JP6575933B2 - Vehicle driving support system - Google Patents

Description

  The present invention relates to a vehicle driving support system, and more particularly to a vehicle driving support system using a driver model.

  In recent years, use of a driver model has been proposed to support vehicle control (see, for example, Patent Document 1). Patent Document 1 describes a driver model particularly related to driving operation. In the driver model processing apparatus of Patent Document 1, an individual driver model created for the vehicle and a best driver model created by a driver model server outside the vehicle are used. In this device, the best driver model created based on data from a large number of vehicles is used as a reference, and when there is a difference between the individual driver model and the best driver model, advice is given to the driver corresponding to the difference. It has come to be offered.

JP 2009-237937 A

  However, in the future, if an attempt is made to create a highly accurate personal driver model using artificial intelligence in a vehicle, the computation load of the in-vehicle processor becomes very high, and there is a risk that vehicle control will be delayed.

  The present invention has been made to solve such a problem, and provides a vehicle driving support system capable of preventing a delay in vehicle control while creating a personal driver model in a vehicle. Objective.

In order to achieve the above object, a vehicle driving support system according to the present invention includes a first in-vehicle computing unit having artificial intelligence for learning a personal driver model unique to a driver based on driving data of the driver, and executes the vehicle control process, comprising: a second in-vehicle calculator for correcting the vehicle control process based on personal driver model, and a first in-vehicle computing unit and the second vehicle computing unit is configured separately The vehicle control process includes a plurality of control processes, and the plurality of control processes include a vehicle travel safety control process involving any one or more of automatic accelerator control, automatic brake control, and automatic steering control. 2 The vehicle-mounted computing unit is characterized in that the vehicle travel safety control process is not corrected based on the individual driver model .

According to the present invention configured as described above, the vehicle includes the first in-vehicle operation unit that learns the individual driver model in the vehicle and the second in-vehicle operation unit that executes the vehicle control process. Learning and execution of the vehicle control process can be performed by separate calculation units. For this reason, it is prevented that the vehicle control process is delayed due to the learning of the individual driver model having a high calculation load. In addition, since the first in-vehicle operation unit and the second in-vehicle operation unit are configured separately, the second in-vehicle operation unit continues to be used, and only the first in-vehicle operation unit having artificial intelligence whose development progress is fast. It will be easy to replace. In addition, according to the present invention, the control (for example, automatic accelerator control, automatic brake control, and automatic steering control) related to running, stopping, and turning is not corrected based on the individual driver model. It is possible to avoid unsafe risks due to poor calculation by intelligence.

In the present invention, it is preferable that the individual driver model includes a driver's emotion estimation model.
In order to create a driver's emotion estimation model, a large amount of driving data indicating the state of the driver and the vehicle is required, which increases the calculation load. Therefore, as in the present invention, the first in-vehicle operation unit that creates the individual driver model including the emotion estimation model is configured separately from the second in-vehicle operation unit, thereby adversely affecting the processing of the second in-vehicle operation unit. Can be prevented.

  Specifically, the vehicle travel safety control process includes a risk avoidance control process for preventing a collision with an obstacle or a deviation from the travel path.

  ADVANTAGE OF THE INVENTION According to this invention, the vehicle driving assistance system which can prevent the delay of vehicle control can be provided, creating a personal driver model in a vehicle.

1 is a configuration diagram of a vehicle driving support system according to an embodiment of the present invention. It is a functional block diagram of each component of the vehicle driving assistance system by embodiment of this invention. It is a functional block diagram in the vehicle control block in the vehicle by embodiment of this invention. It is explanatory drawing of the flow of data in the shared server by the embodiment of this invention, a personal server, and a vehicle-mounted controller. It is explanatory drawing of operation | movement of the synchronous engine by embodiment of this invention. It is explanatory drawing of the parameter update process by embodiment of this invention. It is explanatory drawing of the control recommendation process by embodiment of this invention. It is a block diagram of the vehicle driving system which concerns on the modification of embodiment of this invention.

  Hereinafter, a vehicle driving support system according to an embodiment of the present invention will be described with reference to the accompanying drawings. First, the configuration of the vehicle driving support system will be described with reference to FIGS. FIG. 1 is a configuration diagram of a vehicle driving support system, FIG. 2 is a functional block diagram of each component of the vehicle driving support system, and FIG. 3 is a functional block diagram in a vehicle control block in the vehicle.

As shown in FIG. 1, the vehicle driving support system S includes a shared server 1, a personal server 3 in the vehicle A, and an in-vehicle controller (ECU) 5 in the vehicle A. These are communicably connected to each other via a wireless or wired communication line N.
The shared server 1 and the personal server 3 are computer systems that constitute artificial intelligence, and learn and update the general driver model and the personal driver model, respectively. The shared server 1 includes a calculation unit 1a, a storage unit 1b, a communication unit 1c, and the like. Similarly, the personal server 3 includes a calculation unit 3a, a storage unit 3b, a communication unit 3c, and the like.

  Similarly, the in-vehicle controller 5 includes a calculation unit 5a (processor), a storage unit 5b, a communication unit 5c, and the like. The in-vehicle controller 5 executes a vehicle control process for the vehicle A. The vehicle control process includes a driving support process in addition to the driving control process.

  As shown in FIG. 2, in the shared server 1, a learning engine 11 (computing unit) composed of artificial intelligence receives driving data received from a plurality of personal servers 3 and general data received from an external information system (information providing service center or the like) 7a. As a result of learning in 1a), the general driver model Ma is constructed. The general driver model Ma is sequentially updated as new driving data and general data are learned. The operation data and general data are stored as storage data Da in the storage unit 1b.

  The general data is, for example, human (driver) voice data, behavior data, image data, and the like. The general data is mainly used to construct an emotion estimation model that constitutes a part of the general driver model Ma. Note that the update of the general driver model Ma requires learning of big data including driving data from a plurality of personal servers 3 and general data from an external information source. For this reason, the processing speed (update interval) of the general driver model Ma is very slow (for example, 1 hour or more).

  The general driver model Ma is a model for application to a general driver rather than a specific driver driving the vehicle A. Therefore, driving data for a plurality of drivers other than the specific driver and emotion estimation data (general data) representing general human characteristics are used. The general driver model Ma includes a plurality of submodels. The learning engine 11 learns changes in the behavior and state of a general driver from driving data and general data for each given submodel theme, or creates a submodel. And update. The sub-model includes a driver behavior tendency model in various situations, a driving judgment reference model, an emotion estimation model, and the like. Furthermore, the general driver model Ma has general knowledge data. General knowledge data is various general information acquired from driving data or general data, such as cautionary points for specific road locations, entertainment location information (restaurant information, etc.).

  An example of a sub model of the general driver model Ma is shown. A voice model for general driver's emotions (feeling emotional, especially when feeling fun). Fun state model (relationship model between various states (driver, surrounding environment, vehicle) and fun). A sub-model related to the occurrence of a sidewalk or snoozing that is generated based on map data, driver facial expression data, etc. (for example, a specific point or a driver state (for example, travel time) that is likely to occur is specified). A driving operation characteristic model (for example, an obstacle avoidance action start position) generated based on travel history data, operation history data, and the like.

  When learning the accelerator and brake operation models, the learning engine 11 uses data relating to the driver, the driving location, the surrounding environment, the time zone, the accelerator opening, the brake sensor data, and the like included in the driving data as a group of data. Thus, a model is created by learning how the accelerator and the brake are operated according to the position, speed, number, etc. of the vehicle ahead and the pedestrian.

  In addition, when learning the smile determination model included in the emotion estimation model, the learning engine 11 analyzes voice data of a general driver or a general person and image data related to the voice data, and the general driver feels fun. Analyze the characteristics of facial expressions when you are. As a result, a smile determination model is generated and updated that indicates the relationship between the change in the characteristic portion extracted from the appearance (that is, facial expression) (mouth angle, eye corner angle, etc.) and smile. By using the smile determination model, it is possible to estimate whether or not the driver is smiling (or whether or not he / she feels fun) from the information on the characteristic part (such as the angle of the mouth corner). The characteristic part may be a part designated in advance, or a part newly detected by the learning engine 11.

  Further, the general driver model Ma includes a relationship model (emotion occurrence model) between the vehicle state and the driver's emotional state. The emotional state is analyzed from the driver state data. The learning engine 11 analyzes the transition of the emotional state and determines the vehicle state (motion state: vehicle speed, lateral acceleration, longitudinal acceleration, etc., on-vehicle equipment operating state: air conditioning temperature, seat position, music, etc.) that affected the emotional state. Analyze and learn about the relationship between the vehicle state and the emotional state as to what kind of vehicle state affects the emotional state. The vehicle state (control factor) that affects the emotional state may be set in advance, or may be additionally set by artificial intelligence finding a new control factor through analysis.

  The driver state data used for the analysis of the emotion state is voice data, image data, brain wave data, and the like. For emotion analysis, for example, voice data (uttered voice) is analyzed, and frequency analysis of sound waves based on involuntary movement of the vocal cords is performed. In addition, facial expression analysis based on image data and facial color analysis for blood flow change are performed. In addition, a sympathetic / parasympathetic ratio analysis of the autonomic nervous system is performed. By using one or more of these analyses, for example, it is possible to specify an emotional state on an emotion map or a Russell ring model that expresses emotions on coordinates. The learning engine 11 analyzes changes in emotional state (that is, movement on an emotion map or a ring model) and changes in vehicle state.

  For example, when learning the relationship model between the temperature environment and the emotional state, the learning engine 11 includes data relating to the driver, emotional state, travel location, time zone, inside / outside temperature, weather, etc. included in the driving data. Are used as a group of data, for example, the influence of temperature environment such as the difference between the temperature inside the vehicle and the temperature outside the vehicle, the weather, etc. on the emotional state is learned and the relevance model is updated.

  In addition, when new control factors that are not included in the existing relevance model are learned to change the emotional state (for example, when a plurality of items of the vehicle state are combined) A new relevance model based on the control factors is generated. In this way, the learning engine 11 constructs a model by detecting control factors that affect the driver's emotional state from the big data.

  In the personal server 3 provided in the vehicle A, driving data (including voice data) received from the vehicle A, general data acquired from the external information system 7b, communication data acquired from the portable information terminal device 7c of the specific driver ( The personal driver model Mb is constructed by learning the call voice data, mail text data, device setting information, etc.) by the learning engine 31 (calculation unit 3a) made of artificial intelligence. This personal driver model Mb is also updated sequentially. The personal server 3 learns temporal changes or history of the driver's behavior, vehicle behavior, vehicle performance, etc. using the driving data and the like. For this reason, the processing speed of the individual driver model Mb is lower than the various control processes in the vehicle A (for example, 1 second or more).

  The general data acquired by the personal server 3 includes voice data, behavior data, and the like of a plurality of drivers included in a driver group (for example, a group of drivers having vehicles of the same vehicle type) considered to have a common driving tendency. Image data. The device setting information is, for example, bookmark information registered in the Internet browser application of the portable information terminal device.

The voice data acquired by the microphone of the vehicle A is also included in the driving data acquired via the second synchronization engine 60 of the in-vehicle controller 5, but directly to the personal server 3 in real time via the communication device. Is output. In the personal server 3, this voice data is voice-recognized. Operation data, general data, and voice data are stored as storage data Db in the storage unit 3b.
Further, the first synchronization engine 40 of the personal server 3 performs data conversion on the accumulated data stored in the storage unit 3 b and transmits it to the shared server 1.

  The personal driver model Mb is a model for applying to a specific driver. Therefore, the driving data of the specific driver who drives the vehicle A and general data of other drivers considered to have driving characteristics relatively close to the specific driver are used. Similarly to the general driver model Ma, the individual driver model Mb also includes a plurality of submodels. The personal driver model Mb has surrounding environment state data and vehicle state data extracted from the acquired driving data. The learning engine 31 also learns a plurality of submodels similar to those of the learning engine 11 (for an example of a submodel of the individual driver model Mb, refer to an example of a submodel of the general driver model Ma). The learning engine 31 detects a control factor that affects the emotional state of the driver as in the learning engine 11, updates the model, and constructs a new model.

  In the in-vehicle controller 5, a predetermined vehicle control process is executed by the vehicle control block (calculation unit) 51 based on the sensor data of the vehicle sensor 8. The vehicle control block 51 controls various in-vehicle devices and systems of the vehicle A on a rule basis using a vehicle control algorithm (vehicle control program) 50 that defines a vehicle control process. In other words, various controls are executed according to a predetermined rule (algorithm) based on the sensor data (rule-based process). For this reason, in the vehicle control process by the vehicle-mounted controller 5, a high processing speed is achieved (for example, 10 milliseconds or less).

The vehicle control process includes a driving support process in addition to the driving control process. The driving support processing includes automatic driving support processing, support information presentation processing, and in-vehicle device control processing.
In the automatic driving support process, a command signal is output to the vehicle control system 9d (engine, brake, steering), and the accelerator, brake, and steering driving device are automatically operated.
In the support information presentation process, various support information for assisting the driving operation of the driver is provided via the information presentation device 9a (navigation device, meter, speaker, etc.) in the vehicle, and the information communication device 9c ( Information is provided to an external information system, an information terminal device, a home appliance, etc. via an in-vehicle communication unit, a portable information terminal device, and the like.
In the in-vehicle device control process, the in-vehicle device 9b (air conditioner, window, light, door, etc.) is automatically activated to improve the driving environment. For example, the temperature setting and on / off of the air conditioner are automatically performed, and the opening and closing of the window is automatically performed.

  The vehicle sensor 8 includes an in-vehicle camera, a biological sensor, a microphone, an outside camera, a radar, a navigation device, a vehicle behavior sensor, a driver operation detection sensor, a vehicle-to-vehicle communication device, a vehicle-infrastructure communication device, a remote controller, and the like. It is.

The in-vehicle camera images a driver and other occupants in the vehicle A and outputs in-vehicle image data.
The biometric sensor measures a driver's heartbeat, pulse, sweating, brain waves, and the like, and outputs biometric data.
The microphone collects the voices of the driver and other passengers and outputs voice data.
The vehicle exterior camera captures images of the front, side, and rear of the vehicle A and outputs vehicle exterior image data.

Radar emits radio waves, sound waves, or laser light toward the front, side, and rear of vehicle A, and objects outside the vehicle around vehicle A (preceding vehicles, other vehicles, pedestrians, ground fixed objects, obstacles, etc.) The reflected wave from the vehicle is received, and object data outside the vehicle such as the relative position and relative speed of the object (for example, the position of the preceding vehicle, the relative speed of the preceding vehicle, etc.) is output.
The navigation device acquires vehicle position information and combines navigation data (multiple route information, selected by the driver) with internal map information, traffic congestion information acquired from the outside, and input information (destination, waypoint, etc.). Output route information).

  Vehicle behavior sensors and driver operation detection sensors include speed sensors, longitudinal acceleration sensors, lateral acceleration sensors, yaw rate sensors, accelerator opening sensors, engine speed sensors, AT gear position sensors, brake switch sensors, brake oil pressure sensors, steering An angle sensor, a steering torque sensor, a winker switch position sensor, a wiper switch position sensor, a light switch position sensor, a vehicle interior / external temperature sensor, and the like are included.

  Vehicle-to-vehicle communication device, vehicle-infrastructure communication device, and remote controller acquire communication data from other vehicles, traffic data from traffic infrastructure (traffic jam information, speed limit information, etc.), and remote operation data from outside. And output these.

  Output data from the vehicle sensor 8 is input to the vehicle control block 51 as driving data. Further, the output data is converted into operation data representing various physical quantities suitable for execution of processing in the vehicle control block 51 by a predetermined device (not shown) or by a data processing block in the in-vehicle controller 5. Is input to the vehicle control block 51. By data conversion, one output data is converted into operation data representing one or a plurality of information. Data conversion includes not converting output data.

  For example, the outside image data of the outside camera is converted into position data of a preceding vehicle or lane, offset (deviation) data from a reference line (lane center or set route), and the like. Further, the steering angle data of the steering angle sensor is converted into fluctuation data (steering angle fluctuation data; fluctuation width, fluctuation period, etc.) and the like. In addition, the image data of the in-vehicle camera includes personal identification data (results of driver authentication based on driver images registered in advance and personal data specifying the authenticated driver), smile determination of the driver, etc. Data is converted to facial expression data (mouth angle, eye corner angle, etc.).

The driving data is various data related to the driver, the surrounding environment, and the vehicle, and includes the driver state data, the surrounding environment data, and the vehicle state data. Each of these data consists of a plurality of individual data.
The driver state data is data indicating the physical state of the driver, and includes in-vehicle image data (including driver image data), voice data, biological data (including heartbeat data), and the like.
The ambient environment data is data indicating the status of other vehicles around the vehicle A, pedestrians, obstacles, road objects, traffic objects, and other external objects. Image data outside the vehicle, object data outside the vehicle, navigation data, vehicle Includes inter-vehicle data, inter-vehicle infrastructure data, etc.

The vehicle state data is data representing the vehicle movement state and the operating state of the in-vehicle device, the measurement data by the vehicle behavior sensor, the driver operation data indicating the switch position of the in-vehicle device by the driver operation detection sensor, and the personal identification data. including. Specifically, the vehicle state data includes, for example, vehicle speed, longitudinal acceleration, lateral acceleration, yaw rate, accelerator opening, engine speed, AT gear position, brake switch position, brake hydraulic pressure, front inter-vehicle distance, preceding vehicle and Relative speed, steering angle, steering torque, turn signal switch position, wiper switch position, light switch position, inside / outside temperature, personal identification information, and the like.
The second synchronization engine 60 of the in-vehicle controller 5 converts the operation data once stored in the storage unit 5 b and transmits it to the personal server 3.

As shown in FIG. 3, the vehicle control block 51 includes a current state analysis block 51a, an ideal state analysis block 51b, a difference calculation block 51c, an entertainment (entertainment) control block 52a, and a safety control block 52b.
Operation data is input to the current state analysis block 51a and the ideal state analysis block 51b. In the current state analysis block 51a, the current driver state, the current device operation state, the current surrounding environment state, and the current vehicle state are extracted from the driving data. On the other hand, in the ideal state analysis block 51b, based on the vehicle control model (ideal model) defined by a large number of control parameters P, the ideal driver state, ideal device operation state, ideal peripheral environment state, ideal vehicle state are determined from the driving data. Is calculated.

  The driver state is specified by, for example, the driver's heartbeat data, wobbling analysis data, and the like. The ambient environment state is specified by camera image data, radar measurement data, and the like, for example. The vehicle state is specified from, for example, lateral acceleration data, engine power data, brake wear amount data, and the like.

In the difference calculation block 51c, differences are calculated for various items of the current state and the ideal state (driver state, device operation state, surrounding environment state, vehicle state) output from the current state analysis block 51a and the ideal state analysis block 51b. And output as difference data.
The entertainment control block 52a and the safety control block 52b execute various processes based on the difference data.

  The safety control block 52b controls safety control processing that accompanies the operation of the vehicle control system 9d, and also controls support information presentation processing that accompanies the operations of the information presentation device 9a, the in-vehicle device 9b, and the information communication device 9c. On the other hand, the entertainment control block 52a controls an entertainment control process involving the operation of the information presentation device 9a, the in-vehicle device 9b, and the information communication device 9c, but does not execute a control process involving the operation of the vehicle control system 9d.

  The entertainment control block 52a and the safety control block 52b output an operation instruction to the information presentation device 9a, the in-vehicle device 9b, and the information communication device 9c based on the difference data. The safety control block 52b also outputs an operation instruction to the vehicle control system 9d. Note that data transmitted to the external information system 7b via the information communication device 9c is stored in the information system 7b and can be provided to the personal server 3.

  For example, assume that the current state analysis block 51a analyzes that the driver is driving the vehicle A at 60 km in a normal state with a high consciousness level. On the other hand, the ideal state analysis block 51b calculates (estimates) a planned travel route (including position and speed) when the driver turns a curve 30 meters ahead in this state based on the ideal model of the vehicle control algorithm 50. In addition, the current state analysis block 51a outputs the travel route on which the vehicle A actually traveled as an analysis result in order to continue the state analysis.

  The difference calculation block 51c calculates a difference between the planned travel route by the ideal state analysis block 51b and the actual travel route by the current state analysis block 51a. For example, the safety control block 52b does not perform special processing when the planned speed and the actual speed are substantially equal, and performs processing for generating a brake operation alarm when the speed difference between the planned speed and the actual speed is small. If the speed difference is large, a process for operating the automatic brake is executed.

In addition, a message that prompts the steering operation timing to be advanced is displayed when the actual steering operation timing is later than a predetermined time than the steering operation timing defined by the planned travel route.
Also, in a predetermined situation, when the predicted heart rate predicted by the ideal model is larger than the actual heart rate by a predetermined value or more (estimated as an excited state), the entertainment control block 52a plays a message that prompts a break and music that calms the mood. A process for displaying a message prompting the user is executed.

  Next, the first synchronization engine 40 of the personal server 3 and the second synchronization engine 60 of the in-vehicle controller 5 will be described with reference to FIGS. 4 and 5. FIG. 4 is an explanatory diagram of the data flow in the shared server, personal server, and in-vehicle controller, and FIG.

  As shown in FIG. 4, the in-vehicle controller 5 executes recognition, determination, and action determination on a rule basis based on driving data (information amount “medium”). For this reason, the information speed targeted by the in-vehicle controller 5 is fast (<10 milliseconds). Furthermore, when the vehicle control system 9d, the vehicle-mounted device 9b, and the like receive an operation command (information amount “small”) from the vehicle-mounted controller 5, the vehicle control system 9d operates according to the action determination command. For this reason, the information speed is very fast (<1 ms).

  On the other hand, the personal server 3 learns and grows based on driving data from the in-vehicle controller 5 and data (information amount “large”) from the external information system 7b and the like. For this reason, the information speed in the personal server 3 is slow (> 1 second). Furthermore, the shared server 1 learns and grows based on driving data from a plurality of personal servers 3 and big data (information amount “maximum”) from the external information system 7a and the like. For this reason, the information speed in the shared server 1 is very slow (> 1 hour). In other words, the upper layer has a larger amount of information to be handled than the lower layer, but the information speed is slower. The shared server 1 is the highest and the in-vehicle controller 5 is the lowest.

  For this reason, in this embodiment, information entropy is made uniform so that data processing is smoothly performed in each layer (that is, the flow of data processing does not deteriorate in any layer). Schematically, the processing load (the total number of steps of the processing program) of the processing executed in each layer and the target response time when executing all the processing steps are known, and the data processed per unit time in each layer By adjusting the amount, the processing load at each moment (per unit time) is made uniform.

In this embodiment, the information entropy is defined as “data amount per unit time × processing speed”. The processing speed is defined by “the total number of steps of the processing program (total number of program lines) × target response time”.
Information entropy = amount of data x total number of steps x target response time

  For example, in the in-vehicle controller 5, when the data amount is 10 MB, the number of steps is 1000 rows, and the target response time is 10 milliseconds, the personal server 3 is set to 100 KB, 10000 rows, 1 second, respectively, and the shared server 1 is 10 B respectively. , 100,000 lines, 1000 seconds.

  In order to adjust the amount of data in each layer in this way, when operating data is transmitted from the lower layer to the upper layer, the data conversion of the operating data is performed by the lower layer synchronization engine so that the upper layer can easily process the operating data. Is done. By this data conversion, the operation data is converted in terms of quantity, quality and time. The in-vehicle controller 5 has a second synchronization engine 60, and the personal server 3 has a first synchronization engine 40.

  As shown in FIG. 5, the in-vehicle controller 5 sequentially receives driving data based on the output data of the vehicle sensor 8 and executes a vehicle control process. On the other hand, the second synchronization engine 60 executes a second data conversion process (quantity, quality, time) on the driving data, and transmits the driving data subjected to the data conversion to the personal server 3. In the personal server 3, the received driving data is accumulated in the storage unit 3b as driver's behavior history data and state history data, and used for a predetermined process. Then, the first synchronization engine 40 performs a first data conversion process (quantity, quality, time) on the received operation data, and transmits the operation data subjected to the data conversion to the shared server 1. In the shared server 1, the received driving data is accumulated in the storage unit 1 b as driver behavior history data and state history data and used for predetermined processing.

  The upper layer synchronization request blocks 21 and 41 obtain acquisition requests for requesting transmission of operation data having a necessary information attribute to the lower layer synchronization engines 40 and 60 according to a method required in the upper layer processing. Issue a command. Upon receiving this command, the lower-layer synchronization engine executes data conversion processing according to the required information attribute. The lower layer synchronization engine issues a data conversion instruction to other data processing blocks (not shown) in the same lower layer, converts the data, and outputs the operation data after the data conversion to the upper layer. The lower layer monitors the upper layer. The synchronization request blocks 21 and 41 are, for example, for the first synchronization engine 40 and the second synchronization engine 60, respectively, a data amount reduction method, individual data association and separation method (specification of a plurality of individual data), time axis setting Outputs an acquisition request command that specifies the method (extraction method, statistical processing method).

  In the data conversion process related to the amount of data, the amount of operation data is reduced. For example, data amount reduction processing is performed by feature amount extraction, information amount conversion, and the like. In the first synchronization engine 40 and the second synchronization engine 60, a first data amount reduction process and a second data amount reduction process are executed, respectively.

In the feature quantity extraction, the data size is reduced so as to include the minimum necessary information in the upper layer processing. For example, the image data is converted into feature value data extracted from the image data (mouth angle, lane information, etc.).
In the information amount conversion, the operation data is converted into summary statistics (average, time axis filter, etc.). For example, a deviation amount from the center line (deviation amount data every 10 milliseconds) is converted into average deviation amount data at intervals of 100 seconds. Further, the steering angle data for every 10 milliseconds is converted into the fluctuation degree determination data in units of 5 seconds.
The feature engine extraction and information amount conversion can be executed by another processing block by the synchronization engine.

  In the data conversion process related to the data quality, a data relevance conversion process is performed to convert the relevance between pieces of information included in the operation data. In the first synchronization engine 40 and the second synchronization engine 60, a first data relevance conversion process and a second data relevance conversion process are executed, respectively.

  In relevance conversion, a plurality of individual data are selectively linked. For example, personal identification data and heartbeat data, time data and heartbeat data, position data and heartbeat data are associated with each other. Individual identification data, heart rate data, time data, and position data may be associated with one. The association makes it possible to process the association data as integral data in a certain target process, thereby reducing the processing in the upper layer. For example, mouth angle data (feature data obtained by data amount reduction processing), voice data, driving operation data, and in-vehicle environment data (air conditioning, sound, etc.) can be associated for learning a smile determination model.

  In the conversion of relevance, specific information is deleted from a plurality of pieces of associated information. For example, the personal identification data is separated. In the personal server 3, composite data in which specific individual data and personal identification data are associated is used. However, in the shared server 1, the personal identification data is separated from the composite data in order to ensure anonymity of the composite data. Composite data is used. In addition, when the personal identification data includes first and last names, age, sex, address, etc., only specific items (first name, last name, gender) may be separated.

  In the data conversion process related to the time of data, a time axis change process for performing time axis processing of operation data is performed. In the first synchronization engine 40 and the second synchronization engine 60, a first time axis change process and a second time axis change process are executed, respectively.

  In the time axis processing, predetermined time change data is selectively extracted (sampled) on the time axis. For example, when the data quality is the same, information is thinned out in the time axis direction. For example, heartbeat data at 10 msec intervals is thinned out to heartbeat data at 100 msec intervals. Also, for example, when heart rate data is used to detect arrhythmia, only heart rate data showing a significant numerical value (exceeding a predetermined threshold) is selectively extracted by time axis processing. In the time axis processing, the operation data is converted into summary statistics (averaged, time axis filter) or converted into statistical information (for example, frequency distribution) by statistical processing. In time axis processing, selective extraction time intervals (constant or indefinite) and statistical processing time intervals on the time axis are set according to the update processing time (target response time) of the general driver model Ma and the individual driver model Mb. Is done. Therefore, the longer the target response time, the longer the data interval of operation data output by time axis machining.

  In this embodiment, the amount of data is adjusted between layers (shared server 1, personal server 3, in-vehicle controller 5). However, the present invention is not limited to this, and functional blocks in each layer (for example, personal server 3 The amount of data may be similarly adjusted among the learning engine 31, the parameter update engine 32, the recommendation engine 33, the difference analysis engine 34, and the result verification engine 35).

Next, parameter update processing will be described with reference to FIG. FIG. 6 is an explanatory diagram of parameter update processing. The personal server 3 has a parameter update engine 32.
The parameter update engine 32 acquires the general driver model Ma from the shared server 1, acquires various control parameters P and driving data (including voice data and vehicle state data) that define vehicle control processing from the in-vehicle controller 5, The control parameter P is updated with reference to the personal driver model Mb.

  In principle, the parameter update engine 32 determines the update of the individual driver model Mb performed by the learning engine 31 and updates the vehicle control algorithm 50 related to the update portion in accordance with this update. Specifically, the control parameter P (including the value of the control parameter and the type of the control parameter) included in the vehicle control algorithm 50 is changed.

  For this reason, the parameter update engine 32 compares the personal driver model Mb before update with the latest personal driver model Mb, and extracts an updated portion. Then, the parameter update engine 32 extracts the control parameter P corresponding to the updated portion from the various control parameters P acquired from the vehicle A. Further, the parameter update engine 32 acquires the driver model parameter of the personal driver model Mb corresponding to the control parameter P.

  Then, the acquired driver model parameter is compared with the corresponding control parameter P (difference analysis). If the driver model parameter is related to the control parameter P but does not correspond directly, the driver model parameter is converted so that they correspond directly, and the converted value and the control parameter P are compared.

  As a result of the difference analysis, when the difference exceeds a threshold set according to the type of the control parameter P, the driver model parameter (or conversion value) is set as the update parameter. Further, the parameter update engine 32 determines whether or not a predetermined update condition is satisfied. When the update condition is satisfied, the parameter update engine 32 outputs a control parameter update command in order to update the control parameter P to the update parameter. When receiving the control parameter update command, the in-vehicle controller 5 updates the control parameter P corresponding to the command to a new update parameter.

  In this embodiment, update contents and update time are defined as predetermined update conditions. Regarding the update contents, if the control parameter P to be updated is a control parameter P of vehicle control processing (vehicle driving safety control processing) related to driving safety related to running, stopping, and turning, updating is prohibited. Is done. If the control parameter P of the control process related to vehicle safety is changed, there is a risk of giving the driver a sense of incongruity during the driving operation. The vehicle travel safety control process is specifically a control process involving automatic accelerator control, automatic brake control, and automatic steering control. For example, the vehicle travel safety control process includes a risk avoidance control process for preventing a collision with an obstacle or a deviation from a travel path. Further, the wobble determination process is also included in the vehicle control process related to traveling safety.

  If the update content is updatable, the parameter update engine 32 determines the update time (when stopped, when the IG is off) from the operation data (vehicle state data), and if the update time condition is satisfied, the control parameter Send an update command. In this embodiment, the parameter update engine 32 determines the update condition. However, the in-vehicle controller 5 that has received the control parameter update command may determine the update condition.

  In addition, the update time is defined according to the update content. The update time includes immediately (when the individual driver model Mb is updated), when the vehicle is stopped, and when the ignition is off (when the IG is off). The update time of the control parameter P that is allowed to be changed during traveling is set to “immediate”. Examples of “immediately” are, for example, a smile determination parameter (mouth angle), an air conditioner set temperature, accident information, and the like in the smile determination process.

In addition, the update time of the control parameter that is appropriately updated when the vehicle is stopped is set to “when the vehicle is stopped”. Examples of “when the vehicle is stopped” are, for example, a deadman determination parameter (such as an angle of the driver's body in the driver image data), a vehicle seat position, and a mirror angle in the deadman determination process.
Further, the update time of the control parameter that is appropriately updated when the IG is off is set to “when the IG is off”. An example of “when IG is off” is, for example, general map information.

  In addition, when the updated individual driver model Mb generates a new submodel, or when the learning engine 31 is more effective in response to a predetermined process in a submodel than the existing submodel. If it is determined, a new control parameter P may be added corresponding to these submodels. For example, as a result of learning, it is assumed that the corner of the corner of the eye is more effective for determining the smile of the driver than the angle of the mouth, and a new smile determination model based on the angle of the corner of the corner is generated. In this case, the control parameter P is set instead of or in addition to the existing submodel based on the mouth angle. Specifically, the type of the control parameter P used in the smile determination logic in the smile determination process included in the vehicle control process is changed from the angle of the mouth corner to the angle of the corner of the eye, and the value of the control parameter P is determined from the angle threshold value of the mouth corner. It is changed to the angle threshold of the corner of the eye.

  In addition, an example will be described in which predetermined processing (changing the driving route proposal order, steering vibration, speaker volume increase, etc.) is executed when it is determined that the driver is drowsy in the vehicle control processing. The learning engine 31 learns a sub model based on the magnitude of the steering wobbling angle as a drowsiness determination model. Correspondingly, in the drowsiness determination process in the vehicle A, it is determined that the drowsiness is large when the steering wobbling angle (variation range) exceeds the determination threshold. When the learning engine 31 learns that the vehicle A usually has a large wobble angle, the sub-model is updated, and accordingly, the determination threshold (control parameter value) is updated to a larger value.

On the other hand, when the learning engine 31 learns that the fluctuation period is more effective for determining drowsiness than the magnitude of the steering wobbling angle, a sub model based on the fluctuation period of the steering wobbling angle is used as a drowsiness determination model. to add. Accordingly, in the drowsiness determination process in the vehicle A, the type of the control parameter is changed to the fluctuation cycle of the wandering angle, and the value of the control parameter (determination threshold; fluctuation cycle) is also changed.
When a sleepiness determination submodel based on image data is added, the type of control parameter in vehicle A is changed to a certain feature amount in the image data, and the value of the control parameter (determination threshold) is also changed. .

  Also, a process when the general driver model Ma is considered in the parameter update process will be described. That is, if the specific driver of the vehicle A repeats extreme driving operations different from normal, the personal driver model Mb and the vehicle control process (control parameter P) may be updated so that their safety is reduced. is there. For this reason, when the individual driver model Mb greatly deviates from the general driver model Ma, the control parameter P is updated based on the general driver model Ma in order to ensure safety.

  The parameter update engine 32 acquires the general driver model Ma and the control parameter P. Further, when the personal driver model Mb is updated, the updated part is extracted. And the control parameter P corresponding to this update part is acquired. Furthermore, the personal driver model parameter of the personal driver model Mb and the general driver model parameter of the general driver model Ma in this updated part (or corresponding to the acquired control parameter P) are acquired.

  Next, the parameter update engine 32 compares the acquired personal driver model parameter with the general driver model parameter, and calculates a difference. When the difference is equal to or smaller than the predetermined value, an update parameter for updating the control parameter P is calculated based on the personal driver model parameter. On the other hand, when the difference is larger than the predetermined value, an update parameter for updating the control parameter P is calculated based on the general driver model parameter. The calculation of the update parameter is the same as in the above embodiment.

  When the update parameter is calculated in this manner, a control parameter update command for updating the control parameter P to the update parameter is output when a predetermined update condition is satisfied, as in the above-described embodiment. .

  In the parameter update process, a process when updated based on the general driver model Ma according to a predetermined condition after being updated based on the individual driver model Mb will be described. That is, when the control parameter P is updated by the individual driver model Mb, but no improvement is recognized in the vehicle control process, the updated control parameter P is updated again based on the general driver model Ma.

  The predetermined condition for re-update is whether or not the driver's emotional state has improved before and after the update. If the driver's emotional state does not improve, the control parameter P is updated again. Therefore, the parameter update engine 32 analyzes the driver's emotional state from the driver's emotional analysis data based on the driving data (voice data).

  The emotion analysis may be sequentially executed by the parameter update engine 32, or may be continuously executed by another function block (driver state analysis unit) and stored as an emotion analysis history. The improvement of the emotional state means that the emotional state changes from an unpleasant negative emotion (sadness, disgust, anger, anxiety, tension, dissatisfaction, etc.) to a pleasant positive emotion (joy, fun, relief, relaxation, satisfaction, etc.). Is to move.

  The parameter update engine 32 acquires the control parameter P. Further, when the personal driver model Mb is updated, the updated part is extracted. And the control parameter P corresponding to this update part is extracted. Furthermore, the personal driver model parameter of the personal driver model Mb in this updated part (or corresponding to the acquired control parameter P) is acquired. Then, an update parameter for updating the control parameter P is calculated based on the personal driver model parameter.

  When a predetermined update condition is satisfied, a control parameter update command for updating the control parameter P to the update parameter is output. Further, when receiving the control parameter update command, the in-vehicle controller 5 updates the control parameter P corresponding thereto to a new update parameter.

  The parameter update engine 32 determines whether or not the driver's emotional state has improved before and after the update of the control parameter P resulting from the update of the individual driver model Mb. If it is determined that the driver's emotional state has improved, the control parameter P update process is terminated. On the other hand, when it is determined that the driver's emotional state has not improved, the parameter update engine 32 acquires the general driver model parameter of the general driver model Ma corresponding to the control parameter P.

  Then, a new update parameter for updating the control parameter P is calculated based on the general driver model parameter. When a predetermined update condition is satisfied, a control parameter update command for updating the control parameter P to a new update parameter is output. Further, when receiving the control parameter update command, the in-vehicle controller 5 updates the control parameter P corresponding thereto to a new update parameter.

  For example, regarding a sub model of the steering operation timing when traveling on a curved road, the vehicle travels on a curved road after the corresponding control parameter P of the vehicle A (such as steering operation guidance timing in driving assistance) is updated based on the individual driver model Mb. When the stress (based on heartbeat, voice analysis, etc.) at the time of the update is not lower than before the update, the same control parameter P is updated based on the general driver model Ma.

Next, the recommendation process (control recommendation process) will be described with reference to FIG. FIG. 7 is an explanatory diagram of the control recommendation process. The personal server 3 includes a recommendation engine (vehicle control recommendation unit) 33.
The recommendation engine 33 uses the general driver model Ma acquired from the shared server 1, the driving data (including voice data) acquired from the vehicle A, and the personal driver model Mb to instruct or suggest execution of the recommended process to the in-vehicle controller 5. To do. The in-vehicle controller 5 executes a recommended process when a predetermined condition is satisfied.

The recommendation engine 33 includes a state analysis block 33a and a recommendation block 33b.
The state analysis block 33a analyzes the driver state, the surrounding environment state, and the vehicle state based on the driving data (including voice data), the general driver model Ma, and the individual driver model Mb. Analysis includes analysis of the current state and analysis of the state in the near future (eg, after 30 minutes and 1 hour).

  The recommendation block 33b derives and outputs an optimal action (recommendation process) suitable for the driver based on the analysis output of the state analysis block 33a, the driving data, the general driver model Ma, and the individual driver model Mb. Note that the recommendation engine 33 performs an advanced state analysis using a large amount of accumulated data, so that the recommendation engine 33 operates even while the driver is away from the vehicle A, and derives a recommendation process as appropriate.

  The driver state includes the state of mind, body and behavior of the driver. The mind (emotion) state includes attention state, arousal level, emotion, stress level, driving load, driving motivation, impression, tension level, context, and the like. The body (body) state includes fatigue level, health condition, thermal sensation, device visibility, device operability, riding comfort, sitting comfort, body information, and the like. The action state includes gaze position / object, attention state, gesture, device operation, driving action / operation / posture, dialogue, habit, living action, action intention, and the like.

  The heart state (especially emotional state) may be analyzed directly from voice data (for example, emotion analysis using an endocrine model), driver image data, heartbeat data, or other driving data (driver imaging data). , Including the heartbeat data) using the individual driver model Mb.

The surrounding environment state is an environment around the vehicle A, and includes a traffic / driving environment, risk prior grasp (congestion, road surface freezing, etc.), communication environment, and the like.
The vehicle state is a traveling state of the vehicle A, and includes a driving difficulty level, a wobble, and the like.

The recommendation block performs at least vehicle room space recommendation, travel recommendation, and information presentation recommendation as recommended control.
The vehicle interior space recommendation is a recommended control for providing a vehicle interior environment suitable for the driver, and includes provision of a seat / mirror position / angle, air conditioning, music, a welcome effect, and the like.
The travel recommendation is recommended control for providing a travel route suitable for the driver, and includes presentation of a recommended route, a pleasant route, a challenge route with high driving difficulty, a danger avoidance route, and the like.

  The information presentation recommendation includes presentation of information useful for the driver and presentation of advanced state estimation results at an appropriate timing and an appropriate presentation method. To present useful information to the driver, information such as highlight information on the route (scenery, restaurants, scenic spots, etc.), road traffic / weather / news, alerts (forgotten things, prevention of late arrival), ToDo list, memory images, etc. Is included. The presentation of advanced state estimation results includes presentation of advanced deadman determination and advanced smile determination information.

The recommendation block 33b can derive appropriate recommendation control using a basic association table that describes the relationship between the analysis state by the state analysis block 33a and a predetermined recommendation process. Further, the related table can be learned and updated by the individual driver model Ma and the general driver model Mb.
The in-vehicle controller 5 outputs a command to the information presentation device 9a, the in-vehicle device 9b, the information communication device 9c, and the vehicle control system 9d according to the recommendation signal. For this reason, the vehicle-mounted controller 5 may memorize | store the processing program performed when receiving each recommendation signal.

  In this embodiment, for example, according to each driver state (fatigue level, emotion, stress, attention state, alertness level, etc.) analyzed by the state analysis block 33a, the recommendation block is considered in consideration of the surrounding environment state and the vehicle state. 33b can be realized and an appropriate recommendation process is derived and a recommendation signal is output.

  For example, the state analysis block 33a estimates the body state. Here, it is assumed that it is analyzed that the driver feels fatigue. In response to this, the recommendation block 33b selects and outputs an appropriate recommendation signal according to the situation from the passenger compartment space recommendation signal, the travel route recommendation signal, and the information presentation recommendation signal.

  The vehicle space recommendation signal is, for example, a process for operating an air conditioner, a process for resetting (decreasing) the air conditioning temperature, a process for outputting a predetermined music channel broadcast from a speaker, a process for changing a music channel, and a fatigue level. This is a signal for instructing processing to change the sheet position and the mirror angle. The travel route recommendation signal is a signal for instructing, for example, processing for prompting a change from a currently set route having a high degree of difficulty (eg, a lot of curves) to a new route having a low degree of difficulty (a lot of straight roads). The information presentation recommendation signal is a signal for instructing a display process for displaying a predetermined message for recommending a break or speed reduction on the display screen. Further, when a predetermined body state and a predetermined heart (emotion) state are analyzed, an appropriate recommendation signal is selected.

  The personal driver model Mb includes a sub model that represents the driver's preference (air conditioning temperature, broadcast channel, seat position, mirror angle, etc.) for the setting of the in-vehicle device. For example, when the driver starts driving or when a predetermined body state or a predetermined mind (emotion) state is analyzed, the state analysis block 33a sets the in-vehicle device setting to the driver's preference. If it is analyzed that it is different from the above, the recommendation block 33b outputs a vehicle interior space recommendation signal instructing to change the setting to a desired setting. In this instruction, an air conditioning temperature, a broadcast channel, a seat position, a mirror angle, and the like are designated as operating parameter values (preferred setting values).

  An example of information presentation recommendation will be described. It is assumed that the state analysis block 33a refers to the personal driver model Mb and analyzes that the vehicle A is approaching a specific place that easily induces a sidewalk or drowsiness (peripheral environment state). In response to this, the recommendation block 33b outputs a recommendation signal so as to execute an alerting process by voice or the like (at an appropriate timing) at a point a predetermined distance before the specific location.

  Furthermore, when the state analysis block 33a refers to the general driver model Ma and analyzes that the specific place is a place where a general driver easily induces a sidewalk or sleepiness, the recommendation block 33b further includes the specific place. A recommendation signal is output so that an early warning process is executed at an earlier point (at an appropriate timing).

  The personal driver model Mb is a driver's preference (food, hobbies, sports, etc.) constructed based on communication data (call voice data, mail text data, device setting information, etc.) acquired from the portable information terminal device 7c. ) Is included. For example, in a favorite sub-model of “food”, a favorite dish type (Japanese cuisine, French cuisine, Italian cuisine, etc.) is ranked based on a bookmark or the like that is device setting information.

  Based on this sub-model, when the state analysis block 33a analyzes the expected time when the driver starts to feel hungry, the recommendation block 33b adjusts to the expected time (appropriate timing), and the restaurant of the favorite food type. A recommendation signal for displaying information (cooking type, restaurant name) on the navigation map is output. The recommendation signal may be selected not only when the hunger level is analyzed but also when a predetermined body state is analyzed or when a predetermined heart (emotion) state is analyzed.

  An example of a travel route recommendation will be described. The state analysis block 33a estimates the driver's emotional state or body state from the driving data with reference to the individual driver model Mb and the general driver model Ma. Here, it is assumed that the driver feels that the enjoyment is low (or predicts that the driver will feel bored within one hour with a decrease in driving motivation).

  Receiving the analysis output representing this estimation (prediction), the recommendation block 33b derives a recommendation process for feeling fun (or causing no boredom). For example, on the basis of the personal driver model Mb or the general driver model Ma, a place (a shoreline location) registered as a place where a general driver or a driver of the vehicle A feels pleasure within a predetermined distance range from the current position. Roads, view spots, etc.) are searched, and a recommendation signal that prompts a change to a route that uses this place as a transit point is output. When receiving the recommendation signal, the in-vehicle controller 5 inputs the waypoint included in the recommendation signal to the navigation device. Thereby, a new recommended route is displayed on the display screen by the navigation device performing a new route calculation.

  When the condition analysis block 33a estimates road surface freezing from the outside air temperature, weather, etc., the recommendation block 33b outputs a recommendation signal that indicates a route that avoids a place where road surface freezing is estimated.

  The in-vehicle controller 5 includes a recommendation determination block 53 that determines whether or not to execute a recommended process based on the recommendation signal. The recommendation determination block 53 permits the execution when the recommended process is the entertainment control process, and rejects the execution when the recommended process is the safety control process. In the in-vehicle controller 5, when the execution is permitted by the recommendation determination block 53, the entertainment control block 52a executes a recommendation process according to the recommendation signal.

  In the present embodiment, whether or not the recommended process can be executed is determined depending on whether the recommended process is the entertainment control process or the safety control process. However, the present invention is not limited to this, and even if the recommended process is an entertainment control process, the execution of the recommended process may be rejected when there is a possibility that the vehicle operation becomes unsafe. For example, if the recommended process is a process that delays the alarm generation timing when the brake operation timing is delayed, it will be rejected due to increased insecurity, but if it is a process that accelerates the alarm generation timing, On the other hand, it is allowed because it increases safety.

  The recommendation determination block 53 also rejects the execution of the recommended process even when the recommended process is a process that contradicts or invalidates the vehicle control process by the in-vehicle controller 5. For example, in the vehicle control process, a process for advancing the alarm generation timing for the brake operation delay due to the weather (rainy weather or when visibility is reduced) is executed, but the recommended process is a process for delaying the alarm generation timing. Will be rejected because it will contradict or invalidate the above-mentioned processing caused by the weather, but if it is a processing that accelerates the alarm timing, it will not conflict with the above-mentioned processing caused by the weather and will not be invalidated. Is done.

  Next, the personal driver model supplement process by the general driver model will be described. Since the vehicle A is limited to traveling in a specific region or a specific traveling operation, the reference data (driving data, general data) used for the individual driver model Mb is much larger than the reference data used for the general driver model Ma. Very few. For this reason, a difference occurs between the individual driver model Mb and the general driver model Ma according to such a difference in data amount.

  A comparison block (not shown) in the recommendation engine 33 acquires the individual driver model Mb and the general driver model Ma, and performs comparison processing. By this comparison processing, sub-models and general knowledge that are present in the general driver model Ma but are not present in the individual driver model Mb are extracted. The extracted difference data (submodel, general knowledge, etc.) is stored as accumulated data via the result verification engine 35 and is learned by the learning engine 31. As a result, a submodel, general knowledge, etc. that can be adapted to the vehicle A are added to the individual driver model Mb.

Next, the recommendation instruction verification process will be described with reference to FIG. The personal server 3 further includes an ideal driver model Mi, a difference analysis engine 34, and a result verification engine 35.
The ideal driver model Mi is created based on the driving operation of an expert driver, and the driver enjoys while concentrating on the driving operation in a state where the driving skill and driving difficulty of the driver are balanced. Is an ideal state model.
The difference analysis engine 34 compares the driver state in the ideal driver model Mi with the actual driver state analyzed based on the driver's voice data.

  The result verification engine 35 analyzes the difference data from the difference analysis engine 34 to verify how the driver state is affected by the execution of the recommendation process based on the recommendation signal, and stores the verification result in the storage unit. Accumulate in 3b. The verification result is an evaluation of how much the difference is reduced (approaching the ideal state) or how much the difference is increased (distant from the ideal state) by executing the recommendation process. When the learning engine 31 learns the verification result, the personal driver model Mb is updated so as to be more suitable for the specific driver.

Next, a vehicle driving support system according to a modification will be described with reference to FIG. FIG. 8 is a configuration diagram of a vehicle driving system according to a modified example.
In the vehicle driving support system S2 according to the modified example, the personal server 3 is mounted on the vehicle A as in the vehicle driving support system S of the above embodiment. That is, in-vehicle controller 5 and personal server 3 are mounted on vehicle A so that they can communicate. The data flow is the same as that of the vehicle driving support system S.

  In the vehicle driving support system S2, the in-vehicle controller 5 and the personal server 3 are connected in a divided state via a gateway having a security function. For this reason, the in-vehicle controller 5 and the personal server 3 are configured as separate units.

  When the in-vehicle controller 5 and the personal server 3 are configured as an integrated unit, there is a risk that the vehicle control processing may be delayed due to a temporary shortage of computing capacity for the learning process of the personal driver model that requires advanced processing. There is. However, in the present modified example, since these are configured as separate units, the in-vehicle controller 5 only needs to execute the vehicle control process, as in the above-described embodiment, and therefore a delay in the vehicle control process occurs. Absent.

  Artificial intelligence technology is currently being developed, and progress is rapid. For this reason, in this modification, the personal server 3 configured separately can be easily updated to a higher performance server.

Next, the operation of the vehicle driving support system of this embodiment will be described.
The vehicle driving support system according to the present embodiment includes an in-vehicle personal server 3 (first in-vehicle computing unit) having artificial intelligence that learns a personal driver model Mb unique to the driver based on driving data of the driver, A vehicle controller 5 (second in-vehicle operation unit) that executes the vehicle control process and corrects the vehicle control process based on the individual driver model Mb, and the individual server 3 and the in-vehicle controller 5 are configured separately. ing.

  Thereby, in this embodiment, since it has the personal server 3 which learns the personal driver model Mb in a vehicle, and the vehicle-mounted controller 5 which performs a vehicle control process, learning of the personal driver model Mb and vehicle control process are carried out. Execution and execution can be performed by separate arithmetic units. For this reason, it is prevented that the vehicle control process is delayed due to the learning of the individual driver model Mb having a high calculation load. In addition, since the personal server 3 and the in-vehicle controller 5 are configured separately, the in-vehicle controller 5 can be easily exchanged only for the personal server 3 having artificial intelligence whose development progress is fast while continuing to use. .

  In the present embodiment, the individual driver model Mb includes a driver's emotion estimation model. In order to create a driver's emotion estimation model, a large amount of driving data indicating the state of the driver and the vehicle is required, which increases the calculation load. Therefore, as in this embodiment, the personal server 3 that creates the personal driver model Mb including the emotion estimation model is configured separately from the in-vehicle controller 5, thereby preventing adverse effects on the processing of the in-vehicle controller 5. Can do.

  In the present embodiment, the vehicle control process includes a plurality of control processes, and the plurality of control processes includes a vehicle travel safety control accompanied by any one or more of automatic accelerator control, automatic brake control, and automatic steering control. The process is included, and the in-vehicle controller 5 does not correct the vehicle travel safety control process based on the personal driver model Mb. Thereby, in this embodiment, about control (for example, automatic accelerator control, automatic brake control, and automatic steering control) about run, a stop, and turning, by not performing correction based on personal driver model Mb, The risk of poor calculation due to artificial intelligence can be avoided. Specifically, the vehicle travel safety control process includes a risk avoidance control process for preventing a collision with an obstacle or a deviation from the travel path.

DESCRIPTION OF SYMBOLS 1 Shared server 3 Personal server 5 In-vehicle controller 8 Vehicle sensor 9a Information presentation apparatus 9b In-vehicle apparatus 9c Information communication apparatus 9d Vehicle control system 11 Learning engine 31 Learning engine 32 Parameter update engine 33 Recommendation engine 33a State analysis block 33b Recommendation block 34 Difference analysis Engine 35 Result verification engine 40 First synchronization engine 51 Vehicle control block 51a Current state analysis block 51b Ideal state analysis block 51c Difference calculation block 52a Entertainment control block 52b Safety control block 53 Recommendation determination block 60 Second synchronization engine A Vehicle Da, Db Accumulated data Ma General driver model Mb Individual driver model Mi Ideal driver model P Control parameters S, S2 Vehicle driving support system

Claims (3)

A first in-vehicle computing unit having artificial intelligence for learning a personal driver model unique to the driver based on driving data of the driver;
A second vehicle-mounted arithmetic unit that executes a predetermined vehicle control process and corrects the vehicle control process based on the personal driver model,
Wherein the second in-vehicle computing unit and the first vehicle computing unit is configured separately,
The vehicle control process includes a plurality of control processes,
The plurality of control processes include a vehicle travel safety control process involving any one or more of automatic accelerator control, automatic brake control, and automatic steering control,
The second on-vehicle calculation unit is a vehicle driving support system in which the vehicle travel safety control process is not corrected based on the personal driver model .   The vehicle driving support system according to claim 1, wherein the personal driver model includes a driver's emotion estimation model. The vehicle driving support system according to claim 1 , wherein the vehicle travel safety control process includes a risk avoidance control process for preventing a collision with an obstacle or a departure from a travel path.

Images