioning a computer as an estimation means, support means, construction means, and determination means in the driving support apparatus.

  As described above, according to the present invention, the risk type can be estimated without the risk type input person specifying the risk type, and driving assistance can be provided so that the risk indicated by the estimated risk type is avoided. The effect is obtained.

It is the figure which showed the functional structure of the driving assistance device which concerns on embodiment. It is the figure which showed a mode that dangerous information was input and transmitted with an audio | voice etc. from the onboard equipment which has a communication function. It is the figure which showed a mode that dangerous information was input and transmitted as text information from a portable terminal or a personal computer. It is a figure which shows an example of a danger classification dictionary. It is a figure which shows an example of a danger classification memory | storage part. It is a flowchart which shows an example of the danger classification dictionary construction process which a danger classification dictionary construction part performs. It is explanatory drawing explaining the extraction method of a document. It is a flowchart which shows the other example of the danger classification dictionary construction process which a danger classification dictionary construction part performs. It is a flowchart which shows a document classification | category process. It is a schematic diagram which shows typically the flow of a document classification | category process. It is a flowchart which shows an example of the danger classification estimation process performed by the danger classification estimation part. It is a flowchart which shows an example of the danger classification presentation determination process which a presentation determination part performs. It is a figure which shows the example of presentation of driving assistance information. It is the figure which showed the other example of the functional structure of the driving assistance device.

  Hereinafter, the driving support apparatus 10 according to the present embodiment will be described in detail with reference to the drawings. The driving support device 10 collects danger information and supports driving based on the danger information.

  FIG. 1 is a diagram illustrating an example of a functional configuration of the driving support device 10 according to the present embodiment. The driving support device 10 includes a risk information input unit 12, a risk type dictionary construction unit 14, a risk type dictionary 16, a risk type estimation unit 18, a risk type storage unit 20, and a support unit 28. In the present embodiment, it is assumed that the driving support device 10 is mounted on a vehicle. As another example, at least the support unit 28 is mounted on the vehicle among the elements configuring the driving support device 10, the other components are mounted on an external server or the like, and the support unit 28 is connected to the external server or the like. Necessary information may be exchanged by communication.

  The danger information input unit 12 operates as an interface for communicating with an external device, for example. Various information is input (received) from the external device to the danger information input unit 12. In the present embodiment, information related to traffic danger (hereinafter referred to as danger information) is input from an in-vehicle device having a communication function, a portable terminal, and a personal computer. The danger information includes at least the location information of the dangerous point, such as latitude and longitude, and explanatory information explaining the danger of the dangerous point.

  When the danger information is transmitted from the vehicle-mounted device having the communication function, the danger information can be input and transmitted by voice or the like as shown in FIG. When transmitting danger information from a portable terminal or a personal computer, it can be input and transmitted as text information as shown in FIG. 3, using a keyboard etc., for example. In addition to the position information and the description information, “photographed image of dangerous point” may be included in the dangerous information (details will be described later). The danger information input unit 12 receives the danger information input and transmitted in this way.

  The risk information input to the risk information input unit 12 is output to the risk type dictionary construction unit 14 and the risk type estimation unit 18.

  The risk type dictionary construction unit 14 constructs a risk type dictionary 16 from a document set that is a set of documents related to traffic danger. This document set includes texts of accident cases (texts describing the circumstances under which traffic accidents have occurred), such as newspaper articles relating to traffic accidents and documents of accident investigation agencies. Further, the document set may include a document of danger information input to the danger information input unit 12. A specific example of the construction method of the risk type dictionary will be described later.

  The risk type dictionary 16 is constructed by the risk type dictionary construction unit 14 and is referred to by the risk type estimation unit 18. FIG. 4 shows an example of the danger type dictionary 16. The risk type dictionary 16 is a dictionary that stores a value indicating the degree to which each word corresponds to each risk type in association with each of the plurality of types of words and each of the plurality of types of risk types.

  Examples of the multiple types of danger include jumping out, poor visibility, passing, rear-end collision, traffic jam, many illegal parking, high speed, and heavy traffic. The words registered in the danger type dictionary 16 are the original forms of all the words included in the document set. Note that the registered words are not limited to the words illustrated in FIG. 4, and various words such as elementary school students, curves, and lane changes are stored.

  In the present embodiment, a tfidf value is used as a value indicating the degree of matching. The formula for calculating the tfidf value is shown below.

As shown in the calculation formula (1), the tfidf value is a score calculated based on two indices, tf (word appearance frequency) and idf (reverse document frequency). Here, n _{i, j} indicates how many times the word i appears in the document j (number of occurrences of the word i). Σ _k n _{k, j} indicates the number of occurrences of all words in the document j. idf is a word that appears frequently in many documents (general words such as “ga” and “do”), and its value (importance) decreases, and the word appears only in a specific document. The value goes up. Therefore, the tfidf value is higher for words that appear characteristically for each risk type, whereas for words that appear frequently, the idf value is small (close to zero), so tfidf is also small ( Will be close to zero).

  The risk information is input from the risk information input unit 12 to the risk type estimation unit 18. The risk type estimation unit 18 extracts a word from the description information included in the input risk information, and uses the risk type dictionary 16 to indicate the point indicated by the position information included in the input risk information from the extracted word. Estimate the risk type. The risk type estimation unit 18 stores the estimated risk type in the risk type storage unit 20 in association with the position information.

  FIG. 5 shows an example of the risk type storage unit 20. As shown in FIG. 5, in this embodiment, the estimated number of times (simply indicated as “number of times” in FIG. 5) is stored in association with the point ID indicating the point and the risk type. The risk type estimation unit 18 estimates the risk type for the input risk information every time the risk information is input. However, the risk information estimated at each point is not always the same. Therefore, the risk type estimation unit 18 stores the estimated number of times for each risk type in the risk type storage unit 20 for each point. Each time the risk type is estimated by the risk type estimation unit 18, the number of times corresponding to the estimated risk type increases by one.

  The support unit 28 includes a position information input unit 22, a presentation determination unit 24, and a presentation unit 26.

  Position information indicating the travel position of the host vehicle is input to the position information input unit 22. For example, the position information may be detected by a GPS sensor. The GPS sensor can receive radio waves from a plurality of satellites and use the radio waves to measure the position of the host vehicle.

  Position information is input to the presentation determination unit 24 from the position information input unit 22. The presentation determination unit 24 determines whether or not to present the driving support information to the vehicle driver so that the danger indicated by the danger type is avoided at the point indicated by the input position information. For example, if a risk type having an estimated number of at least one or more corresponding to the point is stored in the risk type storage unit 20, the risk indicated by each of the risk types having the estimated number of at least one or more is avoided. You may determine to show all driving assistance information. Further, it is determined whether there is a risk type whose estimated number is equal to or greater than a threshold, and determination is made so that only driving support information for avoiding the danger indicated by the risk type whose estimated number is equal to or greater than the threshold is presented. May be. In addition, driving support information for avoiding the danger indicated by the danger type with the largest estimated number of times may be presented.

  The presenting unit 26 presents driving support information to the driver so that the risk indicated by the risk type determined to be presented by the presentation determining unit 24 is avoided. For example, the presentation unit 26 may include, for example, a device that emits sound or voice information, may include a device that displays an image or emits light, or a device that vibrates the driver. May be included. That is, the presentation unit 26 presents driving support information to the driver of the vehicle by stimuli such as sound, sound information, image, light, and vibration. In addition, the presentation unit 26 may present information indicating the risk type as the driving support information, or may present instruction information effective for avoiding the risk.

  In addition, each function which comprises the driving assistance apparatus 10 may be implement | achieved by the computer. For example, the computer includes a central processing unit (CPU), a read only memory (ROM), and a random access memory (RAM), and includes a danger information input unit 12, a risk type dictionary construction unit 14, and a risk type estimation. The unit 18 and the support unit 28 may be realized by the CPU executing a program stored in the ROM.

  The storage medium for storing the program is not limited to the ROM, and may be a recording medium such as an HDD (hard disk device), a CD-ROM, or a flash memory.

  Next, an example of the driving support process of this embodiment will be described.

First, a specific example of the construction method of the risk type dictionary 16 will be described.
FIG. 6 is a flowchart illustrating an example of the risk type dictionary construction process performed by the risk type dictionary construction unit 14.

  In step 100, the risk type dictionary construction unit 14 extracts a document for each risk type using a keyword from a document set relating to traffic danger. Keywords are preset for each danger type. Specifically, as illustrated in FIG. 7, a document including the keyword “jumping out and running” is extracted as a document of the risk type “jumping out”, and a document including the keyword “rear collision” is extracted from the risk type “rear collision”. Extract as a document. A document that does not contain any keyword is extracted as a document of danger type “others”.

  In step 102, the risk type dictionary construction unit 14 calculates tfidf values for all words included in the extracted document. The calculation formula of tfidf is as described above, but the document j in the calculation formula (1) of tfidf indicates a document set for each risk type such as “jump out”, “rejection”, and tf is a word set in each document set. It refers to the appearance frequency, and the “total number of documents” in the numerator of idf refers to the number of documents included in each document set.

  In step 104, as shown in FIG. 4, the risk type dictionary construction unit 14 registers the calculated tfidf value for each word and each risk type.

  Another example of the risk type dictionary construction process will be described below.

  FIG. 8 is a flowchart showing another example of the risk type dictionary construction process performed by the risk type dictionary construction unit 14. This process is started in a state in which some documents in the document set relating to traffic danger are classified in advance into one of a plurality of danger types. More specifically, for example, it is performed after manually extracting a small number of document sets for each risk type.

  In step 200, the danger type dictionary construction unit 14 performs document classification processing.

  FIG. 9 is a flowchart showing the document classification process in step 200. FIG. 10 is a schematic diagram schematically showing the flow of document classification processing.

  In step 300, the risk type dictionary construction unit 14, as shown in (1) and (2) of FIG. 10, from the document set, documents already classified for each risk type (hereinafter, a set of these documents is used as learning data). A single document is selected from the remaining documents excluding ()) as a target document, and the similarity with learning data of all risk types is calculated. The cosine similarity is used for the similarity calculation.

  In step 302, the risk type dictionary construction unit 14 determines for each risk type whether the similarity is equal to or greater than a threshold value. If the similarity is equal to or higher than the threshold, the risk type dictionary construction unit 14 adds the target document as learning data for the risk type whose similarity is equal to or higher than the threshold in step 304. If the degree of similarity between the target document and the learning data of the danger type “jump out” is equal to or higher than the darkness value, the document is added as the “jump out” learning data. On the other hand, if there is no risk type whose similarity is equal to or greater than the threshold, step 304 is skipped.

  In step 306, the risk type dictionary construction unit 14 determines whether there is a determination target document. If all the documents in the document set have already been calculated and determined, the determination is affirmative at step 306 and the determination at step 306 is negative otherwise. If the determination in step 306 is affirmative, the document extraction process ends. In this way, the documents included in the document set are classified for each risk type (see also FIG. 10 (3)).

  When the document extraction process is completed, the risk type dictionary construction unit 14 returns to the process of FIG. 8, and in step 202, the risk type dictionary construction unit 14 performs tfidf values for all words included in the classified documents. Is calculated. The tfidf calculation method is the same as that described in step 102 in FIG.

  In step 204, the risk type dictionary construction unit 14 registers the calculated tfidf value for each risk type and each word, as shown in FIG.

  The risk type dictionary construction unit 14 may update (reconstruct) the risk type dictionary 16 periodically or at a timing when a predetermined condition is satisfied. For example, when the danger information acquired by the danger information input unit 12 is also used by the risk type dictionary construction unit 14 as a part of a document set relating to traffic danger, for example, a predetermined number of danger information is acquired. The risk type dictionary 16 may be updated every time the risk type dictionary construction process is performed.

Next, the risk type estimation process performed by the risk type estimation unit 18 will be described in detail.
FIG. 11 is a flowchart illustrating an example of the risk type estimation process performed by the risk type estimation unit 18.

  In step 400, the risk type estimation unit 18 acquires risk information from the risk information input unit 12.

  In step 402, the risk type estimation unit 18 extracts words from the description information from the description information included in the acquired risk information. Here, the explanation information is extracted by morphological analysis and converted into the original form of the word. For example, if the explanatory information is a text document that says “It is dangerous to use a pedestrian bridge, even though it is a pedestrian pedestrian crossing, it is dangerous.” It is troublesome to use the pedestrian bridge, so it is dangerous because pedestrians cross the road.

In step 404, the risk type estimation unit 18 refers to the risk type dictionary 16 and calculates an evaluation value for each risk type from the extracted tfidf value for each word. An example of calculating the evaluation value Score _Tc of the risk type Tc is shown below.

Here, tfidf _word indicates the tfidf value of each word of danger type Tc. In the present embodiment, the sum of tfidf values of each word is calculated for each risk type and used as an evaluation value.

  In step 406, the risk type estimation unit 18 selects the risk type with the highest evaluation value as the risk type of the point indicated by the position information included in the acquired risk information, and stores it in the risk type storage unit 20. Specifically, the number of times of the selected risk type of the point ID corresponding to the point corresponding to the position information included in the risk information is increased by one (see also FIG. 5).

For example, as a result of calculating the evaluation value of each risk type for the explanation information “Dangerous because pedestrians who use the pedestrian bridge cross the road are dangerous because they cross the pedestrian crossing, it is dangerous.” When the evaluation value is reached, the risk type is “jump out (precautions)” and the risk type is “dangerous because pedestrians crossing the pedestrian crossing despite the bicycle-only pedestrian crossing are dangerous.” Choose as.
・ Jump out (note pop out) = 17.3
・ Prospect (poor prospect) = 8.4
・ Passing (passing attention) = 5.6

  It should be noted that all risk types for which the evaluation value is equal to or greater than a preset threshold value may be selected and stored in the risk type storage unit 20. For example, when the threshold value is 7.0, it is also possible to select and store a risk type of “jumping out (caution popping out)” and a risk type of “outlook (poor outlook)”. That is, in the risk type storage unit 20, the estimated number of the selected two risk types is increased by one.

  In step 402, when there is one extracted word, the risk type may be estimated from the magnitude of the tfidf value itself instead of obtaining the evaluation value (total tfidf value). Specifically, a risk type having a maximum tfidf value or a preset threshold value or more may be selected as the risk type of the point.

  Next, the presentation determination process performed by the presentation determination unit 24 will be described in detail.

  FIG. 12 is a flowchart illustrating an example of the risk type presentation determination process performed by the presentation determination unit 24.

  In step 500, the presentation determination unit 24 acquires position information from the position information input unit 22.

  In step 502, the presentation determination unit 24 stores the risk type information (here, information indicating the estimated number of times for each risk type) stored in correspondence with the point indicated by the position information acquired from the risk type storage unit 20. read out.

  In step 504, the presentation determination unit 24 determines whether or not there is a risk type whose estimated number (hereinafter simply referred to as “number of times”) is greater than or equal to a threshold value. If the presentation determination unit 24 determines in step 504 that there is a risk type whose number of times is equal to or greater than a preset threshold value, in step 506, the presentation determination unit 24 avoids the risk indicated by the risk type whose number of times is equal to or greater than the threshold value. The presentation unit 26 is controlled so that the driving support information is presented. If the presentation determination unit 24 determines in step 504 that there is no risk type whose number of times is equal to or greater than a preset threshold, step 506 is skipped.

  FIG. 13 shows an example of presentation of driving support information. In the example shown in Fig. 13, when the vehicle approaches a point where the danger type is "Caution for jumping out", it will be superimposed on the screen of the vehicle-mounted device with "Caution for popping out", and "Please be careful about jumping out 50m ahead" Guidance in

  The presentation determination unit 24 may further present driving support information according to the traveling state of the host vehicle. For example, in step 504, when it is determined that there is a risk type with the number of times equal to or greater than the threshold, and when the host vehicle is traveling, driving support information is presented and it is determined that there is a risk type with the number of times equal to or greater than the threshold. When the host vehicle is stopped, the driving support information may not be presented.

  In addition, here, an example in which driving support information for a risk type whose estimated number of times is equal to or greater than a threshold has been described, but driving support information for a risk type whose estimated number of times is at least one or more may be presented. The driving support information for the risk type with the largest estimated number of times may be presented.

  Here, an example in which driving assistance information is displayed or presented by voice as driving assistance has been described. However, the present invention is not limited to this, and for example, the vehicle may be controlled as driving assistance. . Specifically, examples include increasing the resolution of image processing for pedestrian recognition at an intersection where the risk type is “jumping out” and decreasing the vehicle speed at a point where the risk type is “speed”. The determination of the presentation of the risk type may be used for route guidance. Specifically, a large vehicle is not guided to a point where the danger type is “passing”.

  Further, it is determined whether or not the risk information input to the risk information input unit 12 and whose risk type is estimated by the risk type estimation unit 18 is reliable, and only the risk type estimated based on the risk information determined to be reliable can be determined. You may make it memorize | store in the memory | storage part 20. FIG.

  FIG. 14 shows a driving support device (indicated by reference numeral 11) provided with a function for judging the reliability of the danger information with respect to the driving support device 10 shown in FIG. In FIG. 14, parts that are the same as or equivalent to those in FIG.

  14 includes a danger information input unit 12, a risk type dictionary construction unit 14, a risk type dictionary 16, a risk type estimation unit 18A, a risk information determination unit 19, a risk type storage unit 20, and a support unit 28. It has.

  The risk type estimation unit 18A estimates the risk type in the same manner as the risk type estimation unit 18 of the driving support device 10, but the information on the estimated risk type is not directly stored in the risk type storage unit 20, Together with the danger information input from the danger information input unit 12, it is output to the danger information determination unit 19. The estimation method of the risk type estimation unit 18A is the same as that of the risk type estimation unit 18 of the driving support device 10, but in this example, the output destination of the information of the estimated risk type is the risk information determination unit 19. A symbol A is appended to the end to distinguish it from the risk type estimation unit 18.

  The danger information judgment unit 19 judges the reliability of the danger information for which the danger type is estimated. The risk information determination unit 19 stores the risk type of the risk information determined to be reliable in the risk type storage unit 20.

  For each estimated risk type, the reliability determination method by the risk information determination unit 19 may be different.

  For example, with respect to the danger information in which the danger type of “jump out” or “rear collision” is estimated, the frequency of sudden braking at the point indicated by the position information included in the danger information (for example, the frequency of braking of 0.3 G or more), The frequency (average) of sudden braking at intersections across the country may be compared, and when it is determined that the sudden braking occurs frequently, the risk type may be stored in the risk type storage unit 20 as reliable information. For example, the frequency of sudden braking at each point can be acquired by a system that acquires the vehicle speed, acceleration, and the like of the traveling vehicle at each point.

  Further, with respect to the risk information in which the risk type of “bad visibility” or “passing” is estimated, the reliability can be determined by acquiring curve and road width information from the map information.

  In addition, the risk information for which the risk type of “congestion” is estimated is determined as reliable information when it is determined that the traffic congestion is a frequent occurrence point from the past traffic situation of the point indicated by the position information included in the risk information, The risk type can be stored in the risk type storage unit 20.

  Further, the risk information for which the risk type of “there are many illegal parkings” is determined as reliable information when the number of illegal parking cleared is equal to or greater than the threshold, and the risk type is stored in the risk type storage unit 20. Can be remembered.

  For the danger information whose risk type is “speed”, the average speed at the point indicated by the position information included in the danger information is acquired and compared with a threshold value, or the legal speed and the average speed are It is also possible to take a technique such as taking the difference and supporting it.

  In addition, when the danger information in which the danger type of “the amount of traffic” is estimated includes a photographed image of the point indicated by the position information included in the danger information, the photographed image is disclosed in Japanese Patent No. 4097519 The degree of complexity of the photographed image is calculated by the above technique, etc., and when it is determined to be complicated, it is determined as reliable information, and the risk type is stored in the risk type storage unit 20.

  Thus, more appropriate driving assistance can be performed by determining reliability and storing it in the risk type storage unit 20.

  Although the example in which the risk information determination unit 19 stores the estimation result in the risk type storage unit 20 has been described here, the risk information determination unit 19 inputs the reliability determination result to the risk type estimation unit 18A, and The risk type estimation unit 18A may store the risk type information in the risk type storage unit 20 based on the reliability determination result.

DESCRIPTION OF SYMBOLS 10, 11 Driving support device 12 Danger information input part 14 Danger type dictionary construction part 16 Danger type dictionary 18, 18A Danger type estimation part 19 Danger information judgment part 20 Danger type memory | storage part 22 Position information input part 24 Presentation determination part 26 Presentation part 28 Support Department

Claims (10)

A risk type dictionary that stores values indicating the degree to which each word corresponds to each risk type, corresponding to each of the plurality of types of words and each of the plurality of types of risk types,
Input means for inputting danger information including position information and explanatory information related to the danger of the point indicated by the position information;
A word is extracted from the description information included in the danger information, and the risk type dictionary is associated with the word stored in the danger type dictionary corresponding to the extracted word and each of the plurality of risk types. Estimating means for estimating the risk type of the point indicated by the position information included in the danger information based on the stored value indicating the degree of correspondence;
Support means for supporting driving of the driver so that the risk indicated by the estimated risk type is avoided at the point where the risk type is estimated;
A driving support device including When there is one word extracted from the explanation information, the estimation means has a maximum value indicating the degree of correspondence stored in association with the word stored in the danger type dictionary corresponding to the extracted word. Alternatively, a risk type equal to or higher than a threshold is estimated as the risk type of the point indicated by the position information, and when there are a plurality of words extracted from the explanation information, they are stored in the risk type dictionary corresponding to each of the extracted words. The risk type with the sum total for each risk type of the value indicating the degree of correspondence stored corresponding to each of the recorded words is estimated as the risk type of the point indicated by the position information is a maximum or higher risk type. The driving assistance apparatus as described. A document including a specific keyword predetermined for each danger type is extracted as a document corresponding to the risk type for which the specific keyword is determined from the document set relating to traffic danger, and is included in the extracted document. The driving support device according to claim 1 or 2, further comprising: a construction unit that constructs the risk type dictionary by calculating a value indicating a degree corresponding to the risk type for each word. In a state where a part of documents in the document set relating to traffic hazards is classified in advance into the plurality of types of hazards, the remaining documents obtained by removing the already classified documents from the document set, and the already classified documents And classifying each of the remaining documents into one of the plurality of risk types based on the obtained similarity, and the document is classified for each word included in the classified document. The driving support device according to claim 1, further comprising: a construction unit that calculates a value indicating a degree corresponding to the risk type and constructs the risk type dictionary. The driving support device according to any one of claims 1 to 4, wherein the value indicating the corresponding degree in the danger type dictionary is a tfidf value. The estimation means estimates the danger type for each danger information input by the input means,
The driver assists the driver so that the risk indicated by the risk type for which the estimated number of times for each point is equal to or greater than a predetermined threshold or the risk type for which the estimated number of times is the highest is avoided. The driving support apparatus according to claim 5. A judgment means for judging whether or not the danger information whose risk type is estimated is reliable information;
The driver assists the driver so that the danger indicated by the danger type estimated from the danger information determined to be reliable by the judging means is avoided. The driving support device according to item. Computer
The program for functioning as an estimation means and a support means in the driving assistance apparatus of any one of Claims 1-7. Computer
The program for functioning as an estimation means in the driving assistance apparatus of any one of Claims 3-7, a support means, and a construction means. Computer
The program for functioning as an estimation means in the driving assistance device of Claim 7, a support means, a construction means, and a judgment means.

Images