Control device and control method

Abstract

The present invention obtains a control device and a control method that can appropriately assist with driving by a rider. In a control device (20) and a control method according to the present invention, an execution unit of the control device (20) executes a control mode in which a rider assistance operation for assisting a rider is executed. The execution unit, by using a learning model that has been trained to accept input of surrounding environment information for training, which is past surrounding environment information acquired by a leaning vehicle (1) and / or another leaning vehicle different from the leaning vehicle (1), and to produce behavior information about the leaning vehicle (1) as output, acquires first behavior information as the behavior information about the leaning vehicle (1) on the basis of host vehicle surrounding environment information, which is surrounding environment information at the present time acquired on the basis of output from a host vehicle sensor (14) mounted on the leaning vehicle (1), and executes the control mode on the basis of the first behavior information.

Description

[0001]

Document Name

[0002]

Title of the Invention

[0003]

Technical Field

[0004]

.001

[0005] This disclosure relates to a control device and a control method that can appropriately assist riding by a rider.

[0006]

Background Art

[0007]

.002

[0008] Conventionally, various technologies for assisting riding by a rider of a lean vehicle such as a motorcycle have been proposed. For example, in Patent Document 1, a driver assistance system is disclosed that warns a rider of a motorcycle that they are approaching an obstacle inappropriately based on information detected by a sensor device that detects an obstacle in the traveling direction or substantially in the traveling direction.

[0009]

Prior Art Documents

[0010]

Patent Documents

[0011]

〇003

[0012]

Patent Document 1

[0013]

Summary of the Invention

[0014]

Problems to be Solved by the Invention

[0015]

〇004

[0016] Here, in a lean vehicle, compared to a four-wheeled automobile or the like, the behavior is unstable and the behavior is likely to change sensitively. And in a control mode in which a rider support operation for supporting a rider is executed, it is desired to appropriately grasp the behavior of the lean vehicle and then appropriately assist the riding by the rider.

[0017]

〇005

[0018] This invention was made against the background of the above-mentioned problems, and aims to provide a control device and control method that can appropriately support driving by a LiDAR user.

[0019] [Means for solving the problem]

[0020]

〇 0 0 6

[0021] The control device according to the present invention is a control device for a rider assistance system that assists the rider of a lean vehicle, and comprises an execution unit that executes a control mode in which rider assistance operations are performed to assist the rider, and the execution unit uses a trained model that takes learning ambient environment information, which is past ambient environment information acquired from at least one of the lean vehicle and another lean vehicle different from the lean vehicle, as input and behavior information of the lean vehicle as output, to acquire first behavior information as behavior information of the lean vehicle based on the current ambient environment information of the vehicle, which is ambient environment information acquired based on the output of the vehicle's sensors mounted on the lean vehicle, and executes the control mode based on the first behavior information.

[0022]

〇 0 0 7

[0023] The control method according to the present invention is a control method for a rider assistance system that assists the rider of a lean vehicle, wherein the execution unit of the control device executes a control mode in which rider assistance operations are performed to assist the rider, and the execution unit uses a trained model that takes learning ambient environment information, which is past ambient environment information acquired from at least one of the lean vehicle and another lean vehicle different from the lean vehicle, as input and behavior information of the lean vehicle as output, to acquire first behavior information as behavior information of the lean vehicle based on the current ambient environment information of the vehicle, which is ambient environment information acquired based on the output of the vehicle's sensors mounted on the lean vehicle, and executes the control mode based on the first behavior information.

[0024] [Effects of the Invention]

[0025]

〇 0 0 8

[0026] In the control device and control method according to the present invention, the execution unit of the control device executes a control mode in which rider assistance operations are performed to assist the rider. The execution unit uses a trained learning model, which takes learning ambient environment information, which is past ambient environment information acquired from at least one of the lean vehicle and another lean vehicle different from the lean vehicle, as input and behavior information of the lean vehicle as output, to acquire first behavior information as behavior information of the lean vehicle based on the current ambient environment information of the vehicle, which is ambient environment information acquired based on the output of the vehicle's sensors mounted on the lean vehicle. Based on the first behavior information, the execution unit executes a control mode. As a result, the control mode can be executed after appropriately understanding the behavior of the lean vehicle. Therefore, it is possible to appropriately assist rider driving.

[0027] [Brief explanation of the drawing]

[0028] [ 0 0 0 9 ]

[0029] [Figure 1] This is a schematic diagram showing the general configuration of a lean vehicle according to an embodiment of the present invention.

[0030] [Figure 2] Block diagram showing an example of the functional configuration of a control device according to an embodiment of the present invention.

[0031]

[0032] Furthermore, while the following describes a case where a control unit that controls the hydraulic pressure of the brake fluid (specifically, hydraulic pressure control units 1 and 2 in Figure 1, which will be described later) is used as the control unit for the braking force generated in the wheel, a control unit that controls the position of the wheel's braking part itself by electrical signals (so-called brake-by-wire) may also be used as the control unit for the braking force generated in the wheel.

[0033] [ 0 0 1 4 ]

[0034] Furthermore, the configurations and operations described below are merely examples, and the control device and control method according to the present invention are not limited to such configurations and operations.

[0035] [ 0 0 1 5 ]

[0036] Furthermore, in the following, identical or similar explanations have been simplified or omitted as appropriate. Also, in each figure, identical or similar components or parts have either had their reference numerals omitted or have been given the same reference numeral. In addition, detailed structures have been simplified or omitted as appropriate.

[0037] [ 0 0 1 6 ]

[0038] Clean vehicle configuration >

[0039]

[0040] This includes a hydraulic control unit 12, a display device 13, an ambient environment sensor 14, a front wheel speed sensor 15, a rear wheel speed sensor 16, and a control device 20).

[0041] [ 0 0 1 9 ]

[0042] Engine 11 is an example of a drive source for lean vehicle 1 and is capable of outputting power to drive the drive wheels (specifically, the rear wheels 3). For example, engine 11 is equipped with one or more cylinders in which combustion chambers are formed, a fuel injector that injects fuel into the combustion chamber, and a spark plug. When fuel is injected from the fuel injector, a mixture of air and fuel is formed in the combustion chamber, and this mixture is ignited by the spark plug and combusted. As a result, a piston located in the cylinder reciprocates and the crankshaft rotates. In addition, a throttle valve is provided in the intake manifold of engine 11, and the amount of intake air into the combustion chamber changes according to the throttle opening, which is the opening degree of the throttle valve.

[0043] [ 0 0 2 0 ]

[0044] The hydraulic control unit 12 is a unit responsible for controlling the braking force generated on the wheels. For example, the hydraulic control unit 12 is provided on the oil path connecting the master cylinder and the wheel cylinder, and includes components (such as control valves and pumps) for controlling the brake hydraulic pressure of the wheel cylinder. By controlling the operation of the components of the hydraulic control unit 12, the braking force generated on the wheels is controlled. Note that the hydraulic control unit 12 may control the braking forces generated on both the front wheels 2 and the rear wheels 3, or may control only the braking force generated on one of the front wheels 2 and the rear wheels 3.

[0045]

[0021]

[0046] The display device 13 has a display function for visually displaying information to the rider. Examples of the display device 13 include a liquid crystal display and the like. The display device 13 is provided, for example, in front of the handle in the lean vehicle 1. However, the arrangement of the display device 13 with respect to the vehicle body is not particularly limited.

[0047]

[0022]

[0048] The surrounding environment sensor 14 detects surrounding environment information regarding the environment around the lean vehicle 1. Specifically, the surrounding environment sensor 14 is provided at the front part of the lean vehicle 1 and detects the surrounding environment information in front of the lean vehicle 1. The surrounding environment information detected by the surrounding environment sensor 14 is output to the control device 20.

[0049]

[0023]

[0050] The surrounding environment information detected by the surrounding environment sensor 14 may be information related to the distance or orientation to an object located around the lean vehicle 1 (such as relative position, relative distance, relative speed, relative acceleration, etc.), or may be the characteristics of an object located around the lean vehicle 1 (such as the type of the object, the shape of the object itself, the mark attached to the object, etc.)

[0024]

[0051] For example, the ambient environment sensor 14 is a sensor that receives reflected electromagnetic waves from a transmitted signal. Examples of such sensors include radar, Lidar sensors, ultrasonic sensors, and so on.

[0052] [ 0 0 2 5 ]

[0053] The front wheel speed sensor 15 is a wheel speed sensor that detects the wheel speed of the front wheel 2 (for example, the number of rotations per unit time [rpm] or the distance traveled per unit time [km / h] of the front wheel 2, etc.) and outputs the detection result. The front wheel speed sensor 15 may also detect other physical quantities that can be substantially converted to the wheel speed of the front wheel 2. The front wheel speed sensor 15 is installed on the front wheel 2.

[0054] [ 0 0 2 6 ]

[0055] The rear wheel speed sensor 16 is a wheel speed sensor that detects the wheel speed of the rear wheel 3 (for example, the number of rotations per unit time [rpm] or the distance traveled per unit time [km / h] of the rear wheel 3, etc.) and outputs the detection result. The rear wheel speed sensor 16 may also detect other physical quantities that can be substantially converted to the wheel speed of the rear wheel 3. The rear wheel speed sensor 16 is installed on the rear wheel 3.

[0056] [ 0 0 2 7 ]

[0057] The control device 20 controls the operation of the rider assistance system 100. For example, part or all of the control device 20 is composed of a microcontroller, microprocessor unit, etc. Also, for example, part or all of the control device 20 may be composed of updatable components such as firmware, or a program module executed by commands from a CPU, etc. The control device 20 may be, for example, one unit, or it may be divided into multiple units.

[0028]

[0058] Figure 2 is a block diagram showing an example of the functional configuration of the control device 20. As shown in Figure 2, the control device 20 includes, for example, an acquisition unit 21, an execution unit 22, and a storage unit 23. The control device 20 communicates with each device of the lean vehicle 1 (for example, the engine 11, the hydraulic control unit 12, the display device 13, the ambient environment sensor 14, the front wheel speed sensor 15, and the rear wheel speed sensor 16). The control device 20 can also control the operation of each device of the lean vehicle 1 (for example, the engine 11, the hydraulic control unit 12, and the display device 13).

[0059] [ 0 0 2 9 ]

[0060] The acquisition unit 21 acquires information from each device of the lean vehicle 1 and outputs it to the execution unit 22. For example, the acquisition unit 21 acquires information from the ambient environment sensor 14, the front wheel speed sensor 15, and the rear wheel speed sensor 16. In this specification, information acquisition may include information extraction or generation (e.g., calculation).

[0061] [ 0 0 3 0 ]

[0062] The execution unit 22 can execute a control mode in which rider support operations are performed to assist the rider. In other words, the execution unit 22 performs rider support operations in the above control mode. The execution unit 22 can perform various rider support operations by, for example, controlling the operation of the engine 11, the hydraulic control unit 12, and the display device 13. Examples of rider support operations include behavior stabilization operations to stabilize the behavior of the lean vehicle 1, positional relationship adjustment operations to adjust the positional relationship between the lean vehicle 1 and the preceding vehicle to a target positional relationship, or notification operations to notify the rider. Details of the rider support operations will be described later.

[0063] [ 0 0 3 1 ]

[0064] Furthermore, the above control mode may be executed at all times when the power supply of lean vehicle 1 is turned on, or it may be executed only when a specific operation is performed by the rider. In other words, the rider may permit or disable the above control mode. For example, for some types of rider assistance operations to be performed, the control mode may be executed at all times when the power supply of lean vehicle 1 is turned on, while for some other types, the control mode may be executed only when a specific operation is performed by the rider.

[0065] [ 0 0 3 2 ]

[0066] The memory units 2 and 3 store various types of information. For example, the memory units 2 and 3 store information used in the processing performed by the execution unit 2 and 2 (for example, the learning model described later).

[0067] [ 0 0 3 3 ]

[0068] Control device operation>

[0069] The operation of control device 2〇 according to an embodiment of the present invention will be described.

[0070] [ 0 0 3 4 ]

[0071] As described above, the execution unit 22 of the control device 20 can execute a control mode in which rider assistance operations are performed to support the rider. The rider assistance operations then support the rider's driving. In this case, the lean vehicle 1 is more unstable and prone to sudden changes in behavior compared to a four-wheeled automobile, etc. Therefore, in this embodiment, as will be described later, by making improvements to the processing performed by the execution unit 22, it becomes possible to execute a control mode after appropriately understanding the behavior of the lean vehicle 1, thereby appropriately supporting the rider's driving. An example of processing performed by the control device 20 will be described below.

[0072] [ 0 0 3 5 ]

[0073] Figure 3 is a flowchart showing an example of the processing flow performed by the control device 20. The control flow shown in Figure 3 starts during the execution of a control mode in which LiDAR assistance operation is performed and ends when the control mode ends. Step S101 in Figure 3 corresponds to the start of the control flow shown in Figure 3.

[0074] [ 0 0 3 6 ]

[0075] When the control flow shown in Figure 3 begins, in step S102, the execution unit 22 acquires self-vehicle information, which is the current information of lean vehicle 1, as information to be input to the pre-trained learning model described later. Next, in step S103, the execution unit 22 acquires behavior information of lean vehicle 1 by inputting the self-vehicle information acquired in step S102 into the learning model. The behavior information acquired in step S103 is also called first behavior information. The behavior information may include various information about the behavior of lean vehicle 1. Details of the behavior information will be described later. Now, with reference to Figure 4, the details of the learning model will be explained.

[0076] [ 0 0 3 7 ]

[0077] Figure 4 is a schematic diagram conceptually illustrating the learning model M1. The learning model M1 is a model trained using learning information, which is past information acquired from a lean vehicle other than lean vehicle 1, as input, and behavioral information of lean vehicle 1 as output. The learning model M1 is stored, for example, in memory units 2 and 3. In the following, we will mainly describe an example in which the learning model M1 is trained using learning information acquired in the past from a lean vehicle other than lean vehicle 1, but as will be described later, the learning model M1 may also be trained using learning information acquired in the past from lean vehicle 1.

[0078] [ 0 0 3 8 ]

[0079] As shown in Figure 4, the learning model M1 is a model that, upon receiving various input information I nf 〇 _ IN, outputs various output information I nf 〇 _ UT according to the input information I nf 〇 _ IN. Here, the input information I nf 〇 _ 1 N includes at least information about the surrounding environment. However, as will be described later, the input information I nf 〇 _ 1 N may include other information in addition to the surrounding environment information. Also, the output information I nf 〇 _ UT is behavioral information. In other words, when information including at least information about the surrounding environment is input to the learning model M1, the output information I nf 〇 _ UT, which is behavioral information, is output.

[0080] [ 0 0 3 9 ]

[0081] The ambient environment information input as Input Information I nf 〇 _ IN includes, for example, positional relationship information between lean vehicle 1 and objects stationary around lean vehicle 1. Objects stationary around lean vehicle 1 may include, for example, the road surface, fixed objects provided on the side of the road (e.g., guardrails, signs, vegetation, walls, etc.). The positional relationship information may also include, for example, information such as the relative position, relative distance, relative speed, relative acceleration, relative jerk, or difference in passing time of lean vehicle 1 relative to the objects stationary around lean vehicle 1. Furthermore, the positional relationship information may also be other information that is substantially convertible into this information. [ 0 0 4 0 ]

[0082] Furthermore, the ambient environment information input as input information I nf 〇 _ IN includes, for example, reflection characteristics information relating to the identification of reflected electromagnetic waves from an object stationary around the lean vehicle 1. Electromagnetic waves include, for example, light and radio waves. The reflection characteristics information may include, for example, information such as reflected power and reflection cross-section of the reflected waves from an object stationary around the lean vehicle 1. The positional relationship information may also be other information that can be substantially converted into this information.

[0083] [ 0 0 4 1 ]

[0084] The information input as Input Information I nf 〇 _ IN may include speed information in addition to ambient environment information. Speed ​​information is information about the vehicle's speed. For example, the acquisition unit 2 1 can acquire speed information of lean vehicle 1 based on the outputs of the front wheel speed sensor 1 5 and the rear wheel speed sensor 1 6. Similarly, speed information of other lean vehicles other than lean vehicle 1 can also be acquired. Note that speed information may be acquired based on information transmitted from GPS (Global Positioning System) satellites.

[0085] [ 0 0 4 2 ]

[0086] Furthermore, the information input as input information In nf 〇 _ IN may include braking information in addition to ambient environment information. Braking information is information about the braking state of the vehicle. Note that the braking information only needs to be information about the braking state of at least one of the front wheels 2 and rear wheels 3. For example, the acquisition unit 2 1 can acquire information about the pressure of at least one of the wheel cylinders of the front wheels 2 and rear wheels 3 of lean vehicle 1 as braking information, based on information about the amount of operation of the brake operating part of lean vehicle 1, or the output of a sensor that detects the pressure of the wheel cylinders. Similarly, braking information of other lean vehicles other than lean vehicle 1 can also be acquired.

[0087] [ 0 0 4 3 ]

[0088] The behavior information output as Output Information I nf 〇 _〇 UT includes, for example, the turning attitude information of lean vehicle 1. The turning attitude information is information about physical quantities that reflect the attitude of lean vehicle 1 as it turns. Examples of turning attitude information include the lean angle of lean vehicle 1, the yaw rate of lean vehicle 1, and the lateral acceleration of lean vehicle 1.

[0089] [ 0 0 4 4 ]

[0090] Furthermore, the behavior information output as "Output Information Info — OUT" includes, for example, pitch behavior information of lean vehicle 1. Pitch behavior information is information regarding the behavior of lean vehicle 1 in the pitch direction. Examples of pitch behavior information include the pitch angle of lean vehicle 1.

[0091] [ 0 0 4 5 ]

[0092] Furthermore, the behavior information output as output information Info — OUT includes, for example, longitudinal behavior information of lean vehicle 1. The longitudinal behavior information is information regarding the behavior of lean vehicle 1 in the longitudinal direction. Examples of longitudinal behavior information include information regarding the future stopping of lean vehicle 1, information indicating whether lean vehicle 1 is stopped or not, and the longitudinal acceleration of lean vehicle 1.

[0093] [ 0 0 4 6 ]

[0094] As described above, for example, if ambient environment information (e.g., positional relationship information and reflective properties information), speed information, and braking information are input to the learning model M1 as input information I nf 〇 _ IN, various behavioral information (e.g., turning attitude information, pitch behavior information, and longitudinal behavior information of lean vehicle 1) is output as output information I nf 〇 _ UT.

[0095] [ 0 0 4 7 ]

[0096] The information input as "Inf 〇 _ IN" may include all types of information listed above, or only some of them. Similarly, the information output as "Output 1 nf 〇 _ O UT" may include all types of information listed above, or only some of them. [ 0 0 4 8 ]

[0097] Here, we will describe the training of the learning model M! that has been carried out in the past.

[0098] [ 0 0 4 9 ]

[0099] First, in training the learning model M1, learning information, which serves as training data for the input information I nf 〇 _ IN, is acquired from a different lean vehicle than lean vehicle 1. For example, based on the output of the surrounding environment sensor mounted on the other lean vehicle, the aforementioned surrounding environment information (e.g., positional relationship information and reflection characteristic information) is acquired as learning information. The surrounding environment information acquired as learning information is also called learning surrounding environment information. Furthermore, for example, based on the output of the front wheel speed sensor and rear wheel speed sensor mounted on the other lean vehicle, the aforementioned speed information is acquired as learning information. The speed information acquired as learning information is also called learning speed information. Furthermore, for example, based on the output of the sensor that detects the pressure of the wheel cylinder mounted on the other lean vehicle, the aforementioned braking information is acquired as learning information. The braking information acquired as learning information is also called learning braking information.

[0100] [ 0 0 5 0 ]

[0101] Furthermore, during the training of the learning model M1, the behavior information of the other lean vehicle at the time the learning information was acquired is obtained as training data for the output information I nf 〇 _O UT. For example, based on the output of the inertial measurement device mounted on the other lean vehicle, the behavior information described above (e.g., turning attitude information, pitch behavior information, and longitudinal behavior information of lean vehicle 1) is obtained as training data for the output information I nf 〇 _O UT. The inertial measurement device is, for example, a device that detects acceleration in the three axes and angular velocity around the three axes. Note that the inertial measurement device may be a device that detects only a portion of the acceleration in the three axes and angular velocity around the three axes. In particular, when a sensor that receives reflected electromagnetic waves transmitted as an ambient environment sensor is used, it is difficult to directly and accurately calculate behavior information using the output of the ambient environment sensor. On the other hand, it is easy to directly and accurately calculate behavior information using the output of the inertial measurement device.

[0102] [ 0 0 5 1 ]

[0103] Then, the learning model M1 is trained based on multiple data sets, in which training information equivalent to the training data for input information I nf 〇 _ IN and behavioral information equivalent to the training data for output information I nf 〇 _ O UT are linked. In other words, the above multiple data sets serve as inputs for training the learning model M1. In this way, the learning model M1 is trained using training information, which is past information acquired from a different lean vehicle than lean vehicle 1, as input, and behavioral information of lean vehicle 1 as output.

[0104] [ 0 0 5 2 ]

[0105] Specifically, the learning model M1 is trained until the variability of the behavior information of lean vehicle 1 output by the learning model M1 falls within a reference range (specifically, a predetermined range). For example, the learning model M1 is trained until the standard deviation of the output when training information associated with the same or nearly equivalent behavior information is input to the learning model M1 falls below a predetermined reference value.

[0106] [ 0 0 5 3 ]

[0107] As described above, in step S! 2 of Figure 3, the execution unit 22 executes the learning model M described above.

[0108]

[0109] In step S102 of Figure 3, for example, the execution unit 22 acquires the above-mentioned ambient environment information (e.g., positional relationship information and reflection characteristic information) as vehicle information based on the output of the ambient environment sensor 14 mounted on the lean vehicle 1. The ambient environment information acquired as vehicle information is also called vehicle ambient environment information. Furthermore, for example, the execution unit 22 acquires the above-mentioned speed information as vehicle information based on the output of the front wheel speed sensor 15 and rear wheel speed sensor 16 mounted on the lean vehicle 1. The speed information acquired as vehicle information is also called vehicle speed information. Furthermore, for example, the execution unit 22 acquires the above-mentioned braking information as vehicle information based on the output of the sensor that detects the pressure of the wheel cylinder mounted on the lean vehicle 1. The braking information acquired as vehicle information is also called vehicle braking information.

[0110] [ 0 0 5 5 ]

[0111] In step S! 3 of Figure 3, the execution unit 22 inputs various vehicle information acquired in step S! 2 as input information I nf _ IN to the learning model M1, for example, thereby acquiring behavior information of the lean vehicle 1 (i.e., first behavior information) which is output as output information I nf _ O UT. Here, in this embodiment, an inertial measuring device is not provided on the lean vehicle 1. If an inertial measuring device is provided on the lean vehicle 1, it is easy to directly and accurately calculate behavior information using the output of the inertial measuring device. In this regard, even in the case of a lean vehicle 1 that is not provided with an inertial measuring device, the behavior information of the lean vehicle 1 can be acquired by using the pre-trained learning model M1 through the above processing by the execution unit 22.

[0112] [ 0 0 5 6 ]

[0113] Following step S103 in Figure 3, in step S104, the execution unit 22 executes the control mode based on the behavior information acquired in step S103, and returns to step S102

[0057] .

[0114] As described above, various types of rider assistance operations can be performed in the control mode. In other words, various control modes can be performed in the control mode in which rider assistance operations are performed. The execution unit 22 can execute various control modes based on behavior information.

[0115] [ 0 0 5 8 ]

[0116] For example, the execution unit 22 can execute a control mode in which a behavior stabilization operation to stabilize the behavior of the lean vehicle 1 is performed as a rider assistance operation. The execution unit 22 may then execute such a control mode based on behavior information.

[0117] [ 0 0 5 9 ]

[0118] As a behavior stabilization operation, for example, anti-lock brake control is used to suppress slippage caused by the braking force of the wheels of the lean vehicle 1 by controlling the braking force of the lean vehicle 1. In anti-lock brake control, for example, braking force control is performed to suppress slippage of the wheels by reducing the braking force acting on the wheels. For example, the execution unit 22 may change the degree of reduction of the braking force in the braking force control based on behavior information (for example, turning posture information). Alternatively, for example, the execution unit 22 may change whether or not to perform braking force control based on behavior information (for example, longitudinal behavior information).

[0119] [ 0 0 6 0 ]

[0120] Furthermore, as a behavior stabilization operation, for example, traction control is used to suppress slippage caused by the driving force of the lean vehicle 1's wheels by controlling the driving force of the lean vehicle 1. In traction control, for example, driving force control is performed to suppress slippage of the wheels by reducing the driving force acting on the wheels. For example, the execution unit 22 may change the degree of reduction of the driving force in the driving force control based on behavior information (for example, turning posture information). Alternatively, for example, the execution unit 22 may change whether or not to perform driving force control based on behavior information (for example, longitudinal behavior information).

[0121] [ 0 0 6 1 ]

[0122] Furthermore, for example, the execution unit 22 can execute a control mode in which a positional relationship adjustment operation is performed as a rider assistance operation to adjust the positional relationship between the leaning vehicle 1 and the target vehicle to the target positional relationship. The execution unit 22 may execute such a control mode based on behavior information.[0 0 6 2] An example of a positional relationship adjustment operation is adaptive cruise control. Note that the positional relationship adjustment operation may be an operation in which the target positional relationship changes according to the amount of accelerator operation performed by the rider. In adaptive cruise control, for example, the future driving trajectory of the leaning vehicle 1 is estimated, and the target vehicle is determined based on the estimated driving trajectory. For example, the execution unit 22 may estimate the future driving trajectory of the leaning vehicle 1 based on behavior information (e.g., turning attitude information), and determine the target vehicle based on the estimated driving trajectory.

[0123] [ 0 0 6 3 ]

[0124] Furthermore, for example, the execution unit 22 can execute a control mode in which a notification operation to notify the rider is performed as a rider support operation. The execution unit 22 may then execute such a control mode based on behavioral information.

[0125] [ 0 0 6 4 ]

[0126] Examples of notification actions include warning the rider that the likelihood of a collision between the leaning vehicle 1 and an object such as a target vehicle exceeds a certain threshold. For example, the above notification is made using a display device 13. However, the above notification may also be made using a display device provided on the rider's equipment (e.g., a helmet), or using a sound output device or vibration generating device provided on the leaning vehicle 1 or the rider's equipment. In such a notification, for example, the future trajectory of the leaning vehicle 1 is estimated, and the above object is determined based on the estimated trajectory. For example, the execution unit 22 may estimate the future trajectory of the leaning vehicle 1 based on behavior information (e.g., turning attitude information), and determine the above object based on the estimated trajectory.

[0127] [ 0 0 6 5 ]

[0128] The above describes an example of processing performed by the control device 20. However, the processing performed by the control device 20 may be a modified version of the processing example described above.

[0129] [ 0 0 6 6 ]

[0130] For example, the above explanation assumes the use of a learning model M1 that is basically compatible with one vehicle type. However, learning models M1 may be prepared to accommodate multiple vehicle types. For example, vehicle type identification information (e.g., information on the installation height of the ambient environment sensor 14, vehicle type information, damper setting information, suspension type information, etc.) may be added as input information Inf 〇- 1 N. In that case, by inputting the vehicle type identification information into learning model M1, learning model Ml is trained to acquire behavioral information corresponding to the vehicle type of the own vehicle from among multiple vehicle types. Alternatively, for example, a separate learning model M1 may be prepared for each vehicle type, and the learning model M1 corresponding to the vehicle type of the own vehicle may be used from among the multiple learning models M1. Note that if multiple learning models M1 are prepared, the multiple learning models Ml may be stored separately in different control devices.

[0131] [ 0 0 6 7 ]

[0132] Furthermore, for example, the above description referred to a lean vehicle 1 that is not equipped with an inertial measuring device. However, the lean vehicle according to the present invention may be equipped with an inertial measuring device. Figure 5 is a schematic diagram showing the general configuration of a modified lean vehicle 1A. As shown in Figure 5, the modified lean vehicle 1A differs from the lean vehicle 1 described above in that it is additionally equipped with an inertial measuring device 17.

[0133] [ 0 0 6 8 ]

[0134] In the lean vehicle 1A, the execution unit 22 can acquire behavioral information of the lean vehicle 1A based on the detection results of the inertial measuring device 17 mounted on the lean vehicle 1A. Here, the behavioral information acquired based on the detection results of the inertial measuring device 17 is also called the second behavioral information. The execution unit 22 may then execute a control mode based on the first behavioral information (i.e., the behavioral information acquired using the learning model M1 as described above) and the second behavioral information.

[0135] [ 0 0 6 9 ]

[0136] For example, if there is behavior information included in the first behavior information but not in the second behavior information, the execution unit 22 may supplement the second behavior information with such behavior information. If the inertial measuring device 17 can detect the lean angle but cannot detect the pitch angle, the second behavior information will not include the pitch angle. In that case, the execution unit 22 may add the pitch angle included in the first behavior information to the second behavior information.

[0137] [ 0 0 7 0 ]

[0138] Furthermore, for example, the execution unit 22 basically uses the information contained in the second behavior information as behavior information used in the control mode, but may use the information contained in the first behavior information as behavior information used in the control mode for only some of the information. For example, if the information contained in the first behavior information is more accurate than the information contained in the second behavior information with respect to specific information (for example, forward / backward behavior information), the execution unit 22 may prioritize using the information contained in the first behavior information with respect to such specific information.

[0139] [ 0 0 7 1 ]

[0140] Furthermore, for example, the execution unit 22 may execute a control mode based on the comparison result between the first behavior information and the second behavior information. For example, the execution unit 22 may evaluate the reliability of the second behavior information based on the comparison result between the first behavior information and the second behavior information. For example, if the discrepancy between the first behavior information and the second behavior information is excessively large, the execution unit 22 evaluates the reliability of the second behavior information as being lower than the standard. In that case, the execution unit 22 may, for example, prohibit processing that utilizes the second behavior information (for example, processing that executes a control mode based on the second behavior information). In this case, the execution unit 22 may, for example, execute a control mode based on the first behavior information. Furthermore, if the discrepancy between the first behavior information and the second behavior information is excessively large, the execution unit 22 may notify the rider that the reliability of the second behavior information is lower than the standard.

[0141] [ 0 0 7 2 ]

[0142] Furthermore, while the above mainly described an example in which the learning model M1 is trained using training information previously acquired from a lean vehicle other than lean vehicle 1, the learning model Ml may also be trained using training information previously acquired from the vehicle itself. For example, if an inertial measurement device 17 is installed on the vehicle, as in the lean vehicle 1A described above, the behavior information of the vehicle can be acquired as training data for the output information Inf _OUT based on the output of the inertial measurement device 17. Therefore, in this case, the learning model Ml can be trained using training information previously acquired from the vehicle itself. In other words, the learning model Ml should be trained using learning environment information, which is past surrounding environment information acquired from at least one of the vehicle itself and another lean vehicle different from the vehicle, as input, and behavior information as output.

[0143] [ 0 0 7 3 ]

[0144] <Effects of the control device>

[0145] The effects of the control device 2〇 according to the embodiment of the present invention will be described.

[0146] [ 0 0 7 4 ]

[0147] The control device 20 includes an execution unit 22 that executes a control mode in which rider assistance actions are performed to assist the rider. The execution unit 22 uses a trained learning model M1, which takes learning ambient environment information (past ambient environment information acquired from at least one of the lean vehicle 1 and another lean vehicle different from lean vehicle 1) as input and behavior information of lean vehicle 1 as output, to acquire first behavior information as behavior information of lean vehicle 1 based on the current ambient environment information (self-vehicle ambient environment information acquired based on the output of the self-vehicle sensor (ambient environment sensor 14 in the above example) mounted on lean vehicle 1), and executes a control mode based on the first behavior information. This allows the control mode to be executed after appropriately understanding the behavior of lean vehicle 1. Therefore, it is possible to appropriately assist rider driving.

[0148] [ 0 0 7 5 ]

[0149] Preferably, in the control device 20, the learning environment information includes positional relationship information between the lean vehicle 1 and an object stationary around the lean vehicle 1. This appropriately enables the training of a learning model M1 that takes positional relationship information as input and outputs behavior information, and appropriately enables the acquisition of first behavior information using the learning model M1. [0 0 7 6]

[0150] Preferably, in the control device 20, the information on the surrounding environment of the vehicle includes positional relationship information between the leaning vehicle 1 and objects stationary around the leaning vehicle 1. This allows for the appropriate acquisition of first behavior information by obtaining the current positional relationship information and inputting it into the learning model M1.

[0151] [ 0 0 7 7 ]

[0152] Preferably, in the control device 20, the vehicle sensor (in the above example, the ambient environment sensor 14) is a sensor that receives reflected electromagnetic waves from the transmitted waves. This appropriately enables the training of a learning model M1 that takes reflection characteristic information as input and outputs behavioral information, and appropriately enables the acquisition of first behavioral information using the learning model M1. Furthermore, it appropriately enables the acquisition of first behavioral information by acquiring the current reflection characteristic information and inputting it into the learning model M1.

[0153] [ 0 0 7 8 ]

[0154] Preferably, in the control device 20, the learning model Ml is trained using learning speed information, which is past speed information of at least one of the lean vehicle 1 and another lean vehicle different from lean vehicle 1, in addition to learning ambient environment information, as input. The execution unit 22 uses the learning model Ml to acquire first behavior information based on the vehicle's own speed information, which is the current speed information of lean vehicle 1, in addition to the vehicle's own ambient environment information. By training the learning model M1 with learning speed information in addition to learning ambient environment information as input, the accuracy of the behavior information output by the learning model M1 can be improved, and the first behavior information can be acquired with greater accuracy.

[0155] [ 0 0 7 9 ]

[0156] Preferably, in the control device 20, the learning model M1 is trained using learning braking information, which is past braking information of at least one of the lean vehicle 1 and another lean vehicle different from lean vehicle 1, in addition to learning ambient environment information. The execution unit 22 uses the learning model M1 to acquire first behavior information based on the self-vehicle braking information, which is the current braking information of lean vehicle 1, in addition to the self-vehicle ambient environment information. By training the learning model M1 using learning braking information in addition to learning ambient environment information as input, the accuracy of the behavior information output by the learning model M1 can be improved, and the first behavior information can be acquired with greater accuracy.

[0157] [ 0 0 8 0 ]

[0158] Preferably, in the control device 20, the first behavior information includes information on the turning posture of the lean vehicle 1. This allows the control mode to be executed after appropriately understanding the turning posture of the lean vehicle 1.

[0159] [ 0 0 8 1 ]

[0160] Preferably, in the control device 20, the first behavior information includes pitch behavior information of the lean vehicle 1. This allows the control mode to be executed after appropriately understanding the pitch behavior of the lean vehicle 1.

[0161] [ 0 0 8 2 ]

[0162] Preferably, in the control device 20, the first behavior information includes longitudinal behavior information of the lean vehicle 1. This allows the control mode to be executed after appropriately understanding the longitudinal behavior of the lean vehicle 1.

[0163] [ 0 0 8 3 ]

[0164] Preferably, in the control device 20, the learning model M1 is trained until the variability of the behavior information of the lean vehicle 1 output by the learning model M1 falls within a reference range. This improves the accuracy of the behavior information output by the learning model M1, and allows for more accurate acquisition of the first behavior information.

[0165] [ 0 0 8 4 ]

[0166] Preferably, in the control device 20, the execution unit 22 acquires second behavior information as behavior information of the lean vehicle 1A based on the detection results of the inertial measuring device 17 mounted on the lean vehicle 1A, and executes a control mode based on the first behavior information and the second behavior information. Thus, when the lean vehicle 1A is equipped with an inertial measuring device 17, the control mode can be executed using the first behavior information acquired using the learning model M1 in addition to the second behavior information acquired based on the detection results of the inertial measuring device 17. Therefore, for example, the second behavior information can be supplemented with information contained in the first behavior information. Also, for example, with respect to specific information, the information contained in the first behavior information can be used preferentially.

[0167] [ 0 0 8 5 ]

[0168] Preferably, in the control device 20, the execution unit 22 executes a control mode based on the comparison result between the first behavior information and the second behavior information. Thereafter, for example, the reliability of the second behavior information

[0169]

Claims

[Document Name] Scope of Claim

1. A control device (2) for a rider assistance system (100) that assists the rider of a lean vehicle (1), The system includes an execution unit (22) that executes a control mode in which rider assistance operations are performed to support the rider, The execution unit (22) is, Using a trained model (Ml) that takes learning ambient environment information, which is past ambient environment information acquired from at least one of the lean vehicle (1) and another lean vehicle different from the lean vehicle (1), as input and behavior information of the lean vehicle (1) as output, first behavior information is acquired as behavior information of the lean vehicle (1) based on the current ambient environment information, which is the ambient environment information acquired based on the output of the self-vehicle sensor (14) mounted on the lean vehicle (1). Based on the aforementioned first behavior information, the control mode is executed. Control device.

2. The learning environment information includes positional relationship information between the lean vehicle (1) and an object stationary around the lean vehicle (1). The control device according to claim 1.

3. The aforementioned information on the environment surrounding the vehicle includes positional relationship information between the lean vehicle (1) and an object stationary around the lean vehicle (1). The control device according to claim 1.

4. The control device according to claim 1, wherein the vehicle sensor (14) is a sensor that receives reflected waves of transmitted electromagnetic waves.

5. The learning model (M l) is trained using, in addition to the learning ambient environment information, learning speed information which is past speed information of at least one of the lean vehicle (1) and another lean vehicle different from the lean vehicle (1), as input. The execution unit (22) uses the learning model (Ml) to acquire the first behavior information based on the vehicle speed information, which is the current speed information of the lean vehicle (1), in addition to the information about the surrounding environment of the vehicle. The control device according to claim 1.

6. The learning model (M l) is trained using, in addition to the learning ambient environment information, learning braking information which is past braking information of at least one of the lean vehicle (1) and another lean vehicle different from the lean vehicle (1), as input. The execution unit (22) uses the learning model (Ml) to acquire the first behavior information based on the vehicle braking information, which is the current braking information of the lean vehicle (1), in addition to the information about the surrounding environment of the vehicle. The control device according to claim 1.

7. The aforementioned first behavior information includes the turning attitude information of the lean vehicle (1), The control device according to claim 1.

8. The aforementioned first behavior information includes pitch behavior information of the lean vehicle (1), The control device according to claim 1.

9. The aforementioned first behavior information includes longitudinal behavior information of the lean vehicle (1), The control device according to claim 1. [Claim 1〇] The learning model (M l) is trained until the variability of the behavior information of the lean vehicle (1) output by the learning model (M l) falls within a reference range. The control device according to claim 1. [Claim 1 1] The execution unit (22) is, Based on the detection results of the inertial measuring device (17) mounted on the lean vehicle (1A), second behavior information is acquired as behavior information of the lean vehicle (1A). A control device according to any one of claims 1 to 10, which executes the control mode based on the first behavior information and the second behavior information. [Claim 1 2] The execution unit (22) executes the control mode based on the comparison result between the first behavior information and the second behavior information. The control device according to claim 11. [Claim 1 3] A control method for a rider assistance system (100) that assists the rider of a lean vehicle (1), The execution unit (22) of the control device (20) executes a control mode in which rider assistance operations are performed to assist the rider. The execution unit (22) Using a trained model (Ml) that takes learning ambient environment information, which is past ambient environment information acquired from at least one of the lean vehicle (1) and another lean vehicle different from the lean vehicle (1), as input and behavior information of the lean vehicle (1) as output, first behavior information is acquired as behavior information of the lean vehicle (1) based on the current ambient environment information, which is the ambient environment information acquired based on the output of the self-vehicle sensor (14) mounted on the lean vehicle (1). Based on the aforementioned first behavior information, the control mode is executed. Control method.

Images