JP7236319B2 - vehicle control system - Google Patents

Description

The present invention relates to vehicle control systems.

Patent Literature 1 proposes a driving assistance device that models driving behavior characteristics that differ among individual drivers and performs vehicle control and driving assistance according to the characteristics of the driver. can be used to easily generate a driver model for each driver. As described above, it is possible to estimate a driving behavior that is closer to the driver's driving characteristics, and to realize driving assistance that matches the driver's driving characteristics.

JP 2007-176396 A

In the method of Patent Document 1, feature quantities that affect the driver's driving are learned from among multiple feature quantities selected by the system developer. Therefore, it was difficult to select the feature amount itself, and it was difficult to learn the driver's preferences with high accuracy and reflect them in driving support. In addition, in Patent Document 1, since the driver's driving is learned based on the feature amount detected by each sensor, if there is a feature amount that affects the driver's driving but cannot be detected by the current sensor, that feature There was a problem that it was difficult to provide driving support that matched the driver's intentions because the driver model could not consider the amount.

SUMMARY OF THE INVENTION The present invention has been made in view of the above points, and an object of the present invention is to provide a vehicle control system capable of realizing driving assistance according to the driving characteristics of the driver.

In order to solve the above problems, one desirable aspect of the present invention is as follows.
A vehicle control system according to the present invention includes detection data of an external world recognition sensor that recognizes the external world of the own vehicle, and either image data of the external world of the own vehicle or data of reflected waves from the external world of the own vehicle. a vehicle control system that calculates control parameters for controlling the own vehicle using the travel data of the own vehicle, and performs control for supporting the travel of the own vehicle based on the calculated control parameters, the vehicle control system comprising: a learning unit for extracting a feature amount that affects the control parameter from image data or reflected wave data and learning a correlation between the control parameter and the feature amount; a learning model learned by the learning unit; A control parameter calculator that calculates the control parameter using either one of the image data or the reflected wave data and at least one of the travel data and the detection data.

According to the present invention, the feature amount that affects the control parameter is obtained from image data or reflected wave data without depending on the feature amount selected by the developer of the system and without depending on the detection performance of the current sensor. It automatically extracts from the model, automatically learns the correlation between the feature amount and the control parameter, and provides driving support based on the learned model. As a result, it is possible to calculate control parameters that match the driver's intentions according to the driving environment, and to realize driving support that is suitable for the driver's driving characteristics.

Further features related to the present invention will become apparent from the description of the specification and the accompanying drawings. Further, problems, configurations, and effects other than those described above will be clarified by the following description of the embodiments.

1 is a system configuration diagram in the first embodiment; FIG. 1 is a block diagram of a vehicle control system according to a first embodiment; FIG. The figure which shows the relationship between the control parameter and the data for learning in 1st Embodiment. FIG. 4 is a diagram for explaining a driver model according to the first embodiment; FIG. The figure which shows the example of a vehicle control system in 1st Embodiment of operation. The figure explaining the effect of the vehicle control system in 1st Embodiment. The block diagram of the vehicle control system in 2nd Embodiment. The figure which shows the operation example of the vehicle control system in 2nd Embodiment. The figure explaining the effect of the vehicle control system in 3rd Embodiment. The figure explaining the effect of the vehicle control system in 4th Embodiment.

<First embodiment>
A vehicle control system will be described below with reference to the drawings. In this embodiment, the case where the present invention is applied to an automobile will be described as an example, but the present invention can also be applied to buses and trucks.

FIG. 1 is a diagram showing hardware of a vehicle control system 0. As shown in FIG.
In FIG. 1, the FL wheel means the left front wheel, the FR wheel means the right front wheel, the RL wheel means the left rear wheel, and the RR wheel means the right rear wheel. The vehicle control system 0 is based on the information from the external world recognition sensors 2, 3, 4, and 5 that recognize the external world of the own vehicle, the GNSS 27 that detects the absolute position of the own vehicle, and the external world recognition sensors 2, 3, 4, and 5. A first control device 1 that calculates a target trajectory for a plurality of functions such as automatic driving in one's own lane, automatic lane change, and automatic merging, and a first control based on information from external recognition sensors 2, 3, 4, and 5. A second controller 25 that calculates a target trajectory for the autonomous driving function less than the device 1, a steering control mechanism 10, a brake control mechanism 13, a throttle control mechanism 20, an HMI 23, and a first controller 1. A vehicle motion control device 26 that calculates command values for the mechanisms 10, 13, and 20 based on the target trajectory of the second control device 25, and a steering that controls the steering control mechanism 10 based on the command values of the vehicle motion control device 26. A control device 8, a brake control device 15 that controls the brake control mechanism 13 based on the command value to adjust the brake force distribution to each wheel, and a throttle control mechanism 20 based on the command value to adjust the torque output of the engine. and a throttle control device 19 that

As external recognition sensors, a stereo camera 2 is provided in front of the vehicle, laser radars 3 and 4 are provided on the left and right sides of the vehicle, and a millimeter wave radar 5 is provided behind the vehicle. can detect the relative distance and relative velocity. The stereo camera 2 captures an image of the external environment in front of the vehicle and transmits the image data to the first control device 1 and the second control device 25 . The stereo camera 2 can detect preceding vehicles and pedestrians in front of the vehicle. Also, the front stereo camera 2 can detect the lateral position of the lane marker of the lane on which the vehicle is traveling.

In the present embodiment, a case where the stereo camera 2 is used as an example of a sensor that recognizes the external environment in front of the host vehicle among the external environment recognition sensors will be described, but the configuration is not limited to this. For example, a lidar or millimeter wave radar may be used instead of the stereo camera 2, and data of reflected waves from the outside of the vehicle may be input instead of image data. In addition, in the present embodiment, the combination of the sensors described above is shown as an example of the sensor configuration, but the sensor configuration is not limited to the above, and the combination with an ultrasonic sensor, a monocular camera, an infrared camera, etc. It's okay. GNSS27 detects the absolute position of the own vehicle. Information from the external recognition sensors 2 , 3 , 4 , 5 and the GNSS 27 is input to the first control device 1 and the second control device 25 .

The first control device 1 and the second control device 25 have, for example, a CPU, ROM, RAM and input/output devices, which are not shown in detail in FIG. The ROM stores ACC, LKAS, recognition and judgment programs for realizing automatic lane changes and automatic driving at intersections, and generates the target trajectory and target acceleration for the vehicle motion control device 26. , to the vehicle motion control device 26 . The vehicle motion control device 26 calculates command values for each mechanism 10, 13, 20 so that the vehicle follows the target trajectory and target acceleration, and communicates them to the control devices 8, 15, 19 of each mechanism 10, 13, 20. do. The controllers 8, 15, 19 of the mechanisms 10, 13, 20 receive the command value of the first controller 1 by communication, and control each mechanism based on the command value.

Next, the operation of the brake will be explained. The brake boosts the driver's stepping force on the brake pedal 12 with a brake booster (not shown) and generates hydraulic pressure corresponding to the force with a master cylinder (not shown). Hydraulic pressure generated in the master cylinder is supplied to the wheel cylinders 16 via the brake control mechanism 13 . The wheel cylinders 16FL to 16RR are composed of cylinders (not shown), pistons, pads, etc. The pistons are propelled by hydraulic fluid supplied from the master cylinder, and the pads connected to the pistons are pressed against the disk rotor. be. Note that the disk rotor rotates together with the wheel (not shown). Therefore, the braking torque acting on the disc rotor becomes the braking force acting between the wheel and the road surface. As described above, a braking force can be generated at each wheel according to the operation of the brake pedal by the driver.

Although not shown in detail in FIG. 1, the brake control device 15 has, for example, a CPU, a ROM, a RAM and an input/output device like the first control device 1 . The brake control device 15 includes a combine sensor 14 capable of detecting longitudinal acceleration, lateral acceleration, and yaw rate, wheel speed sensors 8FL to 8RR installed on each wheel, a brake force command from the brake control device 15, A sensor signal is input from a steering wheel angle detection device 21 via a steering control device 8, which will be described later. In addition, the output of the brake control device 15 is connected to a brake control mechanism 13 having a pump (not shown) and control valves to generate arbitrary braking force on each wheel independently of the driver's brake pedal operation. be able to. The brake control device 15 estimates spin of the vehicle, drift-out, and locking of the wheels based on the information of the input sensor signals, and generates a braking force for the corresponding wheels so as to suppress them, thereby improving the steering stability of the driver. play a role in enhancing Moreover, the first control device 1 can generate an arbitrary braking force in the vehicle by communicating a brake command to the brake control device 15 . However, the configuration of the brake control device 15 is not limited to the configuration described above, and other mechanisms such as brake-by-wire may be used.

Next, steering operation will be described. A steering torque detector 7 and a steering wheel angle detector 21 detect the steering torque and the steering wheel angle input by the driver through the steering wheel 6, respectively. generate. Although not shown in detail in FIG. 1, the steering control device 8 also has, for example, a CPU, a ROM, a RAM, and an input/output device, like the first control device 1. FIG. The resultant force of the driver's steering torque and the assist torque from the motor 9 activates the steering control mechanism 10, turning the front wheels. On the other hand, the reaction force from the road surface is transmitted to the steering control mechanism 10 according to the steering angle of the front wheels, and is transmitted to the driver as the road surface reaction force.

The steering control device 8 can generate torque by the motor 9 and control the steering control mechanism 10 independently of the driver's steering operation. Therefore, the first control device 1 can control the front wheels to an arbitrary steering angle by communicating the target steering torque to the steering control device 8 . However, the present embodiment is not limited to the above steering control device, and other mechanisms such as steer-by-wire may be used.

Next, the accelerator will be explained. The amount of depression of the accelerator pedal 17 by the driver is detected by a stroke sensor 18 and input to a throttle control device 19 . Although not shown in detail in FIG. 1, the throttle control device 19 also has, for example, a CPU, ROM, RAM, and input/output devices, like the first control device 1 . A throttle control device 19 adjusts the throttle opening according to the depression amount of the accelerator pedal 17 to control the engine. As described above, the vehicle can be accelerated according to the driver's operation of the accelerator pedal. Further, the throttle control device 19 can control the throttle opening independently of the driver's accelerator operation. Therefore, the first control device 1 can cause the vehicle to generate arbitrary acceleration by communicating the target acceleration to the throttle control device 19 .

As described above, the vehicle control system 0 appropriately controls the speed of the vehicle by adjusting the brakes and throttle according to the situation of the surrounding vehicles, and also controls the steering to automatically control ACC, LKAS, automatic lane changes, and intersections. It is possible to realize automatic driving in

FIG. 2 shows functional blocks of the vehicle control system 0. As shown in FIG.
Various detection results of the external recognition sensors 2 , 3 , 4 and 5 and travel data are input to the data selection section 201 and the control parameter calculation section 204 . The detection results of the external world recognition sensors 2, 3, 4, and 5 include detection data of the external world recognition sensors that recognize the external world of the vehicle and image data of the external world of the vehicle. Travel data of the own vehicle is detected by an internal sensor that detects the state of the vehicle, such as the steering wheel angle detection device 21 . Image data from the stereo camera 2 is input to the data selection unit 201 and the control parameter calculation unit 204 .

The vehicle control system 0 has a learning mode and a driving support mode. In the learning mode, the data selection unit 201, the data storage unit 202, and the learning unit 203 execute processing for learning the driving characteristics of the driver. Then, a process of calculating the control parameters using the learning model is executed in the control parameter calculation unit 204 .

In the learning mode, the data selection unit 201 acquires travel data, detection data, and image data at learning timings set based on the control parameters. The image data acquired from the stereo camera 2 in the learning mode is so-called raw data in which no processing such as correction is applied to the data captured by the stereo camera 2 . For example, as shown in FIG. 3, the data selection unit 201 selects learning data necessary for learning from detection data of the external recognition sensors 2, 3, 4, and 5, image data, and travel data. In this embodiment, the target inter-vehicle time of ACC is learned as a control parameter, so only data before and after the upper limit value (for example, 3 seconds) of the target inter-vehicle time is selected as data for learning. By selecting data in this learning mode, it is possible to exclude data unsuitable for learning and use only data suitable for learning data for learning.

Data selected by the data selection unit 201 is stored in the data storage unit 202 . The learning unit 203 calculates control parameters based on learning data acquired during learning stored in the data storage unit 202 . Then, a feature amount that affects the control parameter is extracted from the image data, which is raw data, and the correlation between the control parameter and the feature amount is learned.

Since the control parameter of this embodiment is the target inter-vehicle time, the learning unit 203 calculates the target inter-vehicle time during learning based on the speed of the own vehicle and the inter-vehicle distance and relative speed to the preceding vehicle. Then, a feature amount that affects the target inter-vehicle time is extracted from the image data, and the correlation between the target inter-vehicle time and the feature amount is learned. The learning unit 203 extracts feature amounts using image data, which is raw sensor data, and at least one of travel data and detection data.

The driver model is composed of a convolutional neural network (CNN) as shown in Fig. 4, and the feature amount that affects the target inter-vehicle time is obtained from image data and detection data from external recognition sensors 2, 3, 4, and 5. and at least one of the travel data, and automatically learn the correlation between the target inter-vehicle time and the feature quantity. However, the driver model is not limited to CNN, and may be Recurrent neural network (RNN) or other models.

The learned driver model is written to the control parameter calculator 204 that is executed in the driving support mode. The control parameter calculation unit 204 inputs at least one of the driving data and the detection data and the image data to the model learned by the learning unit 203, thereby calculating the target inter-vehicle time that matches the driver's preference according to the driving scene. do.

The operating range setting unit 205 limits the operating range of the driving support unit 206 by setting upper and lower limits of the target inter-vehicle time. By limiting the operating range of the driving support unit 206, safety can be ensured even if the control parameters are automatically changed by AI. The driving support unit 206 is equipped with an ACC, which calculates a target longitudinal acceleration based on the target inter-vehicle time, the inter-vehicle distance from a preceding vehicle (not shown), the relative speed, the speed of the own vehicle, and the like. This learning mode may be executed not only during manual driving but also during ACC driving assistance, and may be learned based on the target inter-vehicle time traveled by the driver's switch input or accelerator intervention.

The processing of learning the driver's preferences and calculating the control parameters described above is performed for each driver (drivers A, B, and C) who drives the own vehicle. Therefore, the driver identification unit 208 identifies the driver based on the information of the driver monitor 207, and the data selection unit 201, the data storage unit 202, the learning unit 203, and the control parameter calculation unit 204 are switched based on the driver information.

By doing so, for example, when the driver changes, it is possible to learn driving that suits the driver's preference for each driver, and to implement driving support according to the driver's driving characteristics. The means for identifying the driver is not limited to the driver monitor, and other means such as fingerprint authentication and input by the driver may be used. Also, the unit of learning and support may be divided into groups such as aggressive and eco instead of each driver.

FIG. 5 is a diagram showing a calculation example of the target inter-vehicle time calculated by the driver model learned by the vehicle control system in the first embodiment.
The target inter-vehicle time, which is a control parameter, is calculated by a driver model, as shown in FIG. 5, based on image data, travel data such as speed, and detection results such as the lateral position of the white line. Conventionally, the target inter-vehicle time was constant, as indicated by the broken line in FIG.

FIG. 6 is a diagram for explaining effects of the vehicle control system in the first embodiment.
According to this embodiment, the driver model learned by the vehicle control system 0 can provide the effects shown in FIG. For example, on roads with good visibility, the target inter-vehicle time is set shorter than on roads with poor visibility. When the visibility is good, the target inter-vehicle distance is set shorter than when the visibility is poor. If the type of preceding vehicle is a truck, the target inter-vehicle time is set long, and if the preceding vehicle is a compact vehicle, the target inter-vehicle time is set short. When the load of the own vehicle is large, the target inter-vehicle time is set longer than when the load is small.

These are just examples, and the vehicle control system 0 described in this embodiment can achieve the target inter-vehicle time without depending on the feature amount selected by the system developer and without depending on the detection performance of the current sensor. It is possible to automatically extract influential feature quantities from image data, automatically learn the correlation between the feature quantities and the target inter-vehicle time, and perform driving support based on the learned model. As a result, it is possible to calculate the target inter-vehicle time that meets the driver's intentions according to the driving environment, and to realize driving support that corresponds to the driver's driving characteristics.

In the present embodiment, image data obtained by imaging the external environment of the vehicle with the stereo camera 2 is used as the raw data of the sensor. , data of reflected waves from the outside world of the own vehicle may be used. With such a configuration, the driver model automatically obtains the feature amount that affects the target inter-vehicle time from the reflected wave data and at least one of the detection data of the external recognition sensors 2, 3, 4, and 5 and the driving data. It automatically learns the correlation between the target inter-vehicle time and feature values. Then, the control parameter calculation unit 204 inputs at least one of the driving data and the detection data and the reflected wave data to the driver model learned by the learning unit 203, thereby matching the driver's preference according to the driving scene. Calculate the target inter-vehicle time.

<Second embodiment>
The present embodiment is a vehicle control system that calculates a driver's reliability of the vehicle control system and changes control parameters according to the reliability. Descriptions of portions having the same functions as those of the first embodiment will be omitted.

FIG. 7 shows a functional block diagram of this embodiment. The reliability calculation unit 701 calculates the driver's reliability of the vehicle control system based on operation information such as the cumulative usage time of the vehicle control system for each driver and the frequency of intervention of the brake and the steering wheel. For example, if the cumulative usage time is long, it is determined that the driver has a higher degree of confidence in the vehicle control system than if the cumulative usage time is short. If the frequency of brake or steering intervention is low, it is determined that the driver has a higher degree of reliability in the vehicle control system than if the intervention frequency is high. The reliability may be calculated using information other than the accumulated usage time and the frequency of brake or steering intervention.

The control parameter calculator 204 corrects the target inter-vehicle time based on the reliability. Specifically, when the reliability is high, the target inter-vehicle time calculated by the learned driver model is calculated. Correction is made to the side where the inter-vehicle time is long (the control parameter is set to a large value). As a result, it is possible to provide driving support according to the driver's reliability of the vehicle control system.

FIG. 8 is a diagram showing an operation example of the vehicle control system in this embodiment, and shows a calculation example of the target inter-vehicle time calculated in this embodiment.
A target inter-vehicle time is calculated by a driver model based on image data, travel data such as speed, and detection results such as the lateral position of white lines. By correcting the target inter-vehicle time based on the reliability, when the reliability is low, the target inter-vehicle time is set longer (on the safe side) than when the reliability is high, and when the reliability is high, the driver's preference is set. It is possible to perform driving support suitable for As a result, it is possible to reduce the anxiety of the driver who is unfamiliar with the vehicle control system.

<Third Embodiment>
The vehicle control system of this embodiment uses the amount of offset of the target lateral position of the LKAS system from the lane center line as a control parameter. Description of the same parts as in the first embodiment is omitted.

In this embodiment, the outputs of the learning unit 203 and the control parameter calculation unit 204 in the block diagram of the vehicle control system shown in FIG. 2 are not the target inter-vehicle time but the offset amount of the target lateral position from the lane center line. The driving support unit 206 is LKAS instead of ACC. In this embodiment, the control parameter shown in FIGS. 3, 4, and 5 is not the target inter-vehicle time but the offset amount from the lane centerline.

FIG. 9 is a diagram explaining the effect of the vehicle control system in the third embodiment.
According to this embodiment, it is possible to obtain the effects shown in FIG. For example, in an S-shaped lane, a shortcut path with a smaller radius of curvature than passing through the lane centerline is created. A road with a tight S-curve is set to have a larger offset from the lane center line than a road with a loose S-curve. If there is a large vehicle in the lane next to you, keep more distance than a small vehicle and keep in the lane. When visibility is poor due to rain, fog, etc., the offset from the lane centerline is set smaller than when visibility is good. When the road surface is dry, the offset from the lane centerline is set larger than when the road surface is wet or snowy.

The example shown in FIG. 9 is just an example, and the vehicle control system described in this embodiment does not depend on the feature amount selected by the developer of the system, and does not depend on the detection performance of the current sensor. and automatically extracting from the image data a feature amount that affects the amount of offset of the target lateral position from the lane center line, and the correlation between the feature amount and the amount of offset from the lane center line of the target lateral position. can be automatically learned, and driving support can be provided based on the learned model. As a result, it is possible to calculate the amount of offset from the lane centerline of the target lateral position that matches the driver's intention according to the driving environment, and to realize driving support that corresponds to the driving characteristics of the driver.

<Fourth Embodiment>
The vehicle control system of this embodiment uses the lane change judgment criteria of the automatic lane change system as control parameters. Description of the same parts as in the first embodiment is omitted.

In this embodiment, the outputs of the learning unit 203 and the control parameter calculation unit 204 in the block diagram of the vehicle control system shown in FIG. 2 are not the target inter-vehicle time but the automatic lane change start flag. The driving support unit 206 is an automatic lane change instead of an ACC. The control parameters in FIGS. 3, 4, and 5 are not the target inter-vehicle time, but the threshold for determining the start of lane change (lane change start determination).

FIG. 10 is a diagram for explaining the effect of the vehicle control system in the fourth embodiment.
According to this embodiment, it is possible to obtain the effects shown in FIG. For example, on a road with a gentle curve, the threshold for determining the start of an automatic lane change is set smaller than on a road with a tight curve. That is, when the inter-vehicle distance or relative speed to the vehicle in the next lane is larger than a predetermined value, it is determined that the lane change is to be started, but the predetermined value is set small. In other words, the start of lane change is determined in a state where the vehicle-to-vehicle distance and relative speed are smaller. On roads with poor visibility, the threshold for determining the start of an automatic lane change is set higher than on roads with good visibility. Under conditions of good visibility, such as fog or rain, the threshold for determining the start of automatic lane change is set smaller than under bad conditions. When the road surface is dry, the threshold value for determining the start of automatic lane change is set smaller than when the road surface is wet or snowy.

The example shown in FIG. 10 is only an example, and the vehicle control system described in this embodiment does not depend on the feature amount selected by the system developer, and does not depend on the detection performance of the current sensor. Then, the feature quantity that affects the threshold value for determining the start of lane change is automatically extracted from the image data, the correlation between the feature quantity and the threshold value for determining the start of lane change is automatically learned, and the learned model Based on this, driving support can be performed. As a result, it is possible to calculate a lane change start determination threshold that matches the driver's intention according to the driving environment, and realize driving support that matches the driver's intention.

The vehicle control system 0 in each of the above-described embodiments includes detection data of the external world recognition sensors 2, 3, 4, and 5 that recognize the external world of the own vehicle, image data of the external world of the own vehicle, or data from the external world of the own vehicle. A vehicle control system that calculates control parameters for controlling the own vehicle using either one of the reflected wave data and the own vehicle's running data, and performs control to support the running of the own vehicle based on the calculated control parameters. is. Then, a learning unit that extracts a feature amount that affects the control parameter from the image data or reflected wave data and learns the correlation between the control parameter and the feature amount, a learning model learned by the learning unit, the image data or the A control parameter calculator that calculates a control parameter using any one of the reflected wave data and at least one of the travel data and the detection data.

According to the vehicle control system 0 in each of the above-described embodiments, the feature amount that affects the control parameter is independent of the feature amount selected by the developer of the system and independent of the detection performance of the current sensor. is automatically extracted from image data or reflected wave data. Then, the correlation between the feature amount and the control parameter can be automatically learned, and driving support can be performed based on the learned model. Therefore, it is possible to calculate control parameters that match the driver's intentions according to the driving environment, and to realize driving support that corresponds to the driver's driving characteristics.

Although the embodiments of the present invention have been described in detail above, the present invention is not limited to the above-described embodiments, and various designs can be made without departing from the scope of the invention described in the claims. Changes can be made. For example, the above-described embodiments have been described in detail in order to explain the present invention in an easy-to-understand manner, and are not necessarily limited to those having all the described configurations. Also, part of the configuration of one embodiment can be replaced with the configuration of another embodiment, and the configuration of another embodiment can be added to the configuration of one embodiment. Furthermore, it is possible to add, delete, or replace part of the configuration of each embodiment with another configuration.

0: vehicle control system, 1: first control device, 2: stereo camera (external recognition sensor), 3, 4: laser radar (external recognition sensor), 5: millimeter wave Radar (external recognition sensor) 6 Steering wheel 7 Steering torque detector 8 Steering control device 8FL to 8RR Wheel speed sensor 9 Motor 10 Steering control mechanism 12 Brake pedal 13 Brake control mechanism 14 Combine sensor 15 Brake control device 16FL to 16RR Wheel cylinder 17 Accelerator pedal 18... Stroke sensor 19... Throttle control device 20... Throttle control mechanism 21... Steering wheel angle detection device 23... HMI 25... Second control device 26. Vehicle motion control device 27 GNSS 201 Data selection unit 202 Data storage unit 203 Learning unit 204 Control parameter calculation unit 205 Operation range setting unit, 206... driving support unit, 701... reliability calculation unit.

Claims (7)

Detection data of an external world recognition sensor that recognizes the external world of the own vehicle, either image data of the external world of the own vehicle or data of reflected waves from the external world of the own vehicle, and running data of the own vehicle. A vehicle control system that calculates a control parameter for controlling the own vehicle using
calculating a reference value of a control parameter for the driver using the detection data and either the image data or the reflected wave data acquired while the driver of the vehicle is driving, and the driving data; Extracting a feature quantity that affects the reference value of the control parameter from either the image data or the reflected wave data acquired during the driving , and correlation between the reference value of the control parameter and the feature quantity a learning unit for learning relationships;
a control parameter calculation unit that calculates the control parameter using the learning model learned by the learning unit, either one of the image data or the reflected wave data, and at least one of the travel data and the detection data; ,
A vehicle control system comprising: a data selection unit that acquires the travel data, the detection data, and either the image data or the reflected wave data at a learning timing set based on the control parameter;
2. The vehicle control system according to claim 1, wherein the learning section performs the learning using the data selected by the data selection section. 2. The vehicle control system according to claim 1, wherein the control parameter is either a target inter-vehicle time or an offset amount of the target lateral position from the lane centerline. a reliability calculation unit for calculating the reliability of the driver's system based on operation information by the driver of the own vehicle, and correcting the control parameters based on the reliability calculated by the reliability calculation unit; A vehicle control system according to claim 1. 5. The vehicle control according to claim 4 , wherein when the reliability is low, the reliability calculation unit corrects the target inter-vehicle time, which is the control parameter, to a larger value than when the reliability is high. system. The reliability calculation unit is characterized in that, when the reliability is low, the amount of offset from the lane center line of the target lateral position, which is the control parameter, is set to a smaller value than when the reliability is high, and corrected to the safe side. The vehicle control system according to claim 4. 2. The vehicle control system according to claim 1, further comprising an operation range setting unit that sets upper and lower limits of said control parameter to limit an operation range of said driving support unit.

Images