Vehicle braking control device

Abstract

To provide a brake controller of a vehicle capable of designating responsiveness of brake activation according to an external situation of the vehicle.SOLUTION: A brake controller 100 includes an information acquisition unit 101 that acquires outside situation information concerning an external situation of a vehicle 10, and a designation unit 103 that designates responsiveness of brake activation according to an index outputted from a learner 110 in response to input of the outside situation information, which is acquired by the information acquisition unit 101, to the learner 110.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a braking control device for a vehicle. [Background technology]

[0002] Patent Document 1 discloses that, in manual driving by a vehicle driver and automatic driving, the braking distance when there is no preceding vehicle during manual driving is learned, and the learning results are reflected in the driving characteristics of automatic driving. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. 2018 / 163288 Summary of the Invention [Problem to be solved by the invention]

[0004] However, since external conditions outside the vehicle are diverse, there is room for improvement in the control of the vehicle braking system in accordance with the external conditions. An object of the present invention is to provide a vehicle braking control device that can set an appropriate braking action responsiveness in accordance with the diverse external conditions. [Means for solving the problem]

[0005] A vehicle braking control device for solving the above problem is a device applied to a braking device that applies braking force to wheels of a vehicle, and includes: an information acquisition unit that acquires vehicle exterior situation information, which is information about a situation outside the vehicle; and a setting unit that sets a responsiveness of the braking operation in accordance with an index output from a learning device that has performed machine learning to estimate a probability of a braking operation that applies braking force to the wheels in the braking device based on the situation outside the vehicle, by inputting the vehicle exterior situation information acquired by the information acquisition unit into the learning device.

[0006] As described above, external conditions vary widely. Therefore, it is not easy to calculate the probability by conditional branching using a large number of parameters that define the various external conditions as input. In this regard, the above configuration makes it possible to calculate the probability corresponding to the external condition information acquired by the vehicle by inputting the information to a learning device. Therefore, the responsiveness of the braking operation can be suitably set in accordance with the various external conditions. [Brief explanation of the drawings]

[0007] [Figure 1] FIG. 1 is a schematic diagram of a vehicle according to the first embodiment. [Figure 2] FIG. 2 is a partial cross-sectional view showing a schematic configuration of a friction brake mounted on the vehicle. [Figure 3] FIG. 3 is a flowchart illustrating the flow of control relating to the pre-braking process executed by the braking control device mounted on the vehicle. [Figure 4] FIG. 4 is a schematic diagram showing a machine learning method according to the first embodiment. [Figure 5] FIG. 5 is a flowchart showing a control flow of the learning process executed by the learning device according to the first embodiment. [Figure 6] FIG. 6 is a schematic diagram showing a vehicle and an external server device according to the second embodiment. [Figure 7] FIG. 7 is a sequence diagram showing the flow of pre-braking processing executed in the vehicle and the external server device of the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0008] (First embodiment) The first embodiment will be described below with reference to FIGS. FIG. 1 is a diagram showing a vehicle 10 according to this embodiment, a braking device 30 mounted on the vehicle 10, and a braking control device 100 that controls the braking device 30. As shown in FIG.

[0009] <Brake device> The braking device 30 has a friction brake 20. The friction brake 20 is a braking mechanism that applies a braking force to the corresponding wheel 11.

[0010] The friction brake 20 is, for example, a caliper-type braking mechanism. The friction brake 20 has a rotor 21 as a frictioned part that rotates integrally with the wheel 11, and a brake pad 22 as a friction part.

[0011] 2 is a partial cross-sectional view showing the friction brake 20. The friction brake 20 is configured so that the brake pad 22 can be displaced in an approaching direction and a moving away direction relative to the rotor 21. The approaching direction is a direction in which the brake pad 22 is moved relatively closer to the rotor 21. The moving away direction is a direction in which the brake pad 22 is moved relatively away from the rotor 21.

[0012] As shown by the two-dot chain line in Figure 2, when the friction brake 20 is not applying a braking force to the wheel 11, the brake pad 22 is separated from the rotor 21. On the other hand, when the friction brake 20 is applying a braking force to the wheel 11, as shown by the solid line in Figure 2, the brake pad 22 is in contact with the rotor 21.

[0013] The friction brake 20 includes a wheel cylinder 23. For example, when the driver of the vehicle 10 operates the brake pedal 15, brake fluid is supplied into the wheel cylinder 23, and the WC pressure, which is the fluid pressure in the wheel cylinder 23, increases. This causes the piston 25 of the wheel cylinder 23 to move, and the brake pad 22 approaches the rotor 21 as shown by the white arrow in FIG. 2. When the brake pad 22 comes into contact with the rotor 21, a braking force is applied to the wheel 11.

[0014] 1, the braking device 30 has an actuator (ACT) 31. The braking device 30 is configured to be able to adjust the WC pressure by the operation of the actuator 31. Therefore, the braking device 30 can apply a braking force to the wheels 11 by the operation of the actuator 31 even when the brake pedal 15 is not operated.

[0015] <Sensor system> The sensor system of the vehicle 10 includes, for example, a wheel speed sensor 51, a longitudinal acceleration sensor 52, a yaw rate sensor 53, and a brake switch 56. The wheel speed sensor 51 detects the wheel speed WS, which is the rotational speed of the wheels 11, and outputs a detection signal corresponding to the wheel speed WS to the braking control device 100. The longitudinal acceleration sensor 52 detects the longitudinal acceleration Gx of the vehicle 10 and outputs a detection signal corresponding to the longitudinal acceleration Gx to the braking control device 100. The yaw rate sensor 53 detects the yaw rate Yr of the vehicle 10 and outputs a detection signal corresponding to the yaw rate Yr to the braking control device 100. The brake switch 56 outputs a signal indicating whether the brake pedal 15 is being operated to the braking control device 100.

[0016] <Exterior vehicle monitoring system> The vehicle 10 is equipped with an exterior monitoring system 60 that monitors the situation outside the vehicle 10. The exterior monitoring system 60 has, for example, an imaging device 61 and a radar device 62. The imaging device 61 captures images of the outside of the vehicle 10. The radar device 62 detects, for example, the distance between the vehicle 10 and other vehicles, pedestrians, and obstacles located around the vehicle 10. The exterior monitoring system 60 outputs image information, such as information on images captured by the imaging device 61, and radar information, which is information detected by the radar device 62, to the braking control device 100.

[0017] The braking control device 100 acquires map information from a navigation device (NAV) 70. The map information is information relating to the current position of the vehicle 10. The navigation device 70 may be an on-board navigation device provided in the vehicle 10, or may be a portable terminal owned by the driver of the vehicle 10.

[0018] <Brake control device> The braking control device 100 includes a CPU, which is a calculation unit, and a storage unit. The storage unit includes a ROM. The ROM stores various control programs executed by the CPU and a learned model LM that constitutes the learning device 110. In other words, the braking control device 100 includes the learning device 110.

[0019] The braking control device 100 functions as an information acquisition unit 101, an index acquisition unit 102, a setting unit 103, and a braking processing unit 104 by the CPU executing a control program stored in the storage unit.

[0020] The information acquisition unit 101 acquires outside-vehicle situation information, which is information relating to the outside of the vehicle 10. For example, the information acquisition unit 101 acquires imaging information, radar information, and map information as the outside-vehicle situation information. The information acquisition unit 101 may also acquire information that can be understood from the imaging information, radar information, and map information as the outside-vehicle situation information.

[0021] The index acquiring unit 102 acquires an index IND indicating the probability of a braking operation based on the vehicle exterior situation information acquired by the information acquiring unit 101. In detail, the index acquiring unit 102 inputs the vehicle exterior situation information to the learning device 110 and acquires the value output from the learning device 110 as the index IND.

[0022] Here, the probability of a braking action refers to the probability that a braking action will occur in the braking device 30. The probability of a braking action can also be said to be the possibility that a braking request will be made to the braking device 30, which is a request to apply braking force to the wheels 11. In this embodiment, a braking action is considered to have occurred in the braking device 30 both when a braking force is applied to the wheels 11 due to operation of the brake pedal 15 by the driver and when a braking force is applied to the wheels 11 due to operation of the actuator 31 in a situation where the brake pedal 15 is not being operated.

[0023] The probability of braking is correlated with conditions outside the vehicle. For example, the more other vehicles and pedestrians there are around the vehicle 10, the closer the vehicle is to these vehicles and pedestrians, and the narrower the road on which the vehicle 10 is traveling, the higher the probability of braking. The more traffic signals, intersections, and curves there are on the road on which the vehicle 10 is traveling, and the closer the traffic signals, intersections, and curves are to these traffic signals, intersections, and curves, the higher the probability of braking. When the vehicle 10 is traveling on an ordinary road, the probability of braking tends to be higher compared to when the vehicle 10 is traveling on a highway. When the vehicle 10 is traveling in an urban area, the probability of braking tends to be higher compared to when the vehicle 10 is traveling in a suburban area. When the vehicle 10 is traveling on a downhill road, the probability of braking tends to be higher compared to when the vehicle 10 is traveling on an uphill road.

[0024] The learning device 110 is constructed by a trained model LM that has undergone machine learning to estimate the probability of a braking operation based on the vehicle exterior situation. For example, the trained model LM is a forward propagation type neural network. When vehicle exterior situation information is input, the trained model LM outputs a value indicating the probability of a braking operation according to the vehicle exterior situation information as an index IND. For example, the index IND has a larger value as the probability of a braking operation increases. In other words, the trained model LM maps the vehicle exterior situation to the probability of a braking operation. A method for generating the trained model LM will be described later.

[0025] The setting unit 103 sets the responsiveness of a braking operation according to the index IND acquired by the index acquisition unit 102. For example, when the likelihood of a braking operation indicated by the index IND is equal to or greater than a predetermined threshold, the setting unit 103 executes a process to cause the friction brake 20 to prepare for a braking operation. This preparation is referred to as "braking preparation," and this process is referred to as "braking preparation processing." On the other hand, when the likelihood of a braking operation indicated by the index IND is less than the predetermined threshold, the setting unit 103 does not execute the braking preparation processing.

[0026] An example of the braking preparation process will be described with reference to Fig. 2. The setting unit 103 executes a pre-braking process as the braking preparation process. The pre-braking process is a process in which the actuator 31 is operated to move the brake pads 22 of the friction brakes 20 relatively closer to the rotor 21. By executing this pre-braking process, it is possible to shorten the free running time, which is the time from when a braking request is issued to the braking device 30 to when a braking force corresponding to the braking request is actually applied to the wheels 11.

[0027] In the pre-braking process, the brake pads 22 may not be brought into contact with the rotor 21, or may be brought into contact with the rotor 21. However, when the brake pads 22 are brought into contact with the rotor 21, the braking force applied to the wheel 11 is assumed to be very small.

[0028] The braking processing unit 104 controls the actuator 31 of the braking device 30 when generating a braking force on the vehicle 10. That is, the braking processing unit 104 operates the actuator 31 to adjust the braking force applied to the wheels 11.

[0029] <Control related to pre-braking processing> 3 is a flowchart showing the flow of control relating to the pre-braking process. This control is executed by the braking control device 100.

[0030] In the control shown in FIG. 3, in step S11, the braking control device 100 functions as the information acquisition unit 101 to acquire image information output by the vehicle exterior monitoring system 60. In step S13, the braking control device 100 functions as the information acquisition unit 101 to acquire the radar information output by the vehicle exterior monitoring system 60.

[0031] By analyzing the image information, the braking control device 100 can grasp the situation around the vehicle 10, such as the number of other vehicles and pedestrians traveling around the vehicle 10, the distance between the vehicle 10 and the other vehicles and pedestrians, whether there are traffic signals around the vehicle 10, and the width of the road on which the vehicle 10 is traveling.By analyzing the radar information, the braking control device 100 can grasp, for example, the number of other vehicles and pedestrians traveling around the vehicle 10 and the distance between the vehicle 10 and the other vehicles and pedestrians. The braking control device 100 (information acquisition unit 101) may acquire imaging information and radar information as the vehicle exterior situation information, or may acquire information that can be ascertained from imaging information and radar information as the vehicle exterior situation information.

[0032] In step S15, the braking control device 100 functions as the information acquisition unit 101 to acquire the map information output by the navigation device . Possible map information includes, for example, information about the area in which the vehicle 10 is traveling and information about the road on which the vehicle 10 is traveling. For example, the information about the area in which the vehicle 10 is traveling may include information indicating whether the vehicle 10 is traveling in an urban area or a suburban area. Possible information about the road on which the vehicle 10 is traveling may include information indicating whether the vehicle 10 is traveling on a road with relatively many traffic lights, intersections, and curves, or a road with relatively few traffic lights, intersections, and curves, information indicating whether the vehicle 10 is traveling on an ordinary road or an expressway, and information indicating whether the vehicle 10 is traveling on a downhill road or an uphill road. The braking control device 100 (information acquisition unit 101) acquires at least one of map information and information that can be grasped from the map information as vehicle exterior situation information.

[0033] In step S17, the braking control device 100 functions as the information acquisition unit 101 to acquire, as sensor information, detection values ​​of the various sensors 51 to 53 provided on the vehicle 10, such as at least one of the wheel speed WS, the longitudinal acceleration Gx, and the yaw rate Yr. The braking control device 100 can also acquire, as sensor information, the behavior of the vehicle 10 that can be understood from these detection values. The behavior of the vehicle 10 here may include the traveling speed, acceleration, and turning state of the vehicle 10.

[0034] In step S19, the braking control device 100 functions as the index acquisition unit 102 to input the acquired outside-vehicle situation information and sensor information to the learning device 110. In step S21, the braking control device 100 functions as the index acquisition unit 102 to acquire the value output from the learning device 110 as the index IND.

[0035] In step S23, the braking control device 100 determines whether the index IND is equal to or greater than the index determination value INDTh by functioning as the setting unit 103. The index determination value INDTh is a threshold value for evaluating the likelihood of a braking action indicated by the index IND.

[0036] If the index IND is equal to or greater than the index determination value INDTh (S23: YES), the braking control device 100 proceeds to the process of step S25. On the other hand, if the index IND is less than the index determination value INDTh (S23: NO), the braking control device 100 proceeds to the process of step S27.

[0037] In step S25, the braking control device 100 executes the above-described pre-braking process by functioning as the setting unit 103. As a result, the brake pads 22 of the friction brake 20 approach the rotor 21. This increases the responsiveness of the braking operation of the braking device 30. Thereafter, the braking control device 100 temporarily ends the control shown in FIG. 3.

[0038] In step S27, the braking control device 100 functions as the setting unit 103 to terminate the execution of the pre-braking process. For example, the setting unit 103 stops the driving of the actuator 31 of the braking device 30. As a result, in the friction brake 20, the brake pad 22 separates from the rotor 21. Thereafter, the braking control device 100 temporarily terminates the control shown in FIG. 3.

[0039] <How to generate a trained model LM> A method for generating the learned model LM that constructs the learning device 110 will be described with reference to FIGS.

[0040] As shown in Fig. 4, the learning device 200 is installed outside the vehicle 10. The learning device 200 performs machine learning to estimate the probability of a braking operation based on learning data LD acquired from the vehicle 10. The result of this machine learning is a learned model LM.

[0041] The learning device 200 includes a communication unit 201, a storage unit 202, and a calculation unit 203. A learning program LP is stored in the storage unit 202 of the learning device 200. On the other hand, the vehicle 10 includes a braking control device 100 and a communication unit 120.

[0042] The calculation unit 203 of the learning device 200 executes the learning program LP. As a result, the calculation unit 203 acquires learning data LD from the braking control devices 100 of the multiple vehicles 10 via the communication unit 120, the network 300, and the communication unit 201. The calculation unit 203 stores the acquired learning data LD in the storage unit 202. Then, the calculation unit 203 performs the above-mentioned machine learning using the learning data LD stored in the storage unit 202, and stores a learning result LR, which is a result of the machine learning, in the storage unit 202.

[0043] In this embodiment, the learning data LD includes the following information: Braking external situation information, which is information about the external situation at the time when a braking operation occurs in the vehicle 10, or immediately before or after that. Braking sensor information, which is sensor information at the time when a braking operation occurs in the vehicle 10, or immediately before or after that. Non-braking external situation information, which is external situation information when no braking action is occurring in the vehicle 10. Non-braking sensor information, which is sensor information when no braking action is occurring in the vehicle 10.

[0044] In the following description, the vehicle external situation information during braking and the sensor information during braking will be referred to as "vehicle external situation information during braking, etc." Additionally, the vehicle external situation information during non-braking and the sensor information during non-braking will be referred to as "vehicle external situation information during non-braking, etc." In this case, the learning device 200 can perform supervised learning by using the vehicle external situation information during braking, etc. as correct answer data and the vehicle external situation information during non-braking, etc. as incorrect answer data.

[0045] 5 is a flowchart showing the control flow for the above-mentioned learning process. This control is executed by the calculation unit 203 of the learning device 200. In the following explanation, this control is assumed to be started when the communication unit 201 of the learning device 200 receives learning data LD from the vehicle 10 while a learning completion flag FLG (described later) is set to OFF.

[0046] 5, in step S51, the calculation unit 203 acquires, as learning data LD, information transmitted from the vehicle 10. The information acquired here is information on the external situation of the vehicle when braking or information on the external situation of the vehicle when not braking.

[0047] In step S53, the calculation unit 203 performs machine learning using the learning data LD acquired in step S51. For example, the calculation unit 203 performs machine learning of a neural network. In this case, the calculation unit 203 may provide the configuration of the neural network, the initial values ​​of the connection weights between neurons, and the initial values ​​of the thresholds of each neuron using a template or input by an operator. When re-learning is performed, the calculation unit 203 may create a neural network based on the learning result LR.

[0048] In this case, the calculation unit 203 inputs the learning data LD into the input layer of the neural network and acquires an output value that is a value output from the output layer of the neural network. Then, the calculation unit 203 calculates the error between the acquired output value and a correct value or an incorrect value. Specifically, when the learning data LD is information about the external situation during braking or the like, the calculation unit 203 determines the difference between the output value and the correct value of 1 as the error. On the other hand, when the learning data LD is information about the external situation during non-braking or the like, the calculation unit 203 determines the difference between the output value and the incorrect value of 0 as the error.

[0049] The calculation unit 203 updates the weights of the connections between neurons and the thresholds of the neurons so as to reduce these errors. In this case, the calculation unit 203 can use the well-known back propagation through time method or the stochastic gradient descent method.

[0050] After updating the neural network parameters in this way, the calculation unit 203 proceeds to the process of step S55. In step S55, the calculation unit 203 determines whether or not the machine learning is complete. For example, the calculation unit 203 determines whether or not the number of pieces of data DC in the learning data LD used in the machine learning is equal to or greater than a judgment value DCTh. The judgment value DCTh is a threshold value of the number of pieces of data DC in the learning data LD for determining whether or not the machine learning has been performed sufficiently.

[0051] If the number of data DC is equal to or greater than the judgment value DCTh, the calculation unit 203 determines that machine learning is complete (S55: YES) and proceeds to the process of step S59. On the other hand, if the number of data DC is less than the judgment value DCTh, the calculation unit 203 does not determine that machine learning is complete (S55: NO) and proceeds to the process of step S57.

[0052] In step S57, the calculation unit 203 sets the learning completion flag FLG to OFF, and then the calculation unit 203 temporarily ends this control. In step S59, the calculation unit 203 sets the learning completion flag FLG to ON, and stores the neural network parameters at that time as the learning result LR in the storage unit 202. Thereafter, the calculation unit 203 ends this control.

[0053] In this embodiment, the calculation unit 203 executes the learning program LP upon receiving the learning data LD from the vehicle 10, but this is not limiting. For example, after a predetermined number of pieces of learning data LD are stored in the storage unit 202, machine learning may be performed using the learning data LD as input. Furthermore, preparation of the learning data LD and machine learning may be separate processes.

[0054] A learned model LM is generated through the implementation of the above control. Then, such a learned model LM is provided in the braking control device 100. <Actions and Effects of the Present Embodiment> First, the operation and effect of the method for generating the trained model LM will be explained.

[0055] (1-1) According to this embodiment, the learning device 200 acquires the learning data LD from the vehicle 10 via the network 300. Therefore, the learning device 200 can easily collect the learning data LD. In particular, by acquiring external situation information during braking from the vehicle 10, highly accurate teacher data (correct answer data) can be easily collected.

[0056] (1-2) In this embodiment, the image capture information is used as the learning data LD, so that machine learning can be performed taking into account a wide variety of external surrounding conditions of the vehicle 10. (1-3) In this embodiment, map information is used as learning data LD, so machine learning can be performed taking into account the area in which the vehicle 10 is traveling and the roads on which the vehicle 10 is traveling.

[0057] (1-4) In this embodiment, machine learning is performed using sensor information as learning data LD in addition to vehicle exterior situation information. Here, even if the vehicle exterior situation information is the same, the probability of braking may change depending on the behavior of the vehicle 10. For example, when the vehicle 10 is traveling on a highway, the probability of braking is generally low. However, when the vehicle 10 is traveling at a low speed on a highway, there is a possibility that congestion is occurring in the area where the vehicle 10 is traveling. Therefore, when the vehicle 10 is traveling at a low speed on a highway, the probability of braking is high. Therefore, by performing machine learning using sensor information in addition to vehicle exterior situation information, it is possible to generate a highly accurate trained model LM.

[0058] (1-5) According to this embodiment, by acquiring the learning data LD from multiple vehicles 10 instead of one vehicle 10, it is possible to prevent bias in the data group of the learning data LD due to the habits of a particular driver. In this embodiment, as described above, the learning data LD is acquired from the vehicle 10 via the network 300, so that the learning data LD can be easily acquired from multiple vehicles 10.

[0059] Next, the operation and effects of the braking control device 100 will be described. (1-6) In this embodiment, the learning device 110 is configured with a trained model LM that has undergone machine learning to estimate the probability of braking operation corresponding to the external situation. By inputting external situation information about the vehicle 10 while it is running into the learning device 110, the braking operation responsiveness of the braking device 30 is set according to the value (index IND) output from the learning device 110. This makes it possible to suitably set the braking operation responsiveness according to a wide variety of external situations.

[0060] (Second embodiment) The second embodiment will be described with reference to Figures 6 and 7. The second embodiment differs from the first embodiment in that the learning device 110 is provided in the off-vehicle server device 400. In the following description, differences from the first embodiment will be mainly described, and components that are the same as or equivalent to those in the first embodiment will be assigned the same reference numerals and will not be described again.

[0061] FIG. 6 shows an example of a vehicle control system including a vehicle 10 and an external server device 400. As shown in FIG. <Vehicle> 6, the vehicle 10 includes a braking device 30, a braking control device 100, and a communication unit 120. The braking control device 100 transmits information such as outside-vehicle situation information from the communication unit 120 via a network 300 to an outside-vehicle server device 400.

[0062] <External server device> The external server device 400 includes a calculation unit 401, a storage unit 402, a learning device 110, and a communication unit 403. In the external server device 400, the communication unit 403 receives the external situation information and the like transmitted from the vehicle 10 via the network 300. The external server device 400 inputs the external situation information and the like received by the communication unit 403 to the learning device 110. Then, the external server device 400 transmits index information, which is information related to the index IND, which is a value output from the learning device 110, from the communication unit 403 via the network 300 to the vehicle 10.

[0063] <Control related to pre-braking processing> 7 is a sequence diagram showing the flow of control related to the pre-braking process. The control program related to the pre-braking process on the vehicle 10 side is stored in the memory unit of the braking control device 100 and executed by the calculation unit of the braking control device 100. The control program related to the pre-braking process on the off-vehicle server device 400 side is stored in the memory unit 402 of the off-vehicle server device 400 and executed by the calculation unit 401 of the off-vehicle server device 400.

[0064] The braking control device 100 of the vehicle 10 acquires vehicle exterior situation information in steps S11 to S15. Specifically, the braking control device 100 functions as the information acquisition unit 101 to acquire the imaging information output by the imaging device 61 as vehicle exterior situation information (step S11). The braking control device 100 acquires the radar information output by the radar device 62 as vehicle exterior situation information (step S13). The braking control device 100 acquires map information from the navigation device 70 as vehicle exterior situation information (step S15).

[0065] In step S17, the braking control device 100 functions as the information acquisition unit 101 to acquire sensor information from the sensor system. In step S191, the braking control device 100 functions as the index acquisition unit 102 to transmit the outside-vehicle situation information and the like from the communication unit 120 to the outside-vehicle server device 400.

[0066] When the external server device 400 receives the external situation information transmitted by the vehicle 10 in step S191, it executes the processes of steps S192 to S194. In step S192, the calculation unit 401 of the external server device 400 inputs the external situation information transmitted by the vehicle 10 and the like into the learned model LM of the learning device 110.

[0067] In step S193, the calculation unit 401 acquires the value output from the learned model LM of the learning device 110 as the index IND. In step S194, the calculation unit 401 transmits index information, which is information related to the index IND, from the communication unit 201 to the vehicle 10.

[0068] When the braking control device 100 of the vehicle 10 receives the index information transmitted by the external server device 400 in step S194, it executes the process of step S211. In step S211, the braking control device 100 functions as the index acquisition unit 102 to acquire the index IND indicated by the index information. Then, the braking control device 100 proceeds to the process of step S23. The process flow from step S23 onwards is the same as in the first embodiment, and therefore description thereof will be omitted.

[0069] <Actions and Effects of This Embodiment> According to this embodiment, in addition to the effects equivalent to those of (1-1) to (1-5) above, the following effects can be further obtained.

[0070] (2-1) In this embodiment, the braking control device 100 of the vehicle 10 transmits information about the external situation of the vehicle and the like to the external server device 400. The external server device 400 then inputs the information about the external situation of the vehicle and the like received from the vehicle 10 into the learned model LM, and transmits to the vehicle 10 index information about the index IND, which is a value output from the learned model LM. This causes the braking control device 100 of the vehicle 10 to set the responsiveness of the braking operation according to the index IND indicated by the index information received from the external server device 400. This makes it possible to suitably set the responsiveness of the braking operation according to a wide variety of external situations.

[0071] (2-2) In this embodiment, the learning device 110 is provided in the off-vehicle server device 400, and the learning device 110 is not provided in the braking control device 100. This allows the braking control device 100 to be made smaller.

[0072] (2-3) The learning device 110 is provided in the off-vehicle server device 400. Therefore, if the off-vehicle server device 400 is provided with a function equivalent to the learning device 200 or the off-vehicle server device 400 is connected to the learning device 200, it is possible to provide the indicators IND output from the learning device 110 to the vehicle 10 while re-learning the trained model LM.

[0073] (Example of change) The above-described embodiments can be modified as follows: The above-described embodiments and the following modifications can be combined with each other to the extent that no technical contradiction occurs.

[0074] In the above embodiments, if the index IND is less than the index determination value INDTh, the pre-braking process is not executed. However, this is not limited to this as long as the braking operation responsiveness can be set according to the index IND. For example, multiple index determination values ​​INDTh of different magnitudes may be set, and the distance between the brake pad 22 and the rotor 21 may be shortened in stages as the probability of a braking operation increases. Alternatively, for example, the distance between the brake pad 22 and the rotor 21 may be continuously shortened as the probability of a braking operation increases without setting the index determination value INDTh.

[0075] In the above embodiments, the braking preparation process reduces the amount of actuation of the braking device 30 required from the time a braking request is issued until braking force is applied to the wheels 11. However, this is not limiting. Specifically, the braking preparation process may also be a process that increases the actuation speed of the braking device 30. For example, if the braking device 30 includes an electric pump that supplies brake fluid to the wheel cylinders 23 and a solenoid valve provided in a return flow path connecting the outlet and inlet of the electric pump, the process that increases the responsiveness of the braking operation may be a process that increases the discharge rate of the electric pump while the solenoid valve is open. In this case, by reducing the opening of the solenoid valve in response to the issuance of a braking request, the greater the discharge rate of the electric pump, the more brake fluid can be supplied to the wheel cylinders 23.

[0076] In the above embodiments, the vehicle exterior situation information and the sensor information are used for machine learning, but it is not essential to use the sensor information for machine learning. If only the vehicle exterior situation information is used for machine learning, the index acquisition unit 102 will input only the vehicle exterior situation information out of the vehicle exterior situation information and the sensor information to the learning device 110.

[0077] In the above embodiments, the image information, radar information, and map information are used as the vehicle exterior situation information for machine learning, but it is not necessary to use all of this information for machine learning. In the above embodiments, the vehicle exterior situation information obtained from multiple vehicles 10 is used for machine learning, but the vehicle exterior situation information obtained from a single vehicle 10 (e.g., the subject vehicle) may also be used for machine learning.

[0078] In the above embodiments, machine learning using a neural network is performed, but the trained model is not limited to a neural network. The braking control device 100 is not limited to a device equipped with a CPU and ROM and executing software processing. For example, it may be equipped with a dedicated hardware circuit that performs hardware processing on at least some of the software processes performed in the above embodiments. An example of a dedicated hardware circuit is an ASIC.

[0079] In the above embodiments, a caliper-type friction brake 20 is illustrated, but the friction brake may be a drum-type friction brake. In this case, the friction brake has a brake lining as a friction material and a brake drum as a friction-bearing material.

[0080] In the above embodiments, the hydraulic friction brake 20 is exemplified, but the friction brake may be an electric one. In this case, the friction material is driven by an electric motor. (Technical idea that can be understood from the embodiments, etc.) Next, the technical ideas that can be understood from the above-described embodiments and modifications will be described.

[0081] (i) It is preferable that the learning device is one that has undergone machine learning using the vehicle exterior situation information acquired from a plurality of the vehicles as learning data. (b) It is preferable that the information acquisition unit acquires, as the vehicle exterior situation information, information on an image obtained by capturing an image of the exterior of the vehicle.

[0082] (c) It is preferable that the information acquisition unit acquires, as the vehicle exterior situation information, information detected by a radar device mounted on the vehicle. (d) It is preferable that the information acquisition unit acquires, as the vehicle exterior situation information, information relating to a map including the position where the vehicle is traveling.

[0083] (e) The braking mechanism has a frictional part that rotates integrally with the wheel, and a frictional part that displaces in a direction relatively approaching the frictional part and in a direction relatively moving away from the frictional part, and applies a braking force to the wheel by bringing the frictional part into contact with the frictional part, It is preferable that the setting unit executes a process of bringing the friction portion relatively closer to the frictioned portion.

[0084] (f) The braking mechanism has a frictional part that rotates integrally with the wheel, and a frictional part that displaces in an approaching direction that is a direction that relatively approaches the frictional part and in a separating direction that is a direction that relatively moves away from the frictional part, and applies a braking force to the wheel by bringing the frictional part into contact with the frictional part, It is preferable that the setting unit executes a process of causing the braking device to generate a driving force within a range that allows the friction portion to be kept stationary.

[0085] (G) A vehicle system including a vehicle equipped with a braking device that applies a braking force to a wheel of the vehicle, and an external server device provided outside the vehicle, The vehicle is a vehicle-side communication unit that transmits vehicle exterior situation information, which is information about a situation outside the vehicle, to the vehicle exterior server device; a setting unit that sets a responsiveness of the braking operation in accordance with a probability that a braking operation that applies a braking force will occur in the braking device, The external server device A vehicle system including a server-side communication unit that inputs the vehicle exterior situation information transmitted from the vehicle communication unit into a learning device that has performed machine learning to estimate the probability of braking operation based on the vehicle exterior situation information, and transmits the indicators output from the learning device to the vehicle.

[0086] (H) A learning method for learning a learned model used to control a second vehicle based on learning data acquired from a first vehicle, an acquisition process of acquiring, as teacher data, outside-vehicle situation information relating to a situation outside the first vehicle when a braking operation for applying a braking force to a wheel of the first vehicle occurs; a learning process that performs supervised learning to input the outside vehicle situation information acquired in the acquisition process into the trained model, and estimate the probability of a braking operation that applies braking force to the wheels of the second vehicle occurring based on the situation outside the second vehicle by comparing the value output from the trained model with the result that a braking operation has occurred.

[0087] In this case, the first vehicle and the second vehicle may be different vehicles or may be the same vehicle. [Explanation of symbols]

[0088] 10...Vehicle 11...Wheel 20...Friction brake 21... Frictioned part 22...Friction part 30...braking device 100...Brake control device 101…Information acquisition department 102…Indicator acquisition part 103...Settings section 110...Learning device 120...Communication unit (an example of a vehicle-side communication unit) 400...Outside vehicle server device 403...Communication unit (an example of a server-side communication unit) LM: trained model

Claims

1. The present invention is applied to a braking device that applies braking force to the wheels of a vehicle, an information acquisition unit that acquires vehicle exterior situation information, which is image information relating to a situation outside the vehicle; a setting unit that sets a responsiveness of the braking operation in preparation for a braking operation in which the braking device applies braking force to the wheels, in accordance with an index output from a learning unit that has performed machine learning to estimate the possibility of a braking request being made to the braking device based on image information relating to a situation outside the vehicle, by inputting the vehicle exterior situation information acquired by the information acquisition unit into the learning unit, The setting unit preliminarily sets the responsiveness of the braking operation to be higher as the likelihood indicated by the index becomes higher.

2. A braking control device for a vehicle as described in claim 1, wherein the setting unit, in preparation for the braking operation, reduces in advance the amount of operation of the braking device required from the occurrence of the braking request to the application of braking force to the wheel, the higher the possibility indicated by the indicator.

3. A braking control device for a vehicle as described in claim 1 or claim 2, wherein the setting unit increases the operating speed of the braking device in advance in preparation for the braking operation to the extent that the probability indicated by the indicator is higher.

Images