JP2009237937A - Driver model processor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To create a driver model as a base for a driving operation of a driver. <P>SOLUTION: A vehicle 2 stores an individual driver model 12 and a best driver model 14. The vehicle 2 inputs measurement data 11 in the driver models, predicts a driver operation 13 of the driver of the vehicle 2 and the best operation 15 of the best driver, and presents an advice 17 to the driver by using the difference 16 between them. On the other hand, the vehicle 2 uploads the measurement data 11 to a driver model server 3 during traveling. The driver model server 3 creates a driver model of the driver of the vehicle 2 by using the measurement data 11 and compares it with the existing best driver model. When the new driver model is better, the driver model server 3 updates the best driver model to the new driver model. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a driver model processing device, for example, a device that processes a driver model of a driver driving a vehicle.

In recent years, a technology for sensing a driver's own state and driving behavior and using the data to assist the driver in driving has been developed.
One of such driving support technologies uses a driver model, and there are various types. As an example using statistical data, the following “driving behavior estimation device, Driving support device and vehicle evaluation system ".
JP 2007-176396 A

This technology uses GMM (mixed Gaussian model) to model the driver's driving behavior (driving operation), and predicts the accelerator operation and brake operation performed by the driver when parameters such as vehicle speed are input. It is possible to do.
By using such a driver model, the driver's driving operation with respect to the current driving environment (road environment) is predicted, and appropriate warnings and warnings are given to the driver, the guidance route is changed by navigation, etc. Control can be performed.

By the way, in recent years, the need for more energy-efficient driving operation (hereinafter, eco-driving) has been recognized due to soaring crude oil prices and increasing awareness of global environmental protection.
For example, in the “network type eco-drive diagnosis system” of the following Patent Document 2, vehicles are connected by a network, and vehicles that are performing eco-drive are ranked. As a result, the driver's awareness of improving fuel efficiency can be increased.
JP2007-315335

  However, in this system, the relative ranking of each driver can be known. However, each driver has tried and errored about the technical aspect of how to improve driving efficiency by specifically driving. There was nothing more to discover than myself. For this reason, it was difficult to exceed the limits of eco-driving that I can do.

In order to improve not only eco-driving but also driving technology such as safe driving, it is necessary to give advice according to the characteristics of the driver individually. It is preferable to give an instruction in comparison with a driving operation as a model.
For this purpose, it is considered effective to prepare a driver model that serves as a reference corresponding to a driving operation as a model, but such a driver model has not been created.

  Accordingly, an object of the present invention is to create a driver model that serves as a reference for the driving operation of the driver.

(1) In order to achieve the above object, according to the first aspect of the present invention, communication means that communicates with a plurality of vehicles, and communication with the plurality of vehicles using the communication means, the data of the vehicles are obtained. A driver model processing device comprising: a driver model acquisition unit that acquires a driver model of a driver of the vehicle based on; and an advantage determination unit that determines superiority or inferiority based on a predetermined standard of the acquired driver model. provide.
(2) In the invention according to claim 2, the driver model acquisition unit acquires data from the plurality of vehicles using the communication unit, and creates a driver model using the data, thereby generating a driver model. The driver model processing apparatus according to claim 1 is obtained.
(3) In the invention described in claim 3, the driver model acquisition unit acquires a driver model from the plurality of vehicles using the communication unit. I will provide a.
(4) In the invention according to claim 4, the driver model storage means for storing the driver model determined to be superior by the superiority / inferiority determination means, and the driver model storage means is superior to the stored driver model. 2. An update means for updating a driver model stored in the driver model storage means with the driver model determined to be superior when there is a driver model determined to be present. A driver model processing device according to claim 2 or claim 3 is provided.
(5) The invention according to claim 5, further comprising classification information acquisition means for acquiring classification information for classifying the driver model by category from a vehicle from which the driver model is acquired, wherein The driver model is determined for each category, the driver model storage unit stores a driver model for each category, and the update unit updates the driver model for each category. A driver model processing device according to any one of claims 4 to 4 is provided.
(6) In the invention according to claim 6, the driver model uses time series data of N types of feature amounts detected as the vehicle travels as learning data, and is based on a probability distribution in which each data in the N-dimensional space exists. The driver model processing device according to any one of claims 1 to 5, wherein the driver model processing device is defined.

(1) According to the first aspect of the present invention, it is possible to determine the superiority or inferiority by acquiring the driver model of the driver of the vehicle.
(2) According to the invention described in claim 2, it is possible to receive the data from the vehicle and create a driver model using the data.
(3) According to the invention described in claim 3, it is possible to receive the driver model from the vehicle.
(4) According to the invention described in claim 4, the stored driver model can be updated with the driver model determined to be superior.
(5) According to the invention described in claim 5, it is possible to judge, store and update the superiority or inferiority of the driver model for each category.
(6) According to the invention described in claim 6, the driver model can be configured by the probability distribution.

(1) Outline of Embodiment FIG. 1A is a diagram for explaining the outline of the present embodiment.
The vehicle 2 stores the personal driver model 12 created for the individual driver and, for example, downloads the best driver model 14 which is the best driver individual driver model for the eco-drive from the driver model server 3 and stores it. is doing.
The personal driver model 12 and the best driver model 14 are configured by a model using a probability density distribution such as GMM, for example. When a predetermined parameter such as a vehicle speed is input, the driver's next driving operation (for example, a brake operation) ) Can be predicted.

  As described above, the vehicle 2 prepares the personal driver model 12 and the best driver model 14 and inputs the measurement data 11 such as the vehicle speed to the driver model as a parameter, and the driver of the vehicle 2 will perform. When the driver 13 of the vehicle 2 is the best driver, the best operation 15 that the best driver will perform is predicted.

This difference 16, that is, a difference in driving operation technology that needs to be filled in order for the driver of the vehicle 2 to approach the best driver, and the vehicle 2 presents the advice 17 to the driver using this difference 16.
According to the advice 17, the driver can learn the best driver's driving operation and improve his / her driving skill by imitating the driving operation to be performed from the best driver's driving operation.

On the other hand, the vehicle 2 gives the advice 17 to the driver as described above, and uploads the measurement data 11 to the driver model server 3 while traveling (or collectively after traveling).
The driver model server 3 creates a driver model of the driver of the vehicle 2 using the measurement data 11, and determines superiority or inferiority with the currently stored best driver model according to a predetermined condition such as fuel consumption.

If the newly created driver model is superior, the driver model server 3 updates the stored best driver model 14 to the newly created driver model.
In this way, the driver model server 3 can transmit the most excellent driver model at the present time to the vehicle 2.

FIG. 1B is an operational image of the present embodiment.
The vehicle 2 travels from a departure point (S in the figure) to a destination point (G in the figure) by route search by navigation.
For example, when eco-driving, driving according to the driving parameters with the best fuel efficiency on this route will result in the best fuel efficiency, but driving under the same conditions due to the traffic volume of the route, surrounding vehicles, time, weather, etc. Is impossible. For this reason, the significance of using a driver model arises.
In other words, since the driving operation performed by the best driver can be predicted from the best driver model 14 in the situation where the vehicle 2 is currently placed, the driver performs driving according to the best driver model 14, so The most fuel efficient driving operation can be performed.

According to the driver model, “Depress the accelerator pedal so that it becomes ○ m / s · s when accelerating. ○ The engine speed is ○ × 1000rpm when driving ○ km / h. The inter-vehicle distance has changed from ○ to Δm. Sometimes the engine speed is increased by □ rpm. The speed is set to ○ km / h just before the slope, and the driving operation unique to the driver is modeled. Driving, the driver can improve driving skills.
In this way, since the driving operation for the next possible event can be understood by using the driver model, it is possible to express the difference in the operation by comparing the driver models with each other. You can see the difference in driving operation of the driver.

In this embodiment, the GMM is used as the driver model. However, since the driving operation varies even for the same person, the GMM method for constructing the driver's driving behavior model based on the deviation in the probability density distribution is suitable. Because it is.
In addition, since the conditions (traffic volume, surrounding vehicles) on which the individual driver model 12 and the best driver model 14 are presupposed are not exactly the same, when they are compared, the ratio of the two matches according to the deviation. This is because it is possible to determine whether there is a difference.

(2) Details of Embodiment FIG. 2 is a diagram for explaining a concept related to creation of a driver model according to the present embodiment and estimation of driving operation based on the created driver model.
In the present embodiment, it is assumed that the driving operation of the driver is described by a predetermined feature amount.
Specifically, the driver determines the current vehicle speed V, the inter-vehicle distance F, and the primary dynamic feature amounts ΔV and ΔF (first-order differential values according to time) and the secondary dynamic feature amounts ΔΔV and ΔΔF (time). Accelerator pedal operation (accelerator operation amount G and next dynamic feature amount ΔG) and brake pedal operation (brake operation amount B and primary dynamic feature amount ΔB) Assuming that
Here, the reason why the modeling is performed in consideration of the primary and secondary quantities is to enable smoother and more natural driving operation prediction.

First, the driver model is created as follows.
First, when the vehicle travels, travel data 101 composed of the accelerator operation amount, the brake pedal operation amount, the vehicle speed, and the inter-vehicle distance at times t1, t2,... Is sampled at a predetermined sampling rate. The first-order differential value and the second-order differential value can be calculated from these.

When the EM algorithm is applied to the travel data collected in this way, a driver model by GMM is created.
A method for creating (estimating) a GMM by applying an EM algorithm to collected data is described in, for example, Seiichi Nakagawa, “Voice Recognition Using a Stochastic Model” (The Institute of Electronics, Information and Communication Engineers 1988, P51 to P54). Yes.

In the driver model created as described above, the driving operation is characterized by the simultaneous distribution of feature values, and the driving operation with the highest probability (accelerator pedal operation amount, brake pedal operation amount) under given conditions. Is a problem of maximizing the so-called conditional probability.
In the present embodiment, when a driver's driving operation (for example, an accelerator operation amount or a brake operation amount) is estimated, a corresponding GMM driver model 102 (for example, an accelerator model or a brake model) is used to drive at time t. By calculating the maximum a posteriori probability calculation 104 for the measured data values (V, F, ΔV,...) By numerical analysis using the EM algorithm, the accelerator operation amount 105 that the driver will operate can be estimated. .

More specific description will be given below.
(A) Creation of Driver Model In a driver model using GMM, learning data for learning a driver's driving operation is required, and traveling data 101 is used as a feature amount.
The travel data 101 uses time-series data for each predetermined measurement interval s (s is arbitrary, but s = 0.1 seconds in the present embodiment).
The travel data 101 is data actually driven by a driver for which a driver model is to be generated, and offline learning is possible by using the travel data 101 that has been measured and stored in advance. Further, the travel data 101 measured and collected in real time when the driver is actually driving may be used.

In this embodiment, by generating a GMM for each driver, modeling that matches the characteristics of each driver is possible.
As described above, the driver model feature amount (travel data 101) includes the vehicle speed, the inter-vehicle distance, and the primary, secondary dynamic feature amount, accelerator pedal operation amount, and primary order of the accelerator pedal operation amount. Of dynamic features are used.
As described above, by adding the dynamic feature amount to the feature amount and modeling, the time relationship before and after is taken into consideration, and a smooth and highly natural estimation result can be obtained.
In the description, the case where the primary and secondary dynamic feature quantities are used will be described. However, only the primary dynamic feature quantity may be used.

Similarly, a driver model for the brake pedal is also possible.
When generating a plurality of driver models for the accelerator pedal, the brake pedal, the inter-vehicle distance range, etc., data other than the accelerator pedal operation amount, the brake pedal operation amount, etc. (V, F, ΔV, ΔF,... ) May use the same data.

  In the present embodiment, the dynamic feature amount of the travel data is calculated from the measured values of the accelerator operation amount, the vehicle speed, and the inter-vehicle distance, but may be actually measured.

In this embodiment, a driver model is generated by calculating a mixed Gaussian distribution for the travel data 101.
That is, the joint probability density distribution for the running data 101 is calculated using the EM algorithm, and the parameters of the calculated joint probability density function = {λi, → μi, Σi | i = 1, 2, 3,. The driver model is stored in a storage means such as a database.
Here, λi represents a weight, → μi represents an average vector group, Σi represents a variance-covariance matrix group, and M represents the number of mixtures. In addition, a symbol that is previously displayed as → μi means a vector.
As described above, the GMM according to the present embodiment uses the full-width covariance matrix in consideration of the correlation between the feature dimensions.

(B) Estimation of driving operation (accelerator pedal and brake pedal operation amount) The driver determines the operation amount of the accelerator pedal and the brake pedal based on the current vehicle speed, the inter-vehicle distance, and their primary and secondary dynamic feature amounts. Based on the assumption that it is determined, the motor behavior such as the pedal operation amount is estimated.
That is, from the simultaneous distribution of the feature amount, the motion behavior such as the accelerator pedal operation amount having the highest probability under the given condition is estimated.

This is a problem of maximizing conditional probability, and is based on the calculation of maximum posterior probability.
That is, the accelerator pedal operation amount ∧G (t) and the brake pedal operation amount ∧B (t) are estimates of the value x (t) that maximizes the conditional probability under the condition where y (t) is given. The maximum posterior probabilities are calculated by the following equations (1) and (2), respectively.

∧G (t) = arg max p (G | ΔG, V (t), F (t), ΔV (t), ΔF (t), ΔΔV (t), ΔΔF (t)) Equation (1)
∧ B (t) = arg max p (B | ΔB, V (t), F (t), ΔV (t), ΔF (t), ΔΔV (t), ΔΔF (t)) Equation (2)

Here, like ∧G (t), the previous ∧ is an estimated value.
Also,
p (G | ΔG, V, F, ΔV, ΔF, ΔΔV, ΔΔF)
= {P (G, V, F, ΔV, ΔF, ΔΔV, ΔΔF, ΔG)} / {∫∫ ... ∫p (G, V, F, ΔV, ΔF, ΔΔV, ΔΔF
, ΔG) dΔG, dV, dF, dΔV, dΔF, dΔΔV, dΔΔF}
p (B | ΔB, V, F, ΔV, ΔF, ΔΔV, ΔΔF)
= {P (B, V, F, ΔV, ΔF, ΔΔV, ΔΔF, ΔB)} / {∫∫ ... ∫p (B, V, F, ΔV, ΔF, ΔΔV, ΔΔF, ΔB) dΔB, dV, dF , DΔV, dΔF, dΔΔV, dΔΔF}
It is.

  In Expressions (1) and (2), t represents time, and G, B, V, F, and Δ represent accelerator pedal operation amount, brake pedal operation amount, vehicle speed, inter-vehicle distance, and dynamic feature amount, respectively.

However, the value of the accelerator pedal and the brake pedal that maximizes the conditional probability is obtained by performing numerical integration with a certain step size (for example, 100 steps from 0 to 10,000) in the interval from the minimum value to the maximum value. The value of the accelerator pedal and the brake pedal when the probability is calculated and the probability becomes the maximum may be used as the estimation result.
As described above, the driver model (individual driver model, best driver model) used in the present embodiment generally uses time-series data of N types of feature amounts detected as the vehicle travels as learning data, and in the N-dimensional space. It is defined by the probability distribution that each data exists.

FIG. 3 is a diagram showing a network configuration of the driving operation support system 1 according to the present embodiment.
The driving operation support system 1 includes a driver model server 3, vehicles 2, 2,... (Hereinafter referred to as vehicle 2), AP4, 4,.

The driver model server 3 is connected to the AP 4 (access point) via the communication network, and can communicate with the vehicle 2 via the AP 4.
The driver model server 3 transmits the best driver model to the vehicle 2 in response to the request of the vehicle 2 and receives driver model creation information including measurement data measured by the vehicle 2 from the vehicle 2.

  Then, the driver model server 3 creates a driver model of the driver of the vehicle 2 using the driver model creation information, compares this with the best driver model, and the created driver model is more than the best driver model. If so, update the best driver model with the created driver model.

The vehicle 2 can perform wireless communication with the AP 4 and can communicate with the driver model server 3 via the AP 4.
The vehicle 2 stores the driver's personal driver model, and requests the driver model server 3 for the best driver model when traveling to download and store it.
And while driving | running | working, the vehicle 2 estimates the driving operation which a driver model performs using a personal driver model, and also predicts the driving operation which a best driver performs using a best driver model, and uses the difference of both predictions And give advice to the driver.

The advice can be presented, for example, by presenting an actual guide 43 representing the current speed of the vehicle 2 and a target guide 42 representing the speed by the best driver model on a speedometer.
If the best driver model relates to eco-driving, the driver can perform eco-driving by adjusting the speed so that the target pointer 42 matches the actual pointer 43.
Further, the vehicle 2 performs various measurements and information collection while traveling, generates driver model creation information from these, and sequentially transmits them to the driver model server 3.

  The AP 4 establishes a wireless communication line with the vehicle 2 to perform wireless communication, and connects to the driver model server 3 through the communication network, and relays communication between the vehicle 2 and the driver model server 3.

In the present embodiment, the vehicle 2 creates and stores a personal driver model of the driver from the driving operation of the driver, but the driver model server 3 creates a personal driver using the driver model creation information. And it can also comprise so that this may be transmitted to the vehicle 2 at the time of driving | running | working.
As described above, when the driver model server 3 is configured to manage the personal driver model, for example, when the driver drives another vehicle 2 such as a rental car, the driver model of the driver is transferred from the driver model server 3 to the other driver model. Operation such as downloading to the vehicle 2 and using it becomes possible.

FIG. 4 is a block diagram showing a configuration of the information processing system 20 mounted on the vehicle 2.
The information processing system 20 includes a travel information acquisition unit 22, an external environment information acquisition unit 23, a surrounding environment information acquisition unit 24, a vehicle environment information acquisition unit 25, an information processing unit 21, a communication unit 26, a best driver model storage unit 27, an individual The driver model storage unit 28 and the like function as a driving operation support device.

  The travel information acquisition unit 22 is a functional unit that acquires travel data of the vehicle 2. For example, the travel information acquisition unit 22 detects a gasoline flow rate sensor that detects a gasoline flow rate, an engine speed sensor that detects an engine speed, and an accelerator pedal depressing force. An access pedal force sensor for detecting, a vehicle speed sensor for detecting vehicle speed, a gyro sensor for detecting gyro output, a gear ratio sensor for detecting gear ratio, and the like are provided.

  The external environment information acquisition unit 23 is a functional unit that acquires a general external environment when the vehicle 2 is traveling. For example, a GPS information acquisition device that receives radio waves from a GPS (Global Positioning Systems) satellite. , VICS (Vehicle Information and Communication System) information acquisition device for receiving road traffic information, map DB (database) storing map information, navigation guidance route information acquisition unit for acquiring route guidance by navigation device, acquiring current time A current time acquisition unit, a weather information acquisition unit for acquiring weather, a road surface information acquisition unit for acquiring road surface information, and the like.

  The surrounding environment information acquisition unit 24 is a functional unit that acquires environment information around the vehicle. For example, a vehicle periphery recognition device that recognizes the periphery of the vehicle using a camera, laser, millimeter wave, and the like, as well as a camera, laser, millimeter An obstacle recognition device that recognizes an obstacle using a wave or the like, an inter-vehicle distance measurement device that measures an inter-vehicle distance, and the like.

  The vehicle environment information acquisition unit 25 is a functional unit that acquires environment information of the vehicle itself. For example, a vehicle code acquisition device that acquires a vehicle code that specifies a vehicle type, year, number, etc., the number of passengers or the weight of the passenger It includes a passenger number acquisition device that measures and acquires the weight, a loading weight acquisition device that measures and acquires the loading weight such as weight.

The information processing unit 21 is formed in an ECU (Engine Control Unit) or the like, and is configured by a device or element such as a CPU (Central Processing Unit), a ROM (Read Only Memory), or a RAM (Random Access Memory). Then, various programs are executed to perform various information processing, control of each part of the vehicle 2, and the like.
As the program executed by the information processing unit 21, for example, a data transmission / reception program that communicates with the driver model server 3 via the AP 4, a predicted value based on the personal driver model and the best driver model is calculated, and advice is presented to the driver based on the difference. A driver model comparison program, a personal driver model creation program for creating a driver's personal driver model from driving data, and other programs.

The communication unit 26 is a communication device that establishes a communication line with the AP 4 and communicates with the driver model server 3 via the AP 4.
The information processing system 20 transmits driver model creation information to the driver model server 3 via the communication unit 26, requests the driver model server 3 for the best driver model, and obtains the best driver model from the driver model server 3. Or download it.
In the present embodiment, the information processing system 20 uses all the information acquired by the travel information acquisition unit 22, the external environment information acquisition unit 23, the surrounding environment information acquisition unit 24, and the vehicle environment information acquisition unit 25 as driver model creation information. To the driver model server 3.

  It is also possible to send the personal driver model created in the vehicle 2 to the driver model server 3 and the driver model server 3 compares it with the best driver model, but sends data to the driver model server 3. By configuring the driver model server 3 to create a driver model, the degree of freedom of data processing in the driver model server 3 can be increased.

The best driver model storage unit 27 is formed in a RAM or the like, for example, and stores the best driver model downloaded from the driver model server 3.
The driver model server 3 predicts the driving operation of the best driver using the best driver model stored in the best driver model storage unit 27.

The personal driver model storage unit 28 is formed in, for example, a hard disk or a non-volatile semiconductor memory, and stores the personal driver model created by the information processing unit 21.
The individual driver model is created for each person who drives the vehicle 2, such as the person 1, the person 2,.
Even with the same driver, it can be used on wide roads such as daytime driving and driving at night, driving on low-traffic roads and driving in heavy traffic, driving in fine weather and driving in rainy weather, multi-lane roads on one side, etc. It is considered that driving characteristics or driving characteristics may change according to driving on narrow roads that use both upper and lower lines in one lane, road pavement conditions, uphill and downhill conditions, etc. Therefore, in the present embodiment, these are classified into categories by labels and stored for each category, so that a personal driver model corresponding to the current driver situation can be used.

FIG. 5 is a block diagram showing the configuration of the driver model server 3.
The driver model server 3 includes a communication unit 31, an information processing unit 32, a best driver model storage unit 33, and the like, and functions as a driver model processing device.
The communication unit 31 is connected to a communication network, and can communicate with the information processing system 20 of each vehicle 2 via the AP 4.
The driver model server 3 uses the communication unit 31 to receive driver model creation information from the vehicle 2 and transmits the best driver model to the vehicle 2 in response to a request from the vehicle 2.
Here, the communication unit 31 functions as a communication unit that communicates with the plurality of vehicles 2.

The information processing unit 32 is constructed using a mass storage device such as a CPU, ROM, RAM, and hard disk, and other information processing devices. Information processing and communication defined by these programs are performed according to various programs. I do.
The program executed by the information processing unit 32 includes a data transmission / reception program that exhibits a function of transmitting and receiving various data by communicating with the vehicle 2, and a driver model creation that creates a driver model using the driver model creation data received from the vehicle 2. A driver model management program that judges the superiority or inferiority of the program and driver model and manages the best driver model so that the best driver model is always the best driver model, and the best driver model that matches the driver of the vehicle 2 according to the request from the vehicle 2 There are driver model selection programs and other programs for selecting.

The best driver model storage unit 33 is configured using, for example, a large-capacity storage device such as a hard disk, and stores the best driver model created by the information processing unit 32 as the best driver model. .
Alternatively, a plurality of driver models may be stored and ranked so that the driver model server 3 recognizes the highest driver model as the best driver model.

In the best driver model storage unit 33, for example, an “eco-driver model”, a “safe driving driver model”, and the like are formed for each use / condition database, and each database is best suited to the use / condition. The driver model is stored.
Here, the eco-driver model is, for example, a driver model for driving with good fuel efficiency and low carbon dioxide emission, and the safe driving driver model is for example driving within a speed limit range, sudden braking, This is a driver model with a low frequency of rapid acceleration.

As described above, various uses and conditions of the driver model can be prepared according to the needs of the driver.
Each application / condition is provided with a mathematical formula and a reference for numerically comparing the superiority and inferiority of the driver model.
For example, the eco-driver model is based on fuel consumption per unit mileage per passenger, and the safe driving driver model uses the speed limit and frequency of excess, and the frequency of sudden braking and acceleration as parameters. The calculation formula for calculating is defined and is based on this.

The best driver model stored in each database is distinguished by a label within the same use / condition.
This label is for providing the vehicle 2 with the best driver model created in a situation as close as possible to the environment of the vehicle 2, and when the driver model server 3 creates the best driver model, This is set based on the data of the vehicle 2 that is the creation source.
The driver model server 3 uses the driver model creation information transmitted from the vehicle 2 to select the best driver model with a label suitable for the environment of the vehicle 2 and transmits it to the vehicle 2.

FIG. 6 is a diagram showing the concept of driver model creation by the driver model server 3.
When the service server 6 receives the driver model creation information from the vehicle 2, the service server 6 sorts the received data according to the label such as the vehicle type and applies the EM algorithm to create a driver model.
Then, the superiority and inferiority of the corresponding label with the best driver model of the corresponding label are compared for each application and condition. If the created driver model is superior, the best driver model is updated accordingly.
In the example of the figure, as an example, “low fuel consumption Meister model by GMM”, “eco-driver model by GMM”, and “eco-indifference model by GMM” are shown for each use and condition, but the example shown in FIG. The "fastest driving model" that runs at the highest speed, the "sports driving model" that enjoys sporty driving, the "elderly model" that provides driving operations suitable for the elderly, and the "beginners" that provide driving operations suitable for the beginners Various models such as “model” can be prepared.

FIG. 7 is a diagram showing a concept that the driver model server 3 sets a label on the driver model.
The driver model server 3 acquires data acquired by the travel information acquisition unit 22, the external environment information acquisition unit 23, the surrounding environment information acquisition unit 24, and the vehicle environment information acquisition unit 25 of the vehicle 2 as driver model creation information. A label is set for each driver model using the driving route, traffic conditions, and so on.
By this label, the driver model server 3 can select the best driver model created in the same vehicle type as the vehicle 2, the same or equivalent travel route, and similar traffic conditions.

When the travel route is included in the label, the best driver model is created for each travel route.
In addition, it is possible to divide the travel route into sections such as intersection units and nodes used for route search by the navigation device, and prepare the best driver model for each section.
In this case, the vehicle 2 downloads, from the driver model server 3, the best driver model of the currently traveling section every time the traveling section changes.

  Alternatively, when the departure point and the destination point are set in the vehicle 2, the vehicle 2 downloads the best driver model for each section from the departure point to the destination point from the driver model server 3, or the destination point is If it is not set, the driver model in the section around the vehicle 2 can be downloaded from the driver model server 3 at once.

In addition, when the driver is in the same environment / situation, it is considered that the driver performs the same driving operation regardless of which road the vehicle is traveling on.
For this reason, if the best driver model in a similar environment is provided to the vehicle 2 without making the best driver model correspond to the driving route, for example, the best driver model of the best driver in Hokkaido is used by the Okinawa driver. Thus, the best driver model at the national level can be provided to the driver model.

Next, the operation of the driving operation support system 1 configured as described above will be described.
FIG. 8 is a flowchart illustrating a procedure in which the driving operation support system 1 supports the driving operation of the driver of the vehicle 2.
The following processing of the vehicle 2 is performed by the information processing system 20 mounted on the vehicle 2.
First, the driver activates the vehicle 2 and inputs self-identifying information to the vehicle 2.
As a result, the vehicle 2 can recognize the driver and select the personal driver model of the driver.
The vehicle 2 accesses the driver model server 3 to establish a communication line, and the driver model server 3 also establishes a communication line with the vehicle 2 to identify the vehicle 2.

When the driver starts driving operation of the vehicle 2, the information processing system 20 determines the driver model from the information acquired by the travel information acquisition unit 22, the external environment information acquisition unit 23, the surrounding environment information acquisition unit 24, and the vehicle environment information acquisition unit 25. Generation information is generated, and transmission of driver model generation information to the driver model server 3 is started (step 5).
Then, the driver model server 3 starts to receive driver model creation information from the vehicle 2 (step 10).
Thereafter, the vehicle 2 transmits driver model creation information to the driver model server 3 every moment.

Next, the driver designates the usage / condition of the best driver model requested in the information processing system 20 as, for example, “eco-drive”.
In response to this, the vehicle 2 requests the driver model server 3 for the best driver model for the application / condition specified by the driver (step 15).

  When the driver model server 3 receives a request for the best driver model from the vehicle 2, the model of the vehicle 2, the route on which the vehicle 2 is traveling, and the vehicle 2 are placed from the database of usage / conditions designated by the driver. The best driver model with a label suitable for the environment is selected and transmitted to the vehicle 2 (step 20).

The vehicle 2 receives the best driver model from the driver model server 3 and stores it in the best driver model storage unit 27, and reads out an individual driver model suitable for the current environment from the personal driver model storage unit 28.
Then, the vehicle 2 performs a driver model comparison process, which will be described later, using both driver models (step 25), and gives an instruction so that the driving operation of the driver becomes a better driving operation following the driving operation of the best driver. .

As described above, the vehicle 2 (information processing system 20) includes a reference driver model requesting unit that requests a reference driver model as a reference for driving operation from a predetermined server such as the driver model server 3, and the predetermined server. Get the reference driver model sent by.
The reference driver model that serves as a reference for driving operation is prepared according to the attributes of the vehicle driven by the driver and the environment in which the driver drives by the label, and the information processing system 20 uses the driver model creation information, A reference driver model corresponding to the attributes of the vehicle 2 and the driving environment is required.

When the vehicle 2 reaches the destination, the vehicle 2 notifies the driver model server 3 of the end of transmission of the driver model creation information (step 30).
When the driver model server 3 receives the notification, the driver model server 3 performs driver model creation processing described later using the driver model creation information received from the vehicle 2 (step 35).

FIG. 9 is a flowchart for explaining the driver model comparison processing in step 25.
The information processing system 20 reads the personal driver model from the personal driver model storage unit 28 into a predetermined area of the RAM (step 50), receives the best driver model from the driver model server 3, and stores it in the best driver model storage unit 27. (Step 55).
In this way, the information processing system 20 acquires the personal driver model acquisition means for acquiring the driver's personal driver model, and the reference driver model acquisition for acquiring the reference driver model (here, the best driver model) that serves as a reference for driving operation. Means.

Next, the information processing system 20 acquires the feature quantity such as the current vehicle speed and the inter-vehicle distance and inputs it to the personal driver model, and the next possible event, that is, the driving operation that the driver will perform next. (Accelerator pedal operation and brake pedal operation) are predicted (step 60).
In this way, the information processing system 20 inputs parameters obtained by inputting parameters to the driver model (features such as vehicle speed) for calculating a predicted value, and the obtained parameters to the individual driver model. Thus, an individual driving operation prediction means for predicting the driving operation of the driver (driver) is provided.

Also, the information processing system 20 predicts the driving operation that would be performed if the driver was the best driver by inputting the feature quantity to the best driver model together with the prediction based on the individual driver model, and this is predicted based on the individual driver model. The predicted value is compared (step 65).
As described above, the information processing system 20 includes the reference driving operation prediction unit that inputs a parameter to the reference driver model (best driver model) and predicts a reference driving operation.

Next, the information processing system 20 determines whether there is a difference in the driving operation as a result of the comparison (step 70). If there is a difference (step 70; Y), the information processing system 20 determines the difference. For example, the driver is notified by a speedometer pointer or the like (step 75).
The determination of whether or not there is a difference may be configured such that if there is a difference greater than or equal to a predetermined threshold, it is determined that there is a difference, and if the difference is less than the predetermined threshold, it is determined that there is no difference. Good.
In this manner, the information processing system 20 includes a difference output unit that outputs a difference between the predicted driving operation of the driver (driver) and the predicted driving operation as a reference.

After notifying the difference in step 75, and if there is no difference in step 70, or if the difference is less than a predetermined threshold (step 70; N), the process proceeds to step 80, and whether or not the operation is completed. Is determined (step 80).
If the driving does not end (step 80; N), the information processing system 20 returns to step 60 and continues the comparison with the individual driver model and the best driver model.
When the driving is finished (step 80; Y), the information processing system 20 finishes the driver model comparison process.

FIG. 10 is a flowchart for explaining the driver model creation processing in step 35.
The driver model server 3 receives driver model creation data from the vehicle 2 every moment (step 100) and stores it.

When the vehicle 2 finishes traveling, the driver model server 3 creates a driver model using the stored driver model creation data (step 105).
In the present embodiment, the driver model is created after the vehicle 2 has finished traveling. However, the driver model creation data may be received from the vehicle 2 and the driver model may be created when data is accumulated to some extent. Good.

As described above, the driver model server 3 includes driver model acquisition means for acquiring data from a plurality of vehicles using communication means and acquiring a driver model by creating a driver model using the data.
Further, when the driver model server 3 is configured to receive the personal driver model created by the vehicle 2, the driver model server 3 communicates with the plurality of vehicles 2 using the communication means, so that the data of the vehicle 2 is obtained. Driver model acquisition means for acquiring the driver model of the driver (driver) of the vehicle 2 based on the above.

Next, the driver model server 3 sets usage / conditions of the created driver model (110).
This use / condition is, for example, the same as the use / condition of the best driver model requested by the driver.
This is because, for example, when the driver requests the driver model server 3 for the best driver model for eco-driving, the driver can expect driving operation while keeping eco-driving in mind. This is because the appropriate usage and conditions for the driver model creation information to be used are eco-driving.

Next, the driver model server 3 analyzes the driver model creation information received from the vehicle 2 and sets a label corresponding to the driver model (step 115).
Here, the usage / conditions and label set in the driver model function as a category for classifying the driver model, and the driver model server 3 labels the driver model from the vehicle 2 from which the driver model is obtained. Classification information to obtain classification information (driver model creation information used to set the best driver model usage / conditions requested by the driver used to set the usage / category) An acquisition means is provided.

Next, the driver model server 3 determines the superiority or inferiority of the best driver model corresponding to the created driver model, the use / condition set in the driver model, and the label (step 120).
This superiority or inferiority determination is made based on a predetermined condition set for the application / condition, for example, when the application / condition is eco-drive, the determination is made based on the fuel consumption per unit distance.
As described above, the driver model server 3 includes superiority / inferiority determining means for determining superiority or inferiority based on a predetermined standard of the driver model.

As a result of the determination, if the created driver model is superior to the best driver model (step 125; Y), the driver model server 3 sets the created driver model as a new best driver model and uses this as the current best driver model. The stored best driver model is overwritten and updated (step 130).
On the other hand, as a result of the determination, if the best driver model is superior to the created driver model (step 125; N), the driver model server 3 discards the created driver model (step 135) and determines the best driver model. Do not update.

As described above, the driver model server 3 includes a driver model storage unit (best driver model storage unit 33) that stores a driver model that is determined to be superior by the superiority / inferiority determination unit, and the stored driver by the superiority / inferiority determination unit. When there is a driver model determined to be superior to the model, an update unit is provided to update the driver model stored in the driver model storage unit with the driver model determined to be superior.
In addition, the driver model server 3 determines whether the driver model is superior or inferior for each category (use / condition and label), stores the driver model for each category, and updates the driver model for each category. Yes.

As described above, the driver model server 3 can obtain the driver model of each driver based on the driver model creation information transmitted from each vehicle 2, and compares the best driver model with the best driver model. By updating, it is possible to always provide the vehicle 2 with the best driver model.
In the present embodiment, the driver model creation information transmitted from the vehicle 2 is processed by the driver model server 3 so as to create a driver model. It can also be configured to transmit to the driver model server 3.

Next, a presentation example of a driving operation instruction performed by the vehicle 2 will be described.
FIG. 11A is a diagram illustrating an example of the meter panel 40 of the vehicle 2.
The meter panel 40 includes a speedometer 41, a tachometer 44, an accelerator display 47, a fuel consumption display 48, and the like.

The speedometer 41 includes an actual pointer 43 that indicates the speed at which the actual vehicle 2 is traveling, and a target pointer 42 that indicates a target value of speed.
The target value of the speed is a speed calculated by using a difference between a predicted value of the driving operation that the driver is going to perform and a predicted value of the driving operation that is predicted to be performed by the best driver.
In this way, the information processing system 20 can make an indication to the driver that, for example, if the vehicle is an eco-drive, the vehicle is driven at a slightly lower speed.
Therefore, the driver can follow the driving operation of the best driver by performing the driving operation toward the target pointer 42.

The tachometer 44 includes an actual pointer 46 that indicates the actual engine speed of the vehicle 2 and a target pointer 45 that indicates a target value of the engine speed.
The target value of the engine speed is a speed calculated by using a difference between a predicted value of a driving operation to be performed by the driver and a predicted value of a driving operation predicted to be performed by the best driver.
Therefore, the driver can follow the driving operation of the best driver by performing the driving operation toward the target pointer 45.

The accelerator display 47 is a display for warning the user that the accelerator pedal has been depressed too much, and the display color changes depending on whether the warning is given or not.
The information processing system 20 calculates the amount of depression of the accelerator pedal that the driver wants to perform and the amount of depression of the accelerator pedal that the best driver wants to perform, and determines whether or not it is excessively depressed based on this difference.

The fuel consumption display 48 displays the current fuel consumption of the vehicle 2 as a bar, and the longer the fuel consumption, the longer the bar.
When the vehicle 2 is traveling on an eco drive, the target value of fuel consumption is displayed together with a display “Eco”.
This target value of fuel consumption is calculated using the difference between the fuel consumption based on the predicted value of the driving operation to be performed by the driver and the fuel consumption based on the predicted value of the driving operation predicted to be performed by the best driver.
Therefore, the driver can follow the driving operation of the best driver by performing the driving operation toward the target value.

FIG. 11B shows the meter panel 40 when the driver performs a driving operation based on an instruction given by the information processing system 20.
In the speedometer 41, the actual pointer 43 matches the target pointer 42, in the tachometer 44, the actual pointer 46 matches the tachometer 44, the warning is released in the accelerator display 47, and the fuel consumption display 48 shows the current fuel consumption and fuel consumption. The target value matches.
In this state, the driver of the vehicle 2 performs the driving operation in the same manner as the driving operation predicted to be performed by the best driver. In this way, the driver performs the same driving operation as the best driver and the best driver. You can learn how to drive.

As described above, this embodiment can provide the best driver model for various applications and conditions. However, due to the recent growing interest in the global environment, eco-driving is one of the major applications and conditions. Raised.
This is evident from the growing demand for fuel-efficient vehicles, but in order to further reduce fuel consumption, we will promote energy-saving technology on the vehicle side and improve driver operation technology.・ It is important to improve.

  In other words, the part of vehicle fuel efficiency that can be solved by the technology on the vehicle side can be improved by the efforts of the vehicle manufacturer, but the parts that depend on the technology and skill of the driving operation are Depending on the driving characteristics (癖) and the external environment such as road type, shape and traffic conditions, and the individual conditions such as total weight such as vehicle weight, number of passengers, and loading capacity, the optimum accelerator operation and braking Since operation, steering operation, etc. are different, it is difficult to improve by yourself.

  Therefore, as in the present embodiment, a personal driver model of a driver who is in a certain vehicle type is compared with the best driver model of a driver who drives the same vehicle type with lower fuel consumption through a communication network, and the difference By teaching the driver how to operate and know-how, the vehicle can contribute to further reduction in fuel consumption of the vehicle.

The following effects can be obtained by the present embodiment described above.
(1) When the information processing system 20 uses the personal driver model and the best driver model, the driving operation predicted to be performed by the driver of the vehicle 2 and when the best driver is driving the vehicle 2 Predicted driving operations can be compared.
(2) As a result of the comparison, the information processing system 20 can present a driving operation instruction based on the difference to the driver model.
(3) The driver can perform the driving operation in accordance with the best driver by receiving an instruction prior to the driving operation to be performed.

(4) The driver model server 3 can receive driver model creation information from each vehicle 2 and create a driver model for each driver.
(5) The driver model server 3 can determine the superiority or inferiority of the created driver model, and can select the best driver model as the best driver model.
(6) The driver model server 3 can transmit the best driver model most suitable for the vehicle 2 by managing the best driver model by categorizing and managing the best driver model according to use / conditions and labels.

In the present embodiment, the driver model of the driver of the vehicle based on the data of the vehicle is acquired by communicating with the plurality of vehicles using the communication unit and the communication unit that communicates with the plurality of vehicles. It is possible to provide a driver model processing apparatus comprising a driver model acquisition unit and an superiority / inferiority determination unit that determines superiority or inferiority based on a predetermined criterion of the acquired driver model (first configuration).
In the first configuration, the driver model acquisition unit is configured to acquire data from the plurality of vehicles using the communication unit, and to acquire a driver model by creating a driver model using the data. (Second configuration).
In the first configuration, the driver model acquisition unit may be configured to acquire a driver model from the plurality of vehicles using the communication unit (third configuration).
In the first configuration, the second configuration, or the third configuration, a driver model storage unit that stores a driver model that is determined to be superior by the superiority or inferiority determination unit, and the driver that is stored by the superiority / inferiority determination unit And updating means for updating the driver model stored in the driver model storage means with the driver model determined to be superior when there is a driver model determined to be superior to the model. It is also possible (fourth configuration).
In any one of the configurations from the first configuration to the fourth configuration, classification information acquisition means for acquiring classification information for classifying the driver model by category from the vehicle from which the driver model is acquired is provided. The superiority / inferiority determining means determines a driver model for each category, the driver model storage means stores a driver model for each category, and the updating means updates the driver model for each category. It can also be configured (fifth configuration).
In any one configuration from the first configuration to the fifth configuration, the driver model uses time-series data of N types of feature amounts detected as the vehicle travels as learning data, and in the N-dimensional space It can also be configured so as to be defined by a probability distribution in which each data exists (sixth configuration).

Further, the present embodiment is a parameter acquisition unit that acquires parameters for calculating a predicted value by inputting to the driver model, a personal driver model acquisition unit that acquires a driver's personal driver model, and a reference. Reference driver model acquisition means for acquiring a reference driver model, personal driving operation prediction means for predicting the driving operation of the driver by inputting the acquired parameter to the acquired individual driver model, and the acquired reference driver model A reference driving operation prediction unit that inputs the acquired parameter to predict a driving operation as a reference, and a difference output unit that outputs a difference between the predicted driving operation of the driver and the predicted driving operation as a reference It is possible to provide a driving operation support device characterized by comprising (first configuration).
1st composition WHEREIN: The reference | standard driver model request | requirement means which requests | requires a reference | standard driver model from a predetermined | prescribed server is provided, The said reference | standard driver model acquisition means is comprised so that the reference | standard driver model which the said predetermined | prescribed server transmitted may be acquired It is also possible (second configuration).
In the second configuration, the reference driver model is prepared corresponding to the attribute of the vehicle driven by the driver and the driving environment, and the reference driver model requesting means is set to the attribute of the vehicle and the driving environment. It can also be configured to request a corresponding reference driver model (third configuration).
In the first configuration, the second configuration, or the third configuration, the personal driver model and the reference driver model use, as learning data, time series data of N types of feature amounts detected as the vehicle travels, It can also be configured such that each data in the N-dimensional space is defined by a probability distribution existing (fourth configuration).

It is a figure for demonstrating the outline | summary of this Embodiment. It is a figure for demonstrating the concept etc. of a driver model. It is the figure which showed the network structure of the driving operation assistance system. It is the block diagram which showed the structure of the information processing system. It is the block diagram which showed the structure of the driver model server. It is the figure which showed the concept of driver model creation. It is the figure which showed the concept which sets a label to a driver model. It is the flowchart which showed the procedure in which a driving operation assistance system supports driving operation. It is a flowchart for demonstrating a driver model comparison process. It is a flowchart for demonstrating a driver model creation process. It is a figure for demonstrating the example of presentation of a driving operation instruction | indication.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Driving operation support system 2 Vehicle 3 Driver model server 20 Information processing system 21 Information processing part 22 Running information acquisition part 23 External environment information acquisition part 24 Ambient environment information acquisition part 25 Vehicle environment information acquisition part 26 Communication part 27 Best driver model memory | storage Unit 28 personal driver model storage unit 31 communication unit 32 information processing unit 33 best driver model storage unit 40 meter panel 41 speedometer 42 target pointer 43 actual pointer 44 tachometer 45 target pointer 46 actual pointer 47 accelerator display 48 fuel consumption display

Claims (6)

Communication means for communicating with a plurality of vehicles;
Driver model acquisition means for acquiring a driver model of a driver of the vehicle based on data of the vehicle by communicating with a plurality of vehicles using the communication means;
Superiority or inferiority judging means for judging superiority or inferiority based on a predetermined standard of the acquired driver model;
A driver model processing apparatus comprising:   The driver model acquisition unit acquires the driver model by acquiring data from the plurality of vehicles using the communication unit and creating a driver model using the data. Driver model processing device.   The driver model processing apparatus according to claim 1, wherein the driver model acquisition unit acquires a driver model from the plurality of vehicles using the communication unit. Driver model storage means for storing the driver model determined to be superior by the superiority / inferiority determination means;
When there is a driver model that is determined to be superior to the stored driver model by the superiority / inferiority determination means, the driver model stored in the driver model storage means is updated with the driver model determined to be superior. Update means;
The driver model processing apparatus according to claim 1, wherein the driver model processing apparatus is provided. Classification information acquisition means for acquiring classification information for classifying the driver model by category from the vehicle from which the driver model is acquired,
The superiority / inferiority determining means determines a driver model for each category,
The driver model storage means stores a driver model for each category,
The driver model processing apparatus according to any one of claims 1 to 4, wherein the updating unit updates a driver model for each category.   2. The driver model is defined by a probability distribution in which each type of data in an N-dimensional space exists using time series data of N types of feature amounts detected as the vehicle travels as learning data. The driver model processing device according to any one of claims 1 to 5.

Images