JP2019096081A - Unsafe driving predicting device - Google Patents

Abstract

To provide an unsafe driving predicting device which can predict unsafe driving.SOLUTION: A driving predicting ECU 100 includes: a behavior-related data acquisition unit 101 for successively acquiring behavior-related data, in association with a vehicle position, that indicates at least one of a vehicle behavior and a travel environment; a symbol transition prediction unit 105 for predicting future behavior-related data on the vehicle from the behavior-related data up to the present time; a transition sequence generation unit 108 for generating a current symbol sequence that represents the transition of a plurality of behavior-related data and includes past behavior-related data and future behavior-related data; a symbol transition pattern acquisition unit 107 for acquiring an unsafe driving time behavior-related data transition sequence at a position corresponding to the current symbol sequence from an unsafe driving database in which unsafe driving time behavior-related data transition sequences are stored in association with positions; and a near-miss driving prediction unit 109 for determining unsafe driving probability on the basis of similarity of the current symbol sequence to the unsafe driving time behavior-related data transition sequences.SELECTED DRAWING: Figure 4

Description

  The present invention relates to an unsafe driving prediction device for predicting unsafe driving.

  In Patent Document 1, the driver's past behavior history is compared with the current behavior of the driver to determine whether the current behavior is in an unsafe state.

JP, 2016-110449, A

  In Patent Document 1, it is determined whether the current state is an unsafe state. However, if it can be predicted in advance that it will become unsafe driving, it will be possible to take effective measures in safe driving rather than measures that can be taken after unsafe driving is performed.

  The present disclosure has been made based on the above circumstances, and an object of the present disclosure is to provide an unsafe driving prediction apparatus capable of predicting unsafe driving.

  The above object is achieved by a combination of the features of the independent claims, and the subclaims define further advantageous embodiments. The reference numerals in parentheses described in the claims indicate the correspondence with specific means described in the embodiments described later as one aspect, and do not limit the disclosed technical scope.

One disclosure for achieving the above object is
An unsafe driving prediction device for predicting that driving of a vehicle will be unsafe driving,
A behavior related data acquisition unit (101) for sequentially acquiring behavior related data, which is data representing at least one of the behavior of the vehicle and the traveling environment around the vehicle, in association with the position of the vehicle;
A behavior related data prediction unit (105) for predicting future behavior related data of a vehicle from past behavior related data obtained up to the present time obtained by the behavior related data acquisition unit being acquired sequentially;
A vehicle including a behavior related data transition sequence representing transitions of a plurality of behavior related data, the past behavior related data acquired by the behavior related data acquisition unit, and future behavior related data predicted by the behavior related data prediction unit A transition sequence generation unit (108) that generates current behavior related data transition sequences of
The unsafe operation of the position corresponding to the current behavior related data transition sequence generated by the transition sequence generation unit from the unsafe operation database storing the behavior related data transition sequence during unsafe operation in association with the position. A transition sequence acquisition unit (107) for acquiring a behavior related data transition sequence;
An unsafe driving prediction unit that determines an unsafe driving probability that is a probability that the driving of the vehicle becomes unsafe driving based on the similarity between the current behavior related data transition string and the behavior related data transition string during unsafe driving And (109).

  This unsafe driving prediction apparatus has a probability of unsafe driving of the vehicle based on the similarity between the current behavior related data transition sequence and the behavior related data transition sequence during unsafe driving. Determine the probability.

  The current behavior related data transition sequence also includes future behavior related data. Therefore, future unsafe operation can be predicted based on the similarity between the current behavior related data transition sequence and the behavior related data transition sequence during unsafe operation.

  In addition, current behavior related data transition sequences include not only future behavior related data but also past behavior related data. Therefore, compared with the case where unsafe driving is predicted only with future behavior related data, the prediction accuracy of unsafe driving is higher.

  In addition, unsafe driving differs depending on the position where it is likely to occur. Therefore, this unsafe driving prediction apparatus stores behavior related data transition sequences during unsafe driving in association with positions, and behavior related during unsafe driving of the position corresponding to the current behavior related data transition sequence. The similarity between the data transition sequence and the current behavior related data transition sequence is calculated. This also increases the prediction accuracy of unsafe driving.

FIG. 1 is an entire configuration diagram of a near-miss operation prediction system 1; FIG. 6 is a diagram for explaining a near-miss operation database 5; It is a figure which shows the correspondence of the symbol in a symbol transition pattern PTN1, and a position. It is a block diagram which shows the function with which driving | running | working prediction ECU100 of FIG. 1 is provided. It is a figure which illustrates the future symbol which the symbol transition prediction part 105 of FIG. 4 estimated. It is a flowchart which shows the process which the near-miss operation prediction part 109 performs in 2nd Embodiment. It is a figure explaining _total distance _dtotal and reach distance d _r . It is a flowchart which shows the process which the near-miss operation prediction part 109 performs in 3rd Embodiment.

First Embodiment
[overall structure]
Hereinafter, embodiments will be described based on the drawings. The whole block diagram of the near-miss operation prediction system 1 is shown in FIG. In addition, near-miss driving means unsafe driving. The near-miss driving prediction system 1 includes an on-vehicle system 10 mounted on the vehicle 2 and a server 3.

  The in-vehicle system 10 is mounted on a number of vehicles 2. The vehicle 2 is not particularly limited as long as it can travel on the road. The vehicle 2 includes four-wheeled vehicles, motorcycles, bicycles and the like.

[Configuration of server 3]
The server 3 is disposed in a data center or the like outside the vehicle. The server 3 includes a communication unit 4, a near-miss operation database 5, and a control unit 6. The communication unit 4 wirelessly directly or indirectly communicates with the wireless communication unit 12 included in the in-vehicle system 10 to mutually transmit and receive signals. Examples of indirect communication include communication via a base station provided in a public communication network, communication via a roadside apparatus, and the like.

  The near-miss driving database 5 is a database storing behavior-related data transition sequences of the vehicle 2 when the near-miss driving has been performed in the past. The near-miss driving means driving that did not lead to an accident, but it is not strange that an accident will occur. In the case where the vehicle is inadvertently forward, it is likely to collide with an obstacle, and a rapid deceleration is performed, this is an example of a near-miss operation.

  The behavior related data is data related to the behavior of the vehicle 2 and, in addition to the vehicle behavior data which is data in the vehicle in which the behavior of the vehicle 2 changes, traveling environment data representing the traveling environment around the vehicle 2 is behavior related Included in the data. This is because the behavior of the vehicle 2 may change depending on the surrounding driving environment. The behavior related data transition sequence is a transition sequence of symbolized behavior related data, and in the present embodiment, the behavior related data transition sequence is stored as a symbol transition pattern. The near-miss operation database 5 will be described with reference to FIG.

  In the near-miss operation mode 5 of the present embodiment, a symbol transition pattern is stored in association with a near-miss operation position and a route to the near-miss operation position. The near-miss driving database 5 corresponds to the unsafe driving database, and the near-hit driving position corresponds to the unsafe driving position. The symbol transition pattern is one in which multiple transitions of behavior-related data up to the near-miss driving position are symbolized, that is, a symbol string, and the symbol string is a patterned or typified symbol. . In FIG. 2, positions A and B are shown as near-miss operation positions, symbol transition patterns PTN1 to PTN4 are shown for position A, and symbol transition patterns PTN5 and PTN6 for position B. .

  The symbol in the present embodiment is a combination of an alphabet of one letter and a numeral of one letter. The alphabet indicates traveling conditions such as acceleration and slowing, and the numbers indicate the degree of the traveling conditions.

  The number of symbols included in the symbol transition pattern can be set variously. For example, it can be a fixed distance to the near-miss driving position or the number of symbols obtained in a fixed time. Alternatively, if the symbol transition pattern when the near-miss operation is performed and the symbol transition pattern when the near-miss operation is not performed can be compared, and the number of symbols of the portion where the both differ can be determined, the difference Let the number of symbols of the portion to be processed be the number of symbols of the symbol transition pattern.

  FIG. 3 shows transitions of symbols and positions by taking the symbol transition pattern PTN1 as an example. The symbol transition pattern PTN1 is a symbol transition pattern PTN in the case where behavior-related data that become the symbols c2, a3, b1, d4, and e2 at the positions P1, P2, P3, P4, and A, respectively, are acquired.

  The control unit 6 is connected to the communication unit 4 and the near-miss operation database 5. The control unit 6 is a computer including a CPU, a ROM, a RAM, an I / O, and a bus line connecting these components. The ROM stores a program for causing a general purpose computer to function as the control unit 6.

  The control unit 6 acquires part or all of the symbol transition patterns stored in the near-miss driving database 5 and transmits the acquired symbol transition patterns from the communication unit 4 to the on-vehicle system 10.

[Configuration of in-vehicle system 10]
The in-vehicle system 10 includes a satellite positioning unit 11, a wireless communication unit 12, a periphery monitoring sensor 13, a driving support ECU 14, an in-vehicle LAN 15, and a driving prediction ECU 100. The driving prediction ECU 100 is an unsafe driving prediction device. In the following description, regarding the elements other than the driving prediction ECU 100, the contents related to the driving prediction ECU 100 will be mainly described.

  The satellite positioning unit 11 receives a navigation signal transmitted by a navigation satellite included in a Global Navigation Satellite System (GNSS) that is a satellite navigation system, and sequentially calculates the current position based on the received navigation signal. The satellite positioning unit 11 outputs the calculated current position to the driving prediction ECU 100.

  The wireless communication unit 12 performs wireless communication with the communication unit 4 included in the server 3. The wireless communication unit 12 can also perform inter-vehicle communication by short-distance wireless communication. The periphery monitoring sensor 13 is a sensor that detects an object present around the vehicle 2. For example, the periphery monitoring sensor 13 is an on-vehicle camera that captures the periphery of the vehicle 2. Also, in addition to the on-vehicle camera or in place of the on-vehicle camera, any one or more of radar, LIDAR, and sonar that transmits radio waves, light and sound as transmitted waves to the surroundings of the vehicle 2 and detects reflected waves It can also be used as the perimeter monitoring sensor 13.

  The driving support ECU 14 executes various support control related to the driving of the vehicle 2. Specific contents of the driving support control are, for example, notification control, braking control, and the like. The driving support ECU 14 is provided with the probability of becoming a near-miss operation from the driving prediction ECU 100 (hereinafter referred to as near-miss operation probability). The driving support ECU 14 performs driving support that is determined based on the near-miss driving probability.

  A driving prediction ECU 100 is connected to the in-vehicle LAN 15, and various other ECUs or sensors mounted on the vehicle 2 are connected thereto. The driving prediction ECU 100 acquires various vehicle behavior data via the in-vehicle LAN 15. The vehicle behavior data is data indicating an amount affecting the behavior of the vehicle 2 and includes, for example, data indicating a vehicle speed, a steering angle, an accelerator opening degree, and a brake pressure.

  The operation prediction ECU 100 is a computer provided with a CPU, a ROM, a RAM, an I / O, a bus line connecting these components, and the like. The ROM stores a program for causing a general purpose computer to function as the operation prediction ECU 100.

  In the control unit 6 included in the operation prediction ECU 100 and the server 3, the CPU executes a program stored in the ROM while using the temporary storage function of the RAM. When the CPU executes a program, a method corresponding to the program is executed. Note that the storage medium for storing the program executed by the CPU is not limited to the ROM, and may be stored in a non-transitory tangible storage medium. For example, the program may be stored in a flash memory. In addition, part or all of the functions of the driving prediction ECU 100 and the control unit 6 may be realized using one or more ICs (in other words, as hardware). In addition, part or all of the functions of the operation prediction ECU 100 and the control unit 6 may be realized by a combination of software execution by the CPU and hardware members.

  The driving prediction ECU 100 predicts the near-miss driving probability and outputs the near-miss driving probability to the driving support ECU 14. The operation prediction ECU 100 realizes the functions of the units 101 to 109 shown in FIG. 4 by the CPU executing a program stored in the ROM or the like.

  As shown in FIG. 4, the driving prediction ECU 100 includes, as functions executed by the CPU, a behavior related data acquisition unit 101, a symbolization unit 104, a symbol transition prediction unit 105, a travel route calculation unit 106, and a symbol transition pattern acquisition unit 107. A transition sequence generation unit 108 and a near-miss operation prediction unit 109 are provided. Further, the storage unit 120 is provided. The storage unit 120 is a writable storage unit, and can use a RAM.

  The behavior related data acquisition unit 101 includes a vehicle behavior data acquisition unit 102 and a traveling environment data acquisition unit 103. The vehicle behavior data acquisition unit 102 acquires the various vehicle behavior data described above via the in-vehicle LAN 15, and sequentially stores the acquired vehicle behavior data in the storage unit 120 together with the current position at the time of acquisition.

  The traveling environment data acquisition unit 103 acquires traveling environment data, and sequentially stores the acquired traveling environment data in the storage unit 120. The traveling environment data includes data acquired by the surrounding area monitoring sensor 13. In addition to this, the weather and the road surface condition may be acquired as traveling environment data. In addition, data such as the position, the traveling direction, and the speed of the moving object present around the vehicle 2 may be acquired as traveling environment data. The data of the mobile present around the vehicle 2 is directly or indirectly acquired from the communication device used by the mobile via the wireless communication unit 12.

  The symbolization unit 104 divides the behavior related data sequentially stored in the storage unit 120 according to the running state of the vehicle 2 serving as a division unit, and the running state of the vehicle 2 grasped from each behavior related data classified A symbol string is generated by representing by the symbol corresponding to. That is, in the multi-dimensional space represented by each data constituting the behavior related data, the encoding unit 104 sets each of the traveling states of the vehicle 2 grasped from the behavior related data as a cluster, and performs the behavior at each predetermined timing. Statistically process which cluster the related data belongs to. As a result, the behavior related data is divided for each traveling state (that is, for each cluster) of the vehicle 2 serving as the division unit, and a corresponding symbol is given to each divided portion.

  Each symbol shown in FIG. 3 is a symbol given in this manner, and each symbol is associated with the position at which the behavior related data represented by the symbol is obtained. The predetermined timing is, for example, each time the vehicle travels a fixed distance or every fixed time. The symbol generated by the symbolization unit 104 is stored in the storage unit 120.

  The in-vehicle system 10 sequentially transmits to the server 3 a symbol string generated in this way, which is the final symbol of the position where it is determined from the vehicle behavior data that the near-miss driving has occurred. The near-miss operation database 5 of the server 3 is generated by performing typification based on the symbol string transmitted from the in-vehicle system 10.

  The symbol transition prediction unit 105, the travel route calculation unit 106, the symbol transition pattern acquisition unit 107, the transition sequence generation unit 108, and the near-miss operation prediction unit 109, which will be described later, have a cycle in which symbols are newly generated in the symbolization unit 104. Execute the process

  The symbol transition prediction unit 105 predicts the transition of symbols representing future behavior related data of the vehicle 2 from the past symbol strings stored in the storage unit 120 up to the present time. The symbol transition prediction unit 105 corresponds to a behavior related data prediction unit.

  The current point means the latest point stored in the storage unit 120. Various methods are possible as methods for predicting future symbol transitions. For example, n-gram can be used. In addition, a symbol transition database storing symbol transition rules may be provided, and future symbols may be predicted according to the symbol transition rules stored in the symbol transition database.

  A future transition of a plurality of symbols may be predicted by generating a driving scene composed of a symbol string including a plurality of symbols and predicting a future driving scene. Furthermore, one driving scene may be represented as one symbol. The driving scene is to segment behavior-related data obtained in time series at the switching point of the driving scene felt by the driver. As a specific segmentation method, for example, there is a method using a dual segmentation analyzer that segments by the unsupervised driving scene segmentation method using a dual segmentation structure.

  FIG. 5 exemplifies future symbols predicted by the symbol transition prediction unit 105. In FIG. 5, the symbols enclosed by solid lines are the symbols up to the present time that the encoding unit 104 has encoded. On the other hand, symbols d1 and e4 surrounded by broken lines are future symbols predicted by the symbol transition prediction unit 105.

  The travel route calculation unit 106 sequentially acquires the current position from the satellite positioning unit 11 and calculates a future travel route of the vehicle 2. In this calculation, when the route guidance is being executed, and the guidance route can be acquired from the route guidance device, the route after the current position in the guidance route is taken as the future travel route. On the other hand, when the guidance route can not be acquired, a future traveling route is predicted based on the current position which can be sequentially acquired from the satellite positioning unit 11 and the traveling history up to the current position. For example, assuming that the road on which the vehicle is traveling continues in the future, the route on which the road on which the vehicle is traveling is advanced in the current traveling direction is taken as the future travel route. Alternatively, if the travel history generated in the current travel matches a part of the past travel history, the travel route may be determined based on the assumption that the vehicle travels along the past travel history this time as well. Good. The travel route calculation unit 106 provides the symbol transition pattern acquisition unit 107 with a future travel route.

  The symbol transition pattern acquisition unit 107 acquires the symbol transition pattern transmitted from the server 3 and received by the wireless communication unit 12 and stores the acquired symbol transition pattern in the storage unit 120. Further, when the symbol transition pattern on the future travel route calculated by the travel route calculation unit 106 is not stored in the storage unit 120, the server 3 is requested and the symbol transition pattern on the future travel route is The acquired symbol transition pattern is stored in the storage unit 120. As a result, at least a part of the symbol transition patterns stored in the near-miss operation database 5 included in the server 3 is stored in the storage unit 120, so the storage unit 120 can also be regarded as the near-miss operation database. The storage unit 120 also stores the symbol transition pattern in association with the near-miss driving position and the route to the near-miss driving position.

  The symbol transition pattern acquisition unit 107 further performs symbol transition corresponding to the position of the current symbol string generated by the transition sequence generation unit 108 described below on the symbol transition pattern in the near-miss operation stored in the storage unit 120. Search if there is a pattern. As a result of the search, when there is a symbol transition pattern in the near-miss operation where the current symbol string and the final position are the same, the symbol transition pattern in the near-miss operation is acquired from the storage unit 120 and provided to the near-miss operation prediction unit 109. . Although the symbol transition pattern is typed, since it is a symbol string, the symbol transition pattern acquisition unit 107 corresponds to a transition sequence acquisition unit.

  The transition string generation unit 108 generates a symbol string including the future symbol predicted by the symbol transition prediction unit 105 and the past symbol string used for the prediction. Hereinafter, this symbol string is taken as the current symbol string. The five symbols shown in FIG. 5 are examples of the current symbol string. The transition string generation unit 108 refers to the symbol transition pattern stored in the storage unit 120 and generates the current symbol string so as to have the same length as the symbol transition pattern.

  If the current symbol string is provided from the transition string generation unit 108 and the symbol transition pattern for the near-miss operation is provided from the symbol transition pattern acquisition unit 107, the near-miss operation prediction unit 109 receives the current symbol string and the near-miss operation. The similarity S with the symbol transition pattern is calculated. Further, the near-miss operation prediction unit 109 determines the near-miss operation probability from the similarity S. The near-miss operation probability represents the probability of performing near-miss operation, that is, unsafe operation. The near-miss driving probability corresponds to the unsafe driving probability, and the near-hit driving prediction unit 109 corresponds to the unsafe driving prediction unit.

  For example, cosine similarity is calculated as similarity S. The closer the current symbol string is to the symbol transition pattern during near-miss operation, the closer the cosine similarity is to one. In the present embodiment, a table in which the correspondence between the similarity S and the near-missing driving probability is defined is stored, and after the similarity S is calculated, the near-missing operating probability is calculated from the calculated similarity S and the above table. decide. The near-miss driving prediction unit 109 provides the determined near-miss driving probability to the driving support ECU 14 together with the position corresponding to the probability. The position corresponding to the near-miss operation probability is the position where the near-miss operation indicated by the symbol transition pattern used to calculate the similarity S has occurred. The driving support ECU 14 executes control that is determined based on the near-miss driving probability.

[Summary of the embodiment]
As described above, in the present embodiment described above, the near-miss operation probability is determined based on the similarity S between the current symbol string and the symbol transition pattern during near-miss operation. The current symbol string also includes the future symbol transitions predicted by the symbol transition prediction unit 105. Therefore, from the similarity S between the current symbol string and the symbol transition pattern during near-miss operation, it is possible to predict the probability of future near-miss operation.

  In addition, the current string includes not only future symbol transitions, but also past strings. Therefore, the prediction accuracy of the near-miss operation becomes higher than the case where the near-miss operation is predicted only by the transition of the future symbol.

  In the case of near-miss operation, situations that are likely to occur differ depending on the position. Therefore, in the present embodiment, the symbol transition pattern in the near-miss operation is stored in association with the position, and the symbol transition pattern in the near-miss operation of the position corresponding to the current symbol string and the similarity of the current symbol string S is calculated. This also increases the prediction accuracy of unsafe driving.

Second Embodiment
Next, a second embodiment will be described. In the following description of the second embodiment, elements having the same reference numerals as the reference numerals used so far are identical to the elements of the same reference numerals in the previous embodiments, unless otherwise stated. In addition, when only a part of the configuration is described, the embodiment described above can be applied to other parts of the configuration.

  In the second embodiment, the near-miss operation prediction unit 109 executes the process shown in FIG. In step (hereinafter, step is omitted) S1, similarity S is calculated. The calculation method of the similarity S is the same as that of the first embodiment.

In S2, the reach distance d _r to reach the position where the near-miss operation has occurred is calculated. The reaching distance d _r is the distance between the near-miss driving position and the current position identified by the symbol transition pattern used for calculating the similarity S.

In S3, the corrected similarity S _d is calculated. The corrected similarity S _d is calculated by the equation shown in FIG. This expression, a value obtained by subtracting the reach _{d r} from the overall distance _{d total} and molecules, the weighting factor the denominator the total distance _{d total,} is an expression to multiply the degree of similarity S. FIG. 7 shows the total distance d _total and the reach distance d _r . The total distance d _total is a distance between two positions respectively corresponding to both ends of the current symbol string. The shorter the reach distance d _{r, the} larger the weighting factor by which the similarity S is multiplied.

In S4, and defining the relationship between the corrected similarity S _d and Incident operation probability table, from the calculated correction similarity S _d in S3, determining the near-miss operation probability. The above table may be the same as that used in the first embodiment. In S5, the near-miss driving probability determined in S4 is output to the driving support ECU 14.

If the degree of similarity S is the same, it can be considered that the smaller the proportion of the predicted symbol in the current symbol string used for calculation of the degree of similarity S, the higher the possibility of a near-miss operation. It is because the prediction part which is uncertain occurs. Therefore, in the second embodiment, as the reaching distance d _r becomes shorter, the weighting coefficient by which the similarity S is multiplied is increased to calculate the corrected similarity S _d . By doing this, it is possible to calculate a more accurate near-miss operation probability.

Third Embodiment
Next, a third embodiment will be described. In the third embodiment, the near-miss operation prediction unit 109 executes the process shown in FIG. The execution cycle of FIG. 8 may be the same as the cycle for determining the near-miss operation probability in the first embodiment and the second embodiment.

  In S11, the near-miss operation probability is determined. The method of determining the near-missing operation probability here may be the method described in the first embodiment or the method described in the second embodiment.

  In S12, a near-miss operation prediction signal is generated. The near-miss operation prediction signal is a signal including the near-miss operation probability determined in S11 and the near-miss operation position. The near-miss operation prediction signal corresponds to the unsafe operation prediction signal.

  In S13, the near-miss driving prediction signal is wirelessly transmitted from the wireless communication unit 12 around the vehicle 2 on which the in-vehicle system 10 is mounted. If another vehicle exists around the vehicle and the in-vehicle system 10 is mounted on the other vehicle, the near-miss driving prediction signal may be transmitted from the in-vehicle system 10 mounted on the other vehicle.

  In S14, it is determined whether the wireless communication unit 12 receives a near-miss driving prediction signal from another vehicle within a predetermined time. The fixed time may be, for example, a cycle for calculating the near-miss driving probability. If the determination in S14 is NO, the process proceeds to S18, and if YES, the process proceeds to S15.

  In S15, it is determined whether the near-miss driving position included in the received near-miss driving prediction signal is the same as the near-miss driving position included in the near-miss driving prediction signal transmitted in S13. If the determination in S15 is NO, the process proceeds to S18, and if YES, the process proceeds to S16.

  In S16, it is determined whether the near-miss operation probability contained in the received near-miss driving prediction signal is higher than the near-miss driving probability contained in the near-miss driving prediction signal transmitted in S13. If the determination in S16 is NO, the process proceeds to S18, and if YES, the process proceeds to S17.

  In S17, the near-miss operation probability determined in S11 is updated to the near-miss operation probability contained in the received near-miss operation prediction signal. Then, it progresses to S18. In S18, the near-miss driving probability is output to the driving support ECU 14.

Summary of Third Embodiment
In the third embodiment, the driving prediction ECU 100 determines the near-missing driving probability determined by itself and the near-missing driving probability determined for the same near-missing driving position by the on-vehicle system 10 used in the other vehicles existing in the surroundings. Compare. If the near-miss driving probability determined by the on-vehicle system 10 used in the other vehicle is a higher value, the near-miss driving probability determined by the on-vehicle system 10 used in the other vehicle is the near-missing driving probability. Update to

  If the near-miss driving position is the same as the near-miss driving position determined by the on-vehicle system 10 used in the other vehicle, it can be estimated that the cause of the near-miss driving is the other vehicle. The near-miss driving probability is also determined using the driving environment data. However, the own vehicle and the other vehicle can not necessarily obtain substantially the same data as traveling environment data that affects the near-miss driving probability because the traveling direction and the position are different. In some cases, only the in-vehicle system 10 mounted on another vehicle may be able to acquire data that affect the near-miss driving probability.

  Therefore, in the third embodiment, when the near-miss driving position is the same as the near-miss driving position determined by the on-vehicle system 10 used in the other vehicle, the near-miss driving probability determined by the other vehicle and the near-miss hat determined by itself. Compare with the driving probability. Then, if the near-miss driving probability determined by the other vehicle is a higher value, the near-miss driving probability is updated to the near-hit driving probability determined by the other vehicle. As a result, it is possible to determine more accurate near-miss driving probability.

  As mentioned above, although embodiment was described, the disclosed art is not limited to the above-mentioned embodiment, and the following modification is also contained in the range indicated, and also within the range which does not deviate from the gist besides the following. Various modifications can be made.

<Modification 1>
In the second embodiment, the weighting factor is calculated, which increases as the reaching distance d _r decreases. Because there is a correlation between distance and time, instead of the reaching distance d _r , a weighting factor that increases as the reaching time to reach the near-miss driving point becomes shorter is calculated, and the weighting factor is calculated according to the second embodiment. It may be used instead of the weighting factor of. For example, in the equation shown in FIG. 6, the weighting factor is set such that the denominator is the total time which is the difference between the two times respectively corresponding to both ends of the current string, and the numerator is the total time minus the arrival time It should be a value.

<Modification 2>
There is an individual difference in the probability of performing near-miss driving, and there are drivers who tend to perform near-miss driving and drivers who are not. Therefore, the degree of similarity S or the corrected degree of similarity S _d calculated in the first embodiment, the second embodiment, or the modification 1 may be further multiplied by a weighting factor (hereinafter referred to as a driver coefficient) determined for each driver. . The driver coefficient is set to a larger value as the driver tends to drive unsafely. The specific value of the driver coefficient is set, for example, in accordance with the number of times the driver has caused a near-miss driving in the past. By doing this, it is possible to determine even more accurate near-miss operation probability.

<Modification 3>
In the near-missing operation database 5, the symbol strings when the near-missing operation is performed are stored in a typified manner. However, the symbol string when the near-miss operation is performed may be stored in the near-miss operation database 5 as it is, without being typed.

1: near-miss driving prediction system 2: vehicle 3: server 4: communication unit 5: near-miss driving database 6: control unit 10: in-vehicle system 11: satellite positioning unit 12: wireless communication unit 13: peripheral monitoring sensor 14: driving support ECU 15 : In-vehicle LAN 100: Driving prediction ECU 101: Behavior related data acquisition unit 102: Vehicle behavior data acquisition unit 103: Traveling environment data acquisition unit 104: Symbolization unit 105: Symbol transition prediction unit 106: Traveling route calculation unit 107: Symbol transition Pattern acquisition unit 108: Transition sequence generation unit 109: Near-miss operation prediction unit 120: Storage unit S: Similarity Sd: Corrected similarity dr: Reached distance

Claims (8)

An unsafe driving prediction device for predicting that driving of a vehicle will be unsafe driving,
A behavior related data acquisition unit (101) for sequentially acquiring behavior related data, which is data representing at least one of the behavior of the vehicle and the traveling environment around the vehicle, in association with the position of the vehicle;
A behavior related data prediction unit (105) for predicting the future behavior related data of the vehicle from the past behavior related data up to the present time obtained by sequentially acquiring the behavior related data acquisition unit;
A behavior related data transition sequence representing transitions of a plurality of the behavior related data, wherein the behavior related data acquired in the past acquired by the behavior related data acquiring unit and the future behavior related data predicted by the behavior related data predicting unit A transition sequence generation unit (108) that generates the current behavior-related data transition sequence of the vehicle including data;
The position of the position corresponding to the current behavior-related data transition sequence generated by the transition sequence generation unit from the unsafety-operation database storing the behavior-related data transition sequence during the unsafe operation in association with the position. A transition sequence acquisition unit (107) for acquiring the behavior related data transition sequence during unsafe operation;
Based on the similarity between the current behavior-related data transition sequence and the behavior-related data transition sequence during the unsafe driving, an unsafe driving probability is determined that is the probability that the driving of the vehicle will be the unsafe driving. An unsafe driving prediction unit (109); The behavior related data is represented by a symbol divided by the state of the vehicle serving as a division unit,
The unsafe driving prediction apparatus according to claim 1, wherein the behavior related data transition sequence is represented by a symbol sequence including a plurality of the symbols.   The unsafe driving prediction according to claim 2, wherein the behavior related data acquisition unit acquires, as the behavior related data, both vehicle behavior data representing the behavior of the vehicle and traveling environment data representing a traveling environment around the vehicle. apparatus.   The transition sequence generation unit generates the current behavior-related data transition sequence that matches the length of the behavior-related data transition sequence during the unsafe operation stored in the unsafe operation database. The unsafe driving prediction device according to any one of to 3.   The unsafe driving prediction unit determines the unsafe driving probability based on the similarity and the reaching distance until the vehicle reaches the position where the unsafe driving has occurred, and the reaching distance The unsafe driving prediction apparatus according to any one of claims 1 to 4, wherein the unsafe driving probability is determined by multiplying the similarity degree by a larger weight as the length becomes shorter.   The unsafe driving prediction unit determines the unsafe driving probability based on the similarity and the reaching time until the vehicle reaches the position where the unsafe driving has occurred, and the reaching time The unsafe driving prediction apparatus according to any one of claims 1 to 4, wherein the unsafe driving probability is determined by multiplying the similarity degree by a larger weight as the length becomes shorter.   The unsafe driving prediction unit determines the unsafe driving probability by multiplying the similarity by a larger weight as the driver of the vehicle tends to perform the unsafe driving. The unsafe driving prediction device according to claim 1 or 2. The unsafe driving prediction signal including the unsafe driving probability predicted by the unsafe driving predicting unit and the unsafe driving position where the unsafe driving may be performed is wirelessly transmitted around the vehicle. A wireless communication unit (12) for receiving the unsafe driving prediction signal transmitted by the other unsafe driving prediction apparatus used in another vehicle existing around the vehicle;
The unsafe driving prediction unit, the unsafe driving position included in the unsafe driving prediction signal received by the wireless communication unit is included in the unsafe driving prediction signal transmitted. The unsafe driving probability included in the received unsafe driving prediction signal is the same as the position and the unsafe driving probability included in the transmitted unsafe driving prediction signal. The unsafe driving prediction according to any one of claims 1 to 7, wherein when it is high, the predicted unsafe driving probability is updated to the unsafe driving probability included in the received unsafe driving prediction signal. apparatus.

Images