Driving control device, driving control method, and driving control computer program

Abstract

To appropriately control the motion of a vehicle.SOLUTION: A travel control device: generates a candidate parameter set including one or more parameters by inputting input data including a peripheral image acquired by a peripheral imaging unit that images the condition of the periphery of a vehicle into a classifier that has been trained so as to output the one or more parameters for controlling the travel of the vehicle in response to input of the input data; predicts a future motion of the vehicle under control of travel with the candidate parameter set; controls the travel of the vehicle by using the candidate parameter set when the future motion of the vehicle predicted for the candidate parameter set does not correspond to any of one or more undesirable motions indicating the undesirable motions of the vehicle stored in a storage unit; and controls the travel of the vehicle without using the candidate parameter set when the future motion of the vehicle predicted for the candidate parameter set corresponds to any of the one or more avoidance motions.SELECTED DRAWING: Figure 7

Description

[Technical Field]

[0001] The present disclosure relates to a driving control device for controlling the driving of a vehicle, a driving control method, and a driving control computer program. [Background technology]

[0002] A driving control device that controls the driving of a vehicle through autonomous driving controls the driving of the vehicle so as to maintain a predetermined distance from objects such as other vehicles and pedestrians that exist around the vehicle. The driving control device appropriately controls the driving of the vehicle by performing processes such as recognizing surrounding objects, predicting the future positions of surrounding objects, creating a driving route, and specifying the operating parameters of the driving mechanism for driving along the driving route.

[0003] In recent years, attention has been drawn to cruise control devices that control vehicle driving using a machine learning model (end-to-end learning model) that is trained to output operating parameters of the driving mechanism based on peripheral data that represents the vehicle's surrounding conditions. The end-to-end learning model can use peripheral data and operating parameters during manual driving as training data in learning, so it can learn more efficiently than machine learning models that are applied to individual processes. Patent Document 1 describes an autonomous vehicle driving system that uses an end-to-end learning model. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2019-153277 Summary of the Invention [Problem to be solved by the invention]

[0005] Because an end-to-end learning model outputs operational parameters based on peripheral data, the vehicle behavior controlled by the output operational parameters may not be appropriate for the vehicle. For example, if the operational parameters output by the end-to-end learning model result in sporty behavior (e.g., a relatively large absolute value of the acceleration in the vehicle's forward direction or a relatively large steering angle), that behavior may not be appropriate for a luxury vehicle, which is expected to have a milder behavior. Furthermore, an end-to-end learning model trained using training data acquired on roads where appropriate manual driving control is difficult due to complex geometries may output operational parameters that result in inappropriate behavior.

[0006] An object of the present disclosure is to provide a cruise control device that can appropriately control the behavior of a vehicle. [Means for solving the problem]

[0007] The gist of the present disclosure is as follows.

[0008] (1) A candidate generation unit that generates a candidate parameter set having one or more parameters by inputting input data including a peripheral image acquired by a peripheral image capturing unit that captures the situation around the vehicle to a classifier that has been trained in advance to output one or more parameters that control the traveling of the vehicle in response to input of the input data; and an estimation unit that estimates a future behavior of the vehicle when the vehicle's traveling is controlled by the candidate parameter set; a storage unit that stores one or more avoidance behaviors that indicate behaviors of the vehicle that should be avoided; a driving control unit that controls driving of the vehicle using the candidate parameter set when a future behavior of the vehicle predicted for the candidate parameter set does not correspond to any of the one or more avoidance behaviors, and controls driving of the vehicle without using the candidate parameter set when a future behavior of the vehicle predicted for the candidate parameter set corresponds to any of the one or more avoidance behaviors; A driving control device comprising:

[0009] (2) the candidate generator generates a plurality of the candidate parameter sets; The driving control device described in (1) above, wherein when the future behavior of the vehicle predicted for at least one candidate parameter set among the generated plurality of candidate parameter sets corresponds to one of the one or more avoidance behaviors, and the future behavior of the vehicle predicted for other candidate parameter sets among the plurality of candidate parameter sets does not correspond to any of the one or more avoidance behaviors, controls the driving of the vehicle using one of the other candidate parameter sets.

[0010] (3) A driving control device that controls the driving of a vehicle inputting input data including a peripheral image acquired by a peripheral image capturing unit that captures images of the situation around the vehicle into a classifier that has been trained in advance to output one or more parameters that control the traveling of the vehicle in response to input of the input data, thereby generating a candidate parameter set including the one or more parameters; predicting a future behavior of the vehicle when the vehicle's running is controlled by the candidate parameter set; When the future behavior of the vehicle predicted for the candidate parameter set does not correspond to any of one or more avoidance behaviors that indicate behaviors of the vehicle to be avoided and that are stored in a storage unit, the driving of the vehicle is controlled using the candidate parameter set, and when the future behavior of the vehicle predicted for the candidate parameter set corresponds to any of the one or more avoidance behaviors, the driving of the vehicle is controlled without using the candidate parameter set. A driving control method comprising:

[0011] (4) inputting input data including a peripheral image acquired by a peripheral image capturing unit that captures the situation around the vehicle into a classifier that has been trained in advance to output one or more parameters that control the traveling of the vehicle in response to input of the input data, thereby generating a candidate parameter set including the one or more parameters; predicting a future behavior of the vehicle when the vehicle is controlled by the candidate parameter set; When the future behavior of the vehicle predicted for the candidate parameter set does not correspond to any of one or more avoidance behaviors that indicate behaviors of the vehicle to be avoided and that are stored in a storage unit, controlling the traveling of the vehicle using the candidate parameter set, and when the future behavior of the vehicle predicted for the candidate parameter set corresponds to any of the one or more avoidance behaviors, controlling the traveling of the vehicle without using the candidate parameter set; A computer program for driving control that causes a computer installed in the vehicle to execute the above.

[0012] According to the cruise control device of the present disclosure, the behavior of the vehicle can be appropriately controlled. [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a schematic configuration diagram of a vehicle in which a driving control device is implemented; [Figure 2] FIG. 2 is a hardware schematic diagram of a driving control device. [Figure 3] FIG. 10 is a diagram illustrating an example of an avoidance behavior list stored in a memory. [Figure 4] 1A is a schematic diagram showing a situation in which a first avoidance behavior is detected, FIG. 1B is a graph showing the first avoidance behavior, and FIG. 1C is a graph showing parameters corresponding to the first avoidance behavior. [Figure 5] 10A is a schematic diagram showing a situation in which a second avoidance behavior is detected, FIG. 10B is a graph showing the second avoidance behavior, and FIG. 10C is a graph showing parameters corresponding to the second avoidance behavior. [Figure 6] FIG. 2 is a functional block diagram of a processor included in the driving control device. [Figure 7] 4 is a flowchart of a driving control process. DETAILED DESCRIPTION OF THE INVENTION

[0014] A cruise control device capable of appropriately controlling vehicle behavior will be described in detail below with reference to the drawings. The cruise control device stores one or more avoidance behaviors indicating vehicle behaviors to be avoided in a memory unit. The cruise control device generates a candidate parameter set having one or more parameters for controlling vehicle traveling by inputting input data including a surrounding image acquired by a surrounding image capturing unit that captures the situation around the vehicle to a classifier. The classifier is trained in advance to output one or more parameters for controlling vehicle traveling in response to input of the input data. The cruise control device predicts future behavior of the vehicle when traveling is controlled using the candidate parameter set. If the future behavior of the vehicle predicted using the candidate parameter set does not correspond to any of the one or more avoidance behaviors, the cruise control device controls the traveling of the vehicle using the candidate parameter set. On the other hand, if the future behavior of the vehicle predicted using the candidate parameter set corresponds to any of the one or more avoidance behaviors, the cruise control device controls the traveling of the vehicle without using the candidate parameter set.

[0015] FIG. 1 is a schematic diagram of a vehicle in which a driving control device is implemented.

[0016] The vehicle 1 has a front camera 2, a side sensor 3, a GNSS (Global Navigation Satellite System) receiver 4, a storage device 5, and a cruise control device 6. The front camera 2, the side sensor 3, the GNSS receiver 4, the storage device 5, and the cruise control device 6 are communicably connected via an in-vehicle network that complies with a standard such as a controller area network.

[0017] The front camera 2 is an example of a peripheral image capturing unit that captures images of the surroundings of the vehicle 1. The front camera 2 has a two-dimensional detector configured with an array of photoelectric conversion elements, such as a CCD or C-MOS, that are sensitive to visible light, and an imaging optical system that forms an image of the area to be captured on the two-dimensional detector. The front camera 2 is mounted, for example, at the upper front part of the vehicle interior, facing forward. The front camera 2 captures images of the surroundings of the vehicle 1 through the windshield at predetermined capture intervals (for example, 1 / 30 to 1 / 10 seconds), and outputs data representing the surroundings of the vehicle 1 as a peripheral image.

[0018] The side sensors 3 are an example of situation sensors that generate situation data for identifying the situation of the vehicle 1. They include a left LiDAR (Light Detection and Ranging) sensor 3-1 mounted on the left side of the vehicle 1 and a right LiDAR sensor 3-2 mounted on the right side of the vehicle 1. The left LiDAR sensor 3-1 and the right LiDAR sensor 3-2 each include a laser that generates infrared laser light and a photoreceiver that two-dimensionally scans the laser light reflected by an object and received through an optical window. The photoreceiver measures the time it takes for the emitted laser light to be reflected by the object and received, thereby generating a distance image in which each pixel has a value corresponding to the distance to the object represented by that pixel. The left LiDAR sensor 3-1 and the right LiDAR sensor 3-2 each output a distance image representing the distance to an object on the side of the vehicle 1 at a predetermined imaging period (e.g., 1 / 30 to 1 / 10 of a second). The distance images output by the left LiDAR sensor 3-1 and the right LiDAR sensor 3-2 are examples of distance information that indicate the distance to an object on the side of the vehicle 1.

[0019] The GNSS receiver 4 is another example of a situation sensor, and receives GNSS signals from GNSS satellites at predetermined intervals and determines the own position of the vehicle 1 based on the received GNSS signals. The GNSS receiver 4 outputs a positioning signal representing the result of determining the own position of the vehicle 1 based on the GNSS signals to the cruise control device 6 via the in-vehicle network at predetermined intervals.

[0020] The storage device 5 is an example of a storage unit, and includes, for example, a hard disk drive or a non-volatile semiconductor memory. The storage device 5 stores map data including information about features such as lane markings in association with positions.

[0021] The cruise control device 6 stores one or more avoidance behaviors. The cruise control device 6 generates a candidate parameter set based on a surrounding image acquired by the forward camera 2. The cruise control device 6 also predicts the behavior of the vehicle 1 when driving is controlled using the candidate parameter set. If the future behavior of the vehicle predicted for the candidate parameter set does not correspond to any of the one or more avoidance behaviors, the cruise control device 6 controls the driving of the vehicle 1 using the candidate parameter set. On the other hand, if the future behavior of the vehicle predicted for the candidate parameter set corresponds to any of the one or more avoidance behaviors, the cruise control device 6 controls the driving of the vehicle 1 without using the candidate parameter set.

[0022] 2 is a hardware schematic diagram of the driving control device 6. The driving control device 6 includes a communication interface 61, a memory 62, and a processor 63.

[0023] The communication interface 61 is an example of a communication unit, and has a communication interface circuit for connecting the driving control device 6 to an in-vehicle network. The communication interface 61 supplies received data to the processor 63. The communication interface 61 also outputs data supplied from the processor 63 to the outside.

[0024] The memory 62 is another example of a storage unit, and includes a volatile semiconductor memory and a non-volatile semiconductor memory. The memory 62 stores various data used for processing by the processor 63, for example, one or more avoidance behaviors that indicate behaviors of the vehicle 1 that should be avoided. In the present disclosure, the behavior of the vehicle 1 refers to the movement of the vehicle 1 that can be observed from outside the vehicle 1. The behavior of the vehicle 1 may include, for example, acceleration in the longitudinal direction (the direction of travel of the vehicle 1), acceleration in the lateral direction (to the right or left relative to the direction of travel of the vehicle 1), and the lateral distance to surrounding objects.

[0025] Fig. 3 is a diagram showing examples of avoidance behaviors stored in memory 62. The avoidance behavior table 621 shown in Fig. 3 lists, in tabular form, examples of combinations of one or more avoidance behaviors stored in memory 62 and the situations of vehicle 1 corresponding to each avoidance behavior.

[0026] One or more avoidance behaviors are included in the avoidance behavior table 621. For example, avoidance behavior (1) is an example of a first avoidance behavior, and indicates that the absolute value of the lateral acceleration of the vehicle 1 exceeds the acceleration threshold X1 while traveling around a curve.

[0027] Fig. 4(a) is a schematic diagram showing a situation in which the first avoidance behavior is detected, and Fig. 4(b) is a graph showing the first avoidance behavior. In the situation shown in Fig. 4(a), the vehicle 1 is traveling on a curved road RD1, and therefore acceleration occurs from the inner periphery of the curve to the outer periphery.

[0028] In graph G1B shown in FIG. 4(b), the vertical axis represents lateral acceleration (positive to the right), and the horizontal axis represents time. Vehicle 1 behavior in which the lateral acceleration changes according to the values ​​indicated by the dashed line corresponds to avoidance behavior because the absolute value of the lateral acceleration exceeds the acceleration threshold X1 (the value of the lateral acceleration is below -X1). Vehicle 1 behavior in which the lateral acceleration changes according to the values ​​indicated by the solid line does not correspond to avoidance behavior because the absolute value of the lateral acceleration does not exceed the acceleration threshold X1 (the value of the lateral acceleration does not fall below -X1).

[0029] Returning to FIG. 3, avoidance behavior (2) is an example of a second avoidance behavior, and indicates that when there is a parallel running vehicle nearby, the lateral distance between vehicle 1 and the parallel running vehicle exceeds margin threshold Y1.

[0030] Fig. 5(a) is a schematic diagram showing a situation in which the second avoidance behavior is detected, and Fig. 5(b) is a graph showing the second avoidance behavior. In the situation shown in Fig. 5(a), a parallel running vehicle 100 is running near the vehicle 1 in a lane L2 adjacent to the lane L1 on which the vehicle 1 is running. When the vehicle 1 runs near the parallel running vehicle 100, the vehicle 1 may increase the lateral distance between the vehicle 1 and the parallel running vehicle 100.

[0031] In graph G2B shown in FIG. 5(b), the vertical axis represents the horizontal distance from vehicle 1 to an object or feature present around vehicle 1, and the horizontal axis represents time. Vehicle 1's behavior in which the horizontal distance changes according to the value indicated by the dashed line corresponds to avoidance behavior because the horizontal distance exceeds margin threshold Y1. Vehicle 1's behavior in which the horizontal distance changes according to the value indicated by the solid line does not correspond to avoidance behavior because the horizontal distance does not exceed margin threshold Y1.

[0032] Returning to FIG. 3, avoidance behavior (3) indicates that the absolute value of lateral acceleration exceeds the acceleration threshold X2, the lateral distance between the vehicle 1 and an object or feature present around the vehicle 1 exceeds the margin threshold Y2, or the absolute value of longitudinal acceleration exceeds the acceleration threshold Z. The acceleration threshold X2 may be set to a value equal to or less than the acceleration threshold X1. It may be set to a value equal to or less than the margin threshold Y1. The circumstances in which avoidance behavior (3) is applied are not particularly limited. Avoidance behaviors applied in predetermined circumstances may be applied in preference to avoidance behaviors for which the circumstances in which they are applied are not particularly limited. That is, in the example of avoidance behaviors shown in the avoidance behavior table 621 shown in FIG. 3, with respect to lateral acceleration, avoidance behavior (1) is applied when driving around a curve, and avoidance behavior (3) is applied in other circumstances.

[0033] For longitudinal acceleration, the memory 62 may store different avoidance behaviors using acceleration thresholds having positive values ​​that define behavior during acceleration, and acceleration thresholds having negative values ​​that define behavior during deceleration.

[0034] Returning to FIG. 2, the memory 62 also stores various application programs, such as a cruise control computer program for causing the cruise control device 6 to execute a cruise control method.

[0035] The processor 63 is an example of a control unit and includes one or more processors and their peripheral circuits. The processor 63 may further include other arithmetic circuits such as a logic unit, a numerical calculation unit, or a graphics processing unit.

[0036] FIG. 6 is a functional block diagram of the processor 63 included in the driving control device 6. As shown in FIG.

[0037] The processor 63 of the driving control device 6 has, as functional blocks, a candidate generation unit 631, an estimation unit 632, and a driving control unit 633. Each of these units in the processor 63 is a functional module implemented by a program executed on the processor 63. A computer program that realizes the functions of each unit in the processor 63 may be provided in a form recorded on a computer-readable portable recording medium such as a semiconductor memory, a magnetic recording medium, or an optical recording medium. Alternatively, each of these units in the processor 63 may be implemented in the driving control device 6 as an independent integrated circuit, microprocessor, or firmware.

[0038] The candidate generation unit 631 acquires a peripheral image from the front camera 2 via the communication interface 61. Based on input data including the peripheral image, the candidate generation unit 631 generates a candidate parameter set having one or more parameters for controlling the traveling of the vehicle 1. In addition to the peripheral image, the input data may include data representing the traveling situation of the vehicle 1, such as vehicle speed data of the vehicle 1 acquired from a vehicle speed sensor (not shown) via the communication interface 61 and direction data of the vehicle 1 acquired from a direction sensor (not shown).

[0039] The candidate generator 631 identifies one or more parameters by inputting acquired input data into a classifier that has been trained in advance to output one or more parameters based on the input data, and sets the one or more parameters as a candidate parameter set. The one or more parameters include parameters for controlling a driving mechanism (not shown) that accelerates, decelerates, and steers the vehicle 1. The driving mechanism includes, for example, an engine or motor that supplies power to the vehicle 1, a brake that reduces the driving speed of the vehicle 1, and a steering mechanism that steers the vehicle 1. The parameters for controlling the driving mechanism include, for example, a target vehicle speed, acceleration, steering amount, etc.

[0040] The classifier may be, for example, a neural network having a convolutional neural network (CNN) in which multiple convolutional layers are connected in series from the input side to the output side, and multiple fully connected layers. The neural network operates as a classifier that outputs parameters based on the input data by training the neural network according to a predetermined learning method, such as backpropagation, using input data including a large number of surrounding images and parameters provided to the traveling mechanism at timings corresponding to the input data as training data. The classifier may output, along with the parameters, a confidence level indicating the likelihood of one or more parameters output based on the input data. The classifier may output multiple parameter sets, each including one or more parameters. For example, the classifier may output multiple parameter sets whose confidence levels are equal to or greater than a predetermined threshold.

[0041] The classifier may use, from the input side to the output side, a CNN, a recurrent neural network (RNN), and a neural network having multiple fully connected layers.

[0042] The candidate generating unit 631 may use as input data data that has been subjected to preprocessing such as resizing or normalization on the peripheral image acquired from the front camera 2 via the communication interface 61.

[0043] The prediction unit 632 predicts the future behavior of the vehicle 1 when its traveling is controlled by the candidate parameter set.

[0044] The estimation unit 632 acquires the candidate parameter set by reading out the area of ​​the memory 62 in which the candidate parameter set generated by the candidate generation unit 631 is stored. The estimation unit 632 acquires the parameters currently being used for controlling the vehicle 1 by reading out the area of ​​the memory 62 in which the parameters used for controlling the traveling mechanism are stored. The estimation unit 632 acquires, via the communication interface 61, the vehicle speed data of the vehicle 1 from a vehicle speed sensor (not shown) and the direction data of the vehicle 1 from a direction sensor (not shown), as the traveling conditions of the vehicle 1.

[0045] The estimation unit 632 acquires the future acceleration and steering amount included in the candidate parameter set. The estimation unit 632 estimates the future vehicle speed based on the current vehicle speed and future acceleration among the parameters used for the current control of the vehicle 1. The estimation unit 632 also estimates the lateral acceleration, which is an example of the future behavior of the vehicle 1, based on the future vehicle speed and future steering amount of the vehicle 1.

[0046] The estimation unit 632 also acquires a distance image from the side sensor 3 via the communication interface 61. The estimation unit 632 identifies the closest distance among the distances represented in the acquired distance image as the current lateral distance. The estimation unit 632 estimates the lateral distance, which is an example of the future behavior of the vehicle 1, based on the current lateral distance and the future acceleration and steering amount included in the candidate parameter set.

[0047] The estimation unit 632 may identify the position of the parallel running vehicle from the peripheral image acquired from the front camera 2, and estimate the future lateral distance based on the identified position of the parallel running vehicle and the future steering amount. For example, the estimation unit 632 inputs the peripheral image into a classifier that has been trained in advance to detect areas corresponding to objects such as vehicles from the image, and identifies an area corresponding to the parallel running vehicle in the peripheral image. The estimation unit 632 references the imaging parameters stored in the memory 62, such as the angle of view of the front camera 2 and the mounting direction on the vehicle 1, and estimates the direction in which the parallel running vehicle exists relative to the vehicle 1. The estimation unit 632 compares the size of a standard vehicle stored in the memory 62 with the size of the area corresponding to the parallel running vehicle identified in the peripheral image, thereby estimating the distance from the vehicle 1 to the parallel running vehicle and determining the lateral distance.

[0048] The driving control unit 633 controls the driving of the vehicle 1 by sending control signals to the driving mechanism of the vehicle 1 via the communication interface 61. The driving control unit 633 first determines whether or not the future behavior of the vehicle 1 predicted for the candidate parameter set corresponds to any one or more avoidance behaviors stored in the memory 62. If the future behavior of the vehicle 1 does not correspond to any one or more avoidance behaviors, the driving control unit 633 controls the driving of the vehicle 1 using the candidate parameter set. On the other hand, if the future behavior of the vehicle 1 corresponds to any one or more avoidance behaviors, the driving control unit 633 controls the driving of the vehicle 1 without using the candidate parameter set.

[0049] The traveling control unit 633 acquires the future behavior of the vehicle 1 by reading out an area of ​​the memory 62 in which the future behavior of the vehicle 1 estimated by the estimation unit 632 is stored. The traveling control unit 633 reads out one or more avoidance behaviors stored in the memory 62.

[0050] The driving control unit 633 also identifies the situation in which the vehicle 1 is traveling. For example, the driving control unit 633 receives a positioning signal from the GNSS receiver 4 via the communication interface 61, and acquires map information about the vicinity of the vehicle's own position indicated in the received positioning signal from the storage device 5. The driving control unit 633 determines whether the vehicle 1 is traveling around a curve based on the acquired map information. The driving control unit 633 also acquires a range image from the lateral sensor 3 via the communication interface 61, and determines whether there is a parallel traveling vehicle based on the acquired range image.

[0051] The traveling control unit 633 determines whether the acquired future behavior of the vehicle 1 corresponds to any one of the one or more read avoidance behaviors. At this time, the traveling control unit 633 may first identify an avoidance behavior that corresponds to the situation in which the vehicle 1 is traveling from the one or more read avoidance behaviors, and determine whether the acquired future behavior of the vehicle 1 corresponds to any one of the identified avoidance behaviors.

[0052] If the future behavior of the vehicle 1 does not correspond to any of the one or more avoidance behaviors, the driving control unit 633 transmits a control signal representing a candidate parameter set to the driving mechanism of the vehicle 1 via the communication interface 61. On the other hand, if the future behavior of the vehicle 1 corresponds to any of the one or more avoidance behaviors, the driving control unit 633 transmits a control signal not representing a candidate parameter set to the driving mechanism. At this time, the driving control unit 633 may transmit, as a control signal not representing a control parameter set, a parameter set in which at least one of the one or more parameters included in the candidate parameter set is changed so that the future behavior of the vehicle 1 does not correspond to any of the one or more avoidance behaviors. The driving control unit 633 can determine whether the future behavior of the vehicle 1 estimated by the estimation unit 632 for the changed parameter set corresponds to any of the one or more avoidance behaviors. Therefore, the driving control unit 633 may repeat the parameter change and behavior estimation until the future behavior for the changed parameter set does not correspond to any of the one or more avoidance behaviors.

[0053] First, an example of driving control corresponding to the first avoidance behavior will be described. It is assumed that the behavior shown by the dashed line in graph G1B shown in Fig. 4(b) is predicted for the candidate parameter set generated in the situation shown in Fig. 4(a). As described above, the behavior shown by the dashed line corresponds to the avoidance behavior, and therefore the driving control unit 633 controls the driving of the vehicle 1 without using the candidate parameter set.

[0054] FIG. 4(c) is a graph showing parameters corresponding to the first avoidance behavior.

[0055] In graph G1C shown in FIG. 4(c), the vertical axis represents the target vehicle speed, and the horizontal axis represents time. The dashed line represents the target vehicle speed included in the candidate parameter set. The solid line represents the target vehicle speed changed by the cruise control unit 633. It is estimated that the changed parameter set will cause the behavior of the vehicle 1 to be as shown by the solid line in graph G1B shown in FIG. 4(b). As described above, the behavior shown by the solid line in graph G1B does not correspond to avoidance behavior, and therefore the cruise control device 6 can appropriately control the driving of the vehicle 1.

[0056] Next, an example of driving control corresponding to the second avoidance behavior will be described. It is assumed that the behavior shown by the dashed line in graph G2B shown in Fig. 5(b) is predicted for the candidate parameter set generated in the situation shown in Fig. 5(a). As described above, the behavior shown by the dashed line corresponds to the avoidance behavior, and therefore the driving control unit 633 controls the driving of the vehicle 1 without using the candidate parameter set.

[0057] FIG. 5(c) is a graph showing parameters corresponding to the second avoidance behavior.

[0058] In graph G2C shown in FIG. 5(c), the vertical axis represents the steering amount, and the horizontal axis represents time. The dashed line represents the steering amount included in the candidate parameter set. The solid line represents the steering amount changed by the cruise control unit 633. It is estimated that with the changed parameter set, the behavior of the vehicle 1 will be as shown by the solid line in graph G2B shown in FIG. 5(b). As described above, the behavior shown by the solid line in graph G2B does not correspond to avoidance behavior, and therefore the cruise control device 6 can appropriately control the driving of the vehicle 1.

[0059] 7 is a flowchart of the driving control process. The processor 63 of the driving control device 6 repeatedly executes the driving control process described below at a predetermined cycle (for example, every 1 / 10 seconds) while the vehicle 1 is driving automatically.

[0060] First, the candidate generator 631 of the processor 63 of the driving control device 6 generates a candidate parameter set by inputting input data including a surrounding image acquired by the front camera 2 to a classifier (step S1).

[0061] The estimation unit 632 of the processor 63 estimates the future behavior of the vehicle 1 when its traveling is controlled by the candidate parameter set (step S2).

[0062] The traveling control unit 633 of the processor 63 determines whether or not the future behavior of the vehicle 1 predicted for the candidate parameter set corresponds to one or more avoidance behaviors (step S3).

[0063] If the future behavior of vehicle 1 corresponds to one or more avoidance behaviors (step S3: Y), driving control unit 633 controls the driving of vehicle 1 without using the candidate parameter set (step S4) and terminates the driving control process.

[0064] If the future behavior of vehicle 1 does not correspond to any of one or more avoidance behaviors (step S3: N), the driving control unit 633 controls the driving of vehicle 1 using the candidate parameter set (step S5) and terminates the driving control process.

[0065] By executing the driving control process in this manner, the driving control device 6 can appropriately control the behavior of the vehicle.

[0066] It should be understood that those skilled in the art can make various changes, substitutions, and alterations thereto without departing from the spirit and scope of the present disclosure. [Explanation of symbols]

[0067] 1 vehicle 6. Driving control device 631 Candidate generation section 632 Guessing part 633 Driving control unit

Claims

1. a storage unit that stores one or more avoidance behaviors that indicate vehicle behaviors that should be avoided; a candidate generation unit that generates a plurality of candidate parameter sets each having one or more parameters by inputting input data including a peripheral image acquired by a peripheral image capturing unit that captures an image of the situation around the vehicle into a classifier that has been trained in advance to output one or more parameters that control the traveling of the vehicle in response to input of the input data; and an estimation unit that estimates future behaviors of the vehicle when the vehicle's running is controlled by the plurality of candidate parameter sets; a travel control unit that, when a future behavior of the vehicle predicted for one of the plurality of candidate parameter sets does not correspond to any of the one or more avoidance behaviors, controls travel of the vehicle using the one candidate parameter set, and, when a future behavior of the vehicle predicted for the one candidate parameter set corresponds to any of the one or more avoidance behaviors, controls travel of the vehicle without using the one candidate parameter set, when a future behavior of the vehicle predicted for the one candidate parameter set corresponds to any one of the one or more avoidance behaviors, and a future behavior of the vehicle predicted for another candidate parameter set different from the one candidate parameter set among the plurality of candidate parameter sets does not correspond to any one of the one or more avoidance behaviors, the traveling control unit controls the traveling of the vehicle using the other candidate parameter set. Driving control device.

2. A driving control device that controls the driving of a vehicle, inputting input data including a peripheral image acquired by a peripheral image capturing unit that captures an image of the situation around the vehicle into a classifier that has been trained in advance to output one or more parameters that control the traveling of the vehicle in response to input of the input data, thereby generating a plurality of candidate parameter sets including the one or more parameters; predicting future behaviors of the vehicle when the vehicle's running is controlled by the plurality of candidate parameter sets; When a future behavior of the vehicle predicted for one of the plurality of candidate parameter sets does not correspond to any of one or more avoidance behaviors that indicate a behavior of the vehicle to be avoided and that are stored in a storage unit, controlling traveling of the vehicle using the one candidate parameter set, and when the future behavior of the vehicle predicted for the one candidate parameter set corresponds to any of the one or more avoidance behaviors, controlling traveling of the vehicle without using the one candidate parameter set, In the control of the traveling, if a future behavior of the vehicle predicted for the one candidate parameter set corresponds to any one of the one or more avoidance behaviors, and a future behavior of the vehicle predicted for another candidate parameter set different from the one candidate parameter set among the plurality of candidate parameter sets does not correspond to any one of the one or more avoidance behaviors, the traveling of the vehicle is controlled using the other candidate parameter set. Driving control method.

3. inputting input data including a peripheral image acquired by a peripheral image capturing unit that captures images of the situation around the vehicle into a classifier that has been trained in advance to output one or more parameters that control the traveling of the vehicle in response to input of the input data, thereby generating a plurality of candidate parameter sets including the one or more parameters; predicting future behaviors of the vehicle when the vehicle's running is controlled by the plurality of candidate parameter sets; When a future behavior of the vehicle predicted for one of the plurality of candidate parameter sets does not correspond to any of one or more avoidance behaviors that indicate a behavior of the vehicle to be avoided and that are stored in a storage unit, controlling the traveling of the vehicle using the one candidate parameter set, and when the future behavior of the vehicle predicted for the one candidate parameter set corresponds to any of the one or more avoidance behaviors, controlling the traveling of the vehicle without using the one candidate parameter set; causing a computer mounted on the vehicle to execute the above; In the control of the traveling, if a future behavior of the vehicle predicted for the one candidate parameter set corresponds to any one of the one or more avoidance behaviors, and a future behavior of the vehicle predicted for another candidate parameter set different from the one candidate parameter set among the plurality of candidate parameter sets does not correspond to any one of the one or more avoidance behaviors, the traveling of the vehicle is controlled using the other candidate parameter set. Computer program for driving control.

Images