JP7503429B2 - Parking assistance device and parking assistance method - Google Patents

Description

This invention relates to a parking assistance device and a parking assistance method.

Some vehicles, such as automobiles, are equipped with parking assistance devices to assist with parking (see, for example, Patent Document 1). The parking assistance device of Patent Document 1 uses the situation around the vehicle and a parking frame set from the situation around the vehicle to generate a parking path from the vehicle's position to a parking frame, and moves the vehicle along the parking path to park the vehicle in the parking frame. The parking assistance device also predicts whether the vehicle will collide with an obstacle as it moves along the parking path to a parking frame. If it predicts that the vehicle will collide with an obstacle, it slows down or stops the vehicle to avoid the collision.

JP 2015-81022 A

In the parking assistance device described in Patent Document 1, when the recognition area for the situation around the vehicle expands in the direction of the vehicle's movement as the vehicle moves toward a parking space, the position of the parking space is updated in the direction of movement to match the expanded recognition area, so that the parking space is corrected to the correct position.

On the other hand, the range for collision prediction (collision prediction range) is set to the range of the parking route that was initially generated, and as described above, even if the parking space is updated from its initial position to the correct position in the direction of travel of the vehicle, the range remains within the range of the parking route that was initially generated. This causes an issue in that the initial collision prediction range is exceeded while the vehicle is moving toward a parking space that has been updated to the correct position in the direction of travel, and collision prediction stops working midway through the process, even though the vehicle is still moving.

Therefore, once collision prediction is no longer functioning, an external emergency braking device installed separately from the parking assistance device will be activated instead for obstacles ahead in the direction of the vehicle's movement.

However, because emergency braking devices apply the emergency brakes to bring the vehicle to an emergency stop, there are problems with the vehicle's ride being uncomfortable when parking with parking assistance, causing great anxiety for the occupants when the emergency braking device is activated.

Therefore, the present invention was made in consideration of these problems.

In response to the above problems, the present invention provides:
A recognition unit that recognizes a situation around the vehicle;
A parking space recognition unit that sets a parking space using the recognition result of the recognition unit;
A route generating unit that generates a parking route from the position of the vehicle to the parking space using the surrounding situation of the vehicle recognized by the recognition unit and the parking space set by the parking space recognition unit;
a vehicle control unit that performs vehicle control for parking assistance based on the parking path generated by the path generation unit,
a collision prediction unit that predicts whether the host vehicle will collide with an obstacle when the host vehicle moves along the parking path, using the parking path generated by the path generation unit and the recognition result of the recognition unit, during a period from the start to the end of the vehicle control by the vehicle control unit,
When the recognition area of the recognition unit expands in the moving direction of the vehicle as the vehicle moves, the parking space recognition unit updates the parking space in the moving direction.
The parking assistance device is characterized in that the collision prediction unit performs collision prediction according to the parking space updated by the parking space recognition unit in the moving direction of the host vehicle.

According to the present invention, with the above-mentioned configuration, even if the recognition area of the recognition unit is expanded and the initial parking space obtained by the parking space recognition unit is updated in the direction of movement of the vehicle, the collision prediction unit can properly perform collision prediction to the end in accordance with the latest updated parking space from the start to the end of vehicle control by the vehicle control unit.

1 is a device configuration diagram of a parking assistance system for a vehicle equipped with parking assistance according to an embodiment of the present invention. FIG. 2 is a diagram showing a state in which a vehicle is parked by the parking assistance device of FIG. 1 . FIG. 2 is a diagram showing a state in which an emergency brake device is activated to bring the vehicle to an emergency stop. FIG. 2 is a functional configuration diagram of the parking assistance device of FIG. 1 . 3A to 3C are schematic diagrams showing examples of a surrounding image, an obstacle map, and a superposition map. FIG. 11 is a diagram showing the relationship between the recognition area of the recognition unit and the initial parking frame. FIG. 4 is a flow chart showing parking assistance processing. 13 is a diagram showing how the parking position correction unit corrects the parking position in the moving direction of the host vehicle when the recognition area of the recognition unit expands due to the movement of the host vehicle; FIG. 13A and 13B are diagrams illustrating a state in which a collision prediction range of a collision prediction unit is corrected in a moving direction. FIG. 11 is a diagram showing how a collision prediction unit performs collision prediction within the range of an initial parking path. 11A and 11B are diagrams illustrating how a collision prediction unit corrects a collision prediction range. 11 is a flowchart showing a collision prediction process of a collision prediction unit. FIG.

Hereinafter, the present embodiment will be described in detail with reference to the drawings.
1 to 12 are intended to explain this embodiment.

<Configuration> The configuration of this embodiment is explained below.

Figure 1 shows the device configuration of a parking assistance system installed in a vehicle. This parking assistance system includes a parking assistance device 3 and a vehicle control device unit 4, as well as sensors such as a vehicle state detection unit 5, a surroundings detection unit 6, and a surroundings photography unit 7, and an HMI unit 8. These are connected to each other via an in-vehicle network such as a Controller Area Network (CAN), or are connected directly to each other.

In addition to the parking assistance system described above, an emergency braking device 9 is connected to an in-vehicle network such as CAN, and the emergency braking device 9, the vehicle control device unit 4, and the surrounding detection unit 6 constitute an emergency braking system.

Figure 2 is a diagram showing how a vehicle equipped with the parking assistance device 3 described above (hereinafter referred to as the vehicle 1) parks using the functions of the parking assistance device 3. A parking lot 11 has multiple parallel parking stall lines 12, and a parking space 13 is formed between two adjacent parking stall lines 12. In this embodiment, an obstacle 14 such as a pole or a trash can is placed at the back of the parking space 13.

The parking assistance device 3 described above assists the driver by automatically or semi-automatically parking the vehicle 1 in the target parking space 13 when a start command is given by an occupant such as the driver in the parking lot 11. The vehicle 1 stops once in the parking lot 11, moves forward a predetermined distance as indicated by arrow a, and moves backward as indicated by arrow b, and is parked (entered) in the target parking space 13. This makes it possible for the vehicle 1 to be parked in the parking space 13 without the driver performing (all) driving operations. Parking assistance can be divided into parking assistance in the narrow sense, in which the driver performs some driving operations, and parking assistance in the broad sense, which includes automatic parking, in which the vehicle is parked fully automatically. The parking assistance in this embodiment is parking assistance in the broad sense. Below, an example of automatic parking will be mainly described.

For example, if the parking assistance device 3 does not operate properly for an obstacle 14 at the back of the parking space 13, the emergency braking device 9 operates instead to bring the vehicle 1 to an emergency stop, as shown in FIG. 3. The emergency braking device 9 brings the vehicle 1 to an emergency stop when the distance between the vehicle 1 and the obstacle 14 falls below a certain value. For convenience, the search waves from the surroundings detection unit 6 for the emergency braking device 9 are shown in the figure as being from the emergency braking device 9.

And FIG. 4 is a diagram showing a specific configuration of the parking assistance system described above. Among these, the parking assistance device 3 constitutes a main part of the parking assistance system, and is an in-vehicle device that controls the vehicle control device unit 4 to execute fully automatic parking control and semi-automatic parking assistance control. Among these, the automatic parking control is a control for automatically driving the host vehicle 1 without the driver performing driving operations such as steering, accelerator operation, or brake operation, and for parking the host vehicle 1 into the target parking space 13 and completing parking of the host vehicle 1. Also, the semi-automatic parking assistance control is a control for driving the host vehicle 1 into the target parking space 13 with at least one driving operation such as steering, accelerator operation, or brake operation by the driver.

The parking assistance device 3 has a computer (ECU (Electric Control Unit) in this embodiment) that includes a processor such as a CPU (Central Processing Unit) or MPU (Micro-Processing Unit), a memory device (also called a main memory device) such as a ROM (Read Only Memory) or RAM (Random Access Memory), a storage device (also called an auxiliary memory device) such as a HDD (hard disk drive) or SSD (Solid State Drive), an interface circuit for connecting sensors and peripheral devices, and an in-vehicle network communication circuit that communicates with other in-vehicle devices via an in-vehicle network. In the parking assistance device 3, the processor executes a computer program stored in the memory device or storage device to realize various functional configurations for automatic parking control and parking assistance control. The functional configuration of the parking assistance device 3 will be described later.

The vehicle control device 4 has various actuators that control the driving of the vehicle 1 in response to a control signal 15 from the parking assistance device 3. These actuators include at least a throttle actuator, a brake actuator, and a steering actuator.

When the host vehicle 1 is a general vehicle (engine vehicle), the throttle actuator controls the amount of air supplied to the engine (throttle opening) in response to a control signal 15 from the parking assistance device 3, thereby controlling the driving force of the host vehicle 1. If the host vehicle 1 is a hybrid vehicle, in addition to the amount of air supplied to the engine, a control signal 15 from the parking assistance device 3 is input to a motor serving as a power source to control the driving force. If the host vehicle 1 is an electric vehicle, a control signal 15 from the parking assistance device 3 is input to a motor serving as a power source instead of the throttle actuator to control the driving force. The motor serving as a power source in these cases constitutes an actuator.

The brake actuator controls a brake system provided in the host vehicle 1 in response to a control signal 15 from the parking assistance device 3, and controls the braking force applied to the wheels of the host vehicle 1. As the brake system, for example, a hydraulic brake system can be used.
The steering actuator controls the drive of an assist motor, which controls the steering torque of the electric power steering system, in response to a control signal 15 from the parking assistance device 3.

The vehicle state detection unit 5 detects the physical amount of a parameter related to the driving of the vehicle 1, or the amount of change in the physical amount, and outputs the detection result to the parking assistance device 3. The parameters include at least parameters required for dead reckoning (autonomous navigation), and the parking assistance device 3 can calculate the current position of the vehicle 1 (hereinafter referred to as "vehicle position 16" (Figure 6)) based on the detection result of the vehicle state detection unit 5.

The vehicle state detection unit 5 of this embodiment includes detectors such as a vehicle speed sensor, an acceleration sensor, and a yaw rate sensor in order to detect such parameters.
The vehicle speed sensor is a detector that detects the speed of the host vehicle 1. As the vehicle speed sensor, for example, a wheel speed sensor that is provided on the wheels of the host vehicle 1 or on an axle (drive shaft) that rotates integrally with the wheels and detects the rotation speed of the wheels is used.
The acceleration sensor is a detector that detects the acceleration of the host vehicle 1. The acceleration sensor includes, for example, a longitudinal acceleration sensor that detects the acceleration in the longitudinal direction of the host vehicle 1, and a lateral acceleration sensor that detects the lateral acceleration of the host vehicle 1.
The yaw rate sensor is a detector that detects the yaw rate (rotational angular velocity) around the vertical axis of the center of gravity of the vehicle 1. For example, a gyro sensor can be used as the yaw rate sensor. Publicly known or well-known technology can be used for dead reckoning (autonomous navigation) based on these detectors.

The periphery detection unit 6 detects obstacles 14 present around the vehicle 1 and outputs the detection results of the obstacles 14 to the parking assistance device 3. The obstacles 14 are stationary objects or moving objects that may impede the travel of the vehicle 1. The obstacles 14 are, for example, stationary objects such as curbs, wheel stops, guard rails, signs, buildings, and other parked vehicles 17 (FIG. 5), or moving objects such as other moving vehicles 17 and pedestrians. In this embodiment, the main obstacles 14 have already been described as poles and trash cans at the back of the parking space 13.

The periphery detection unit 6 includes, for example, a distance measurement sensor as a means for detecting the obstacle 14. The distance measurement sensor is a detector for measuring the distance between the vehicle 1 and the obstacle 14, and emits a search wave such as ultrasonic waves, radio waves, or light (visible light or infrared light), detects the reflected wave that is returned after the search wave is reflected by the obstacle 14, and outputs the detection result to the parking assistance device 3. In this embodiment, a sonar is used as the distance measurement sensor. The distance measurement sensor may be a millimeter wave radar or a laser radar. The periphery detection unit 6 is also used in the emergency braking device 9.

The surroundings photographing unit 7 is equipped with a camera that photographs the surroundings of the vehicle 1 and outputs the surroundings image 18 (FIG. 5) obtained by the photographing to the parking assistance device 3. The surroundings photographing unit 7 can also be used as a detection means for the surroundings detection unit 6.

The vehicle 1 can be provided with a distance measuring sensor of the periphery detection unit 6 and a camera of the periphery photographing unit 7 in an appropriate position according to the number of cameras that can detect or photograph the entire periphery of the vehicle 1 (a 360-degree range centered on the vehicle position 16).

The HMI unit 8 is equipped with an input device that serves as a user interface, and an output device. The output device includes a display device that displays various information on a screen, and a speaker and buzzer that output various information by voice. The input device also includes an operating device (such as a touch panel or an operator) for inputting instructions from the occupant.

The parking assistance device 3 described above includes a vehicle state detection result acquisition unit 21, a surrounding detection result acquisition unit 22, and a recognition unit 23 (surrounding image acquisition unit or surrounding recognition unit). The parking assistance device 3 also includes a vehicle position calculation unit 24, an obstacle map generation unit 25, and a parking space recognition unit 26. The parking assistance device 3 also includes a parking space selection unit 27, a path generation unit 28, and a vehicle control unit 29.

The vehicle condition detection result acquisition unit 21 acquires the detection result of the vehicle condition from the vehicle condition detection unit 5, the surroundings detection result acquisition unit 22 acquires the detection result of the obstacle 14 from the surroundings detection unit 6, and the recognition unit 23 acquires the surroundings image 18 from the surroundings capture unit 7.

The vehicle position calculation unit 24 calculates the vehicle position 16 described above based on the detection result of the vehicle state from the vehicle state detection unit 5. The obstacle map generation unit 25 generates an obstacle map 34 (Figure 5) based on the detection result of the obstacle 14 from the surroundings detection unit 6.

FIG. 5 is a schematic diagram showing an example of the surrounding image 18, the obstacle map 34, and a superimposition map 35, which will be described later.
The peripheral image 18 is an image of the periphery of the vehicle 1 captured by the peripheral image capturing unit 7. The peripheral image 18 is converted into an overhead image of the periphery of the vehicle 1 from a viewpoint position above the vehicle 1. A publicly known or well-known viewpoint conversion process can be used to convert the peripheral image 18 into an overhead image.

The obstacle map 34 is data showing the positions of obstacles 14 (other vehicles 17 in the illustrated example) around the host vehicle 1 .
The superimposition map 35 is data in which an overhead image as the peripheral image 18 is superimposed on the obstacle map 34 .
In this embodiment, the obstacle map 34 is formed as point cloud data in which positions 36 of points (hereinafter referred to as "reflection points") where the search wave emitted by the distance measurement sensor is reflected are mapped onto a reference X-Y coordinate system (e.g., world coordinates), as shown in FIG. 5.

More specifically, regarding the generation of the obstacle map 34, the obstacle map generating unit 25 calculates the position 36 of the reflection point based on the detection result each time it acquires the detection result of the periphery detection unit 6 through the periphery detection result acquiring unit 22. The obstacle map generating unit 25 calculates the position 36 of the reflection point by determining the distance from the vehicle position 16 to the reflection point and the direction of the reflection point as seen from the vehicle position 16 based on the detection result using a publicly known or well-known appropriate method.

Then, the obstacle map generator 25 converts the reflection point positions 36 into positions 36 on the reference X-Y coordinate system to determine the reflection point positions 36 in the reference X-Y coordinate system. As a result, the surface (shape) of the obstacle 14 is depicted in the obstacle map 34 as a collection of the reflection point positions 36.

Returning to FIG. 4, the parking space recognition unit 26 detects the parking space 13 around the vehicle 1 based on the surrounding image 18, and sets a parking frame 37 within the parking space 13 in which the vehicle 1 will be parked.

More specifically, the parking space recognition unit 26 extracts the parking space lines 12 that appear in the peripheral image 18 by image recognition. The parking space lines 12 are lines drawn on the ground (the road surface of the parking lot 11) to divide the parking spaces 13. The parking space recognition unit 26 converts the peripheral image 18 into an overhead image, and generates a superimposition map 35 in which the overhead image is superimposed on an obstacle map 34 based on the vehicle position 16 and the attitude (yaw rate) of the vehicle 1 at the time the peripheral image 18 was captured. The parking space recognition unit 26 then identifies the position of the parking space lines 12 in the superimposition map 35, thereby detecting the position of the parking space 37 in the reference X-Y coordinate system.

The parking frame selection unit 27 extracts a parking space 13 where the vehicle 1 can be parked without coming into contact with surrounding obstacles 14 from among one or more parking spaces 13 detected by the parking space recognition unit 26, by referring to the obstacle map 34 and the overlay map 35. The parking frame selection unit 27 then determines a parking space 13 (target parking space) in which to park the vehicle 1 from among the extracted parking spaces 13 based on an appropriate selection method. The selection method may be either a method in which the occupant selects a target parking space 13, or a method in which the target parking space 13 is automatically selected (without the need for instructions from the occupant). The parking frame selection unit 27 may also be functionally integrated with the parking space recognition unit 26.

The route generation unit 28 determines a specific parking route 38 (FIG. 6) along which the vehicle 1 will travel from the current vehicle position 16 to the parking frame 37 set in the target parking space 13, by referring to the obstacle map 34 and the overlay map 35.

The vehicle control unit 29 performs, for example, automatic driving control (or "automatic parking control") and parking assistance control based on the parking route 38 to automatically drive the vehicle 1 without the driver's operation and enter the parking frame 37 of the target parking space 13.

Specifically, the vehicle control unit 29 generates a control signal 15 that controls the vehicle control device unit 4 based on the parking path 38, and transmits the control signal 15 to the vehicle control device unit 4. The vehicle control device unit 4 controls the actuator for driving the host vehicle 1 according to the control signal 15, so that the host vehicle 1 automatically drives along the parking path 38 and automatically enters the parking frame 37 of the target parking space 13. This completes the basic configuration of the parking assistance device 3.

Figure 7 is a flow diagram showing the parking assistance process performed by the parking assistance device 3 described above.

This parking assistance process is executed when the vehicle 1 is traveling in the parking lot 11. That is, when the parking assistance device 3 detects entry into the parking lot 11 based on, for example, reception of a signal emitted by a navigation device provided in the vehicle 1 or equipment (such as a management device) provided in the parking lot 11, the obstacle map generation unit 25, the parking space recognition unit 26, the parking frame selection unit 27, the route generation unit 28, the vehicle control unit 29, etc. start the obstacle map generation process 41, the parking space detection process 42, and the vehicle control process 43, respectively. The obstacle map generation process 41, the parking space detection process 42, and the vehicle control process 43 are performed synchronously or asynchronously with each other.

In the obstacle map generation process 41, the obstacle map generation unit 25 generates an obstacle map 34 based on the detection results of the obstacle 14 (step Sa1).

Meanwhile, in the parking space detection process 42, the parking space recognition unit 26 detects the position of the parking space 13 based on the surrounding image 18 and sets a parking frame 37 in the parking space 13 (step Sb1). After that, the parking space recognition unit 26 repeatedly updates the position of the parking frame 37 relative to the parking space 13 at predetermined time intervals (step Sb2, FIG. 8). The updating of the position of the parking frame 37 continues until an appropriate timing, for example, when the vehicle 1 leaves the parking lot 11.

Then, in the vehicle control process 43, when the parking frame selection unit 27 determines a target parking space 13 from among the parking spaces 13, for example, according to a user operation or the like (step Sc1: Yes), the path generation unit 28 determines an initial parking path 38 based on the obstacle map 34 and the position of the parking frame 37 of the target parking space 13 (step Sc2). Next, the vehicle control unit 29 generates a control signal 15 based on the initial parking path 38 and outputs the control signal 15 to the vehicle control device unit 4 (step Sc3). This starts parking assistance control (including automatic parking control), and the host vehicle 1 starts traveling along the initial parking path 38.

The vehicle control unit 29 continues to output appropriate control signals 15 to the vehicle control device unit 4 until the vehicle 1 enters the target parking space 13 (until step Sc4 becomes Yes), that is, until the vehicle 1 performs parking assistance control and reaches the end point of the initial parking path 38.

The parking assistance device 3 further includes a parking position correction unit 31 and a collision prediction unit 32 (ACP) (Figure 4).

The parking position correction unit 31 and the collision prediction unit 32 are components related to this embodiment, and during the period T from the start to the end of parking assistance control, the parking position correction process 44 and the collision prediction process 45 are executed individually. The parking position correction process 44 and the collision prediction process 45 are executed by the parking position correction unit 31 and the collision prediction unit 32 synchronously or asynchronously with each other.

While the vehicle control unit 29 is executing parking assistance control, if the parking space recognition unit 26 changes the position of the parking space 37 (determined by the parking space selection unit 27) based on the recognition result of the recognition unit 23, the parking position correction unit 31 performs the correction process required to park the vehicle 1 in the changed parking space 37.

In more detail, while parking assistance control is being executed, the recognition unit 23 continues to capture images of the surroundings of the vehicle 1 using the surrounding image capture unit 7 to sequentially output surrounding images 18, and the parking space recognition unit 26 continues to detect the parking space 37 determined by the parking space selection unit 27 based on these surrounding images 18.

As shown in FIG. 8, while parking assistance control is being performed, the vehicle 1 gradually approaches the target parking space 13, and the distance between the two is shortened, so that the detection accuracy of the target parking space 13 based on image recognition of the peripheral image 18 also gradually increases. As a result, the position of the parking frame 37 of the target parking space 13 may shift in the movement direction 46 of the vehicle 1 from its initial position at the start of the automatic driving control (displacement amount 47).

Then, based on the detection results sequentially output by the parking space recognition unit 26, the parking position correction unit 31 sequentially obtains the difference between the latest value for the position of the parking frame 37 in the target parking space 13 and the depth position of the parking frame 37 at the beginning of the parking assistance control as a correction amount 48 for the position of the parking frame 37 (target parking position correction amount).

While executing parking assistance control, the vehicle control unit 29 corrects the control signal 15 to be sent to the vehicle control device unit 4 based on the correction amount 48 in order to direct the vehicle 1 to the position of the latest parking frame 37 in the target parking space 13, and sends the corrected control signal (corrected control signal 49, FIG. 4) to the vehicle control device unit 4. Then, the vehicle control device unit 4 drives the vehicle 1 according to the corrected control signal 49, so that the vehicle 1 travels beyond the original parking path 38 and heads toward the latest parking frame 37 located further in the moving direction 46 by the correction amount 48.

If it is difficult or impossible for the vehicle 1 to reach the latest parking space 37, the vehicle control unit 29 may output a control signal 15 to the vehicle control device unit 4 to stop the vehicle 1, and discontinue parking assistance control.

7, the parking position correction unit 31 determines the correction amount 48 of the position of the parking space 37 based on the detection result (detection result) of the parking space recognition unit 26, which is updated sequentially (step Sd1), and corrects the control signal 15 output by the vehicle control unit 29 to the vehicle control device unit 4 based on the correction amount 48 (step Sd2). The series of processes of steps Sd1 and Sd2 are repeatedly executed at predetermined time intervals. The corrected control signal (corrected control signal 49) is output to the vehicle control device unit 4 by the vehicle control unit 29 in step Sc3 of the vehicle control process 43. Then, the vehicle control device unit 4 drives the vehicle 1 according to the control signal (corrected control signal 49) as shown by the arrow c in FIG. 8, so that the vehicle 1 automatically drives the original parking path 38 by the correction amount 48 toward the latest parking space 37 located further ahead in the movement direction 46, regardless of the original parking path 38, and the vehicle 1 enters the final parking space 37 (B).

Meanwhile, the collision prediction unit 32 predicts the occurrence of a collision between the host vehicle 1 and the obstacle 14 while the parking assist control described above is being executed. This prediction of a collision by the collision prediction unit 32 is performed separately from the correction process by the parking position correction unit 31. The collision prediction unit 32 will be described later. When the collision prediction unit 32 predicts the occurrence of a collision, it sends a signal to the vehicle control unit 29 to avoid the collision by, for example, slowing down or stopping the host vehicle 1 in front of the obstacle 14.

And in this embodiment, the following configuration can be provided:

(1) As described above (see FIGS. 4 and 6), the parking assistance device 3
A recognition unit 23 (a surrounding image acquisition unit or a surrounding recognition unit) that recognizes the surrounding situation of the vehicle 1;
A parking space recognition unit 26 that sets a parking space 37 using the recognition result of the recognition unit 23;
A route generating unit 28 that generates a parking route 38 from the position of the vehicle 1 (the vehicle position 16) to the parking space 37 (the parking position 51, FIG. 6) using the surrounding situation of the vehicle 1 recognized by the recognition unit 23 and the parking space 37 set by the parking space recognition unit 26;
It also has a collision prediction unit 32 that uses the parking path 38 generated by the path generation unit 28 and the recognition result of the recognition unit 23 to predict whether the vehicle 1 will collide with an obstacle 14 when the vehicle 1 moves along the parking path 38.

Furthermore, in this embodiment, as the vehicle 1 moves, the recognition area 52 of the recognition unit 23 expands in the moving direction 46 of the vehicle 1 (or changes from the recognition area 52 (A) on the front side of FIG. 6 to the recognition area 52 (B) on the back side of FIG. 8). In this case, when the parking space recognition unit 26 updates the position of the parking space 37 in the moving direction 46 (for example, from the position A-A' in FIG. 6 to the position B-B' in FIG. 8),
As shown in FIG. 9, the collision prediction unit 32 performs collision prediction based on the position of the parking space 37 (in the figure, the final parking space 37 (B)) updated by the parking space recognition unit 26 in the moving direction 46 of the vehicle 1 (or performs collision prediction for the corrected collision prediction range 53' (= collision prediction range 53 + corrected portion 54)).

The parking frame 37 is a rectangular frame that indicates the position where the vehicle 1 is parked during parking assistance, and is set to a shape and size that is suitable for the vehicle 1 so that the vehicle 1 can fit inside the rectangular frame. The parking frame 37 is set within the parking space 13 between a pair of adjacent parking section lines 12.

The position of the parking space 37 refers to the position of the parking space 37 itself. The position of the parking space 37 may be based on any part of the parking space 37, such as the front, back, or center, but in this embodiment, a specific position inside the parking space 37, as described below, is used as the reference. This reference position is the parking position 51. The parking position 51 is the target (target stopping position) when parking the vehicle 1 in the parking space 37.

Similarly, the position of the host vehicle 1 refers to the position of the host vehicle 1 itself. The position of the host vehicle 1 may be based on any part of the host vehicle 1, such as the front, rear, or center, but in this embodiment, a specific position on the inside of the host vehicle 1, as described below, is used as the reference. This reference position is the host vehicle position 16 described above.

Specifically, in this embodiment, the vehicle position 16 is set to the center position of the rear wheel axle. The parking position 51 of the parking space 37 is set to the position on the parking space 37 that corresponds to the center position of the rear wheel axle of the vehicle 1 when the vehicle 1 is correctly parked in the parking space 37. As a result, when the parking position 51 of the parking space 37 and the vehicle position 16 almost coincide with each other, the vehicle 1 is correctly parked in the parking space 37, and it becomes possible to express numerically whether the vehicle 1 is correctly parked in the parking space 37.

The parking path 38 is the path that the host vehicle 1 follows during parking assistance. The parking path 38 includes a forward path (arrow a in FIG. 2) and a reverse path (arrow b in FIG. 2), but in this embodiment, the reverse path is the main path. In this embodiment, the parking path 38 is represented as a sequence of points 55 that indicate the displacement of the host vehicle position 16 (the center position of the rear axle of the host vehicle 1) over time, using the host vehicle position 16 described above.

Obstacle 14 refers to an object with which the vehicle 1 moving along the parking path 38 may collide during parking assistance. In this embodiment, the target obstacles 14 are mainly poles or trash cans (Figure 2) located near the back of the target parking space 37.

The recognition area 52 of the recognition unit 23 is the area (range) recognized by the surrounding image 18 (for each frame) acquired by the recognition unit 23. The recognition area 52 is a finite range that depends on the performance of the surrounding image capturing unit 7 (camera), such as the angle of view, and therefore is a part of the range of all the conditions around the vehicle 1.

The recognition area 52 of the recognition unit 23 expands in the movement direction 46 as the vehicle 1 moves means that the position of the surrounding image capturing unit 7 (camera) is displaced in the movement direction 46 together with the vehicle 1, and the recognition area 52 of the recognition unit 23 shifts to the movement direction 46, so that it extends beyond the movement direction 46, and the situation at the back of the parking space 13 becomes (gradually) recognizable. The parking space recognition unit 26 continues to update the parking frame 37 to the latest state in accordance with the transition (expansion) of the recognition area 52.

The moving direction 46 is the direction in which the vehicle 1 moves toward the parking space 37, and in this embodiment, the moving direction 46 is the backward direction of the vehicle 1 (almost backward in the vehicle's fore-and-aft direction). The end of the moving direction 46 is the side in which the vehicle 1 moves (backs up) toward the parking space 37, which is behind the vehicle 1, or toward the back of the parking space 13 or the parking space 37.

The updated parking space 37 is the latest parking space 37 that is replaced over time by the parking space recognition unit 26 in the direction of movement 46 when the vehicle 1 reverses, as opposed to the initially generated parking space 37 (A), and ultimately becomes the parking space 37 (B) in which the vehicle 1 is actually parked.

The collision prediction range 53 is the range in which the collision prediction unit 32 performs collision prediction in the moving direction 46. As shown in FIG. 10, the collision prediction range 53 is set to be within the range of the initially generated parking path 38 (collision prediction range 53 = state of the initial parking path 38). That is, in the basic setting, the collision prediction unit 32 performs collision prediction only within the range of the initial parking path 38. The collision prediction is performed, for example, by calculating the trajectory of the host vehicle 1 to determine whether the four corners of the host vehicle 1, which is approximately rectangular in plan view, will intersect with the obstacle 14 as the host vehicle 1 moves.

Correcting the collision prediction range 53 in the movement direction 46 means that the collision prediction range 53 is essentially corrected in the movement direction 46 so that it is wider than the range of the original parking path 38 (the corrected collision prediction range 53' (FIG. 9) is made larger than the original parking path 38). This makes it possible for the range in which a collision can be predicted to extend to the corrected collision prediction range 53', and the collision prediction unit 32 can make collision predictions beyond the range of the original parking path 38, and beyond the original parking path 38 in the movement direction 46.

The collision prediction range 53 may be modified in the moving direction 46 in any manner as long as it is substantially wider than the range of the original parking path 38 and collisions can be predicted further in the moving direction 46. For example, the collision prediction unit 32 may independently recalculate the parking path to generate a new parking path that is longer in the moving direction 46 than the original parking path 38, and modify the collision prediction range 53 using the new parking path, but in this embodiment, the collision prediction range 53 is substantially modified in a manner described below.

The collision prediction range 53 may be corrected, for example, successively after the generation of the initial parking path 38, or when the vehicle 1 starts to move (back up) toward the initial parking space 37, or when the parking space 37 is updated by the parking space recognition unit 26, or after the position of the vehicle 1 (vehicle position 16) has passed the initial parking path 38, or after the final parking space 37 (B) has been determined. In this embodiment, the collision prediction range 53 is corrected successively after the vehicle 1 starts to move.

To explain the above in more detail, FIG. 6 shows the relationship between the recognition area 52 of the recognition unit 23 and the initial parking space 37(A). According to this figure, initially, the recognition area 52(A) of the recognition unit 23 extends to the front part of the parking space line 12 (almost half of it). The remaining part of the parking space line 12 (almost half of it) is an unrecognized area 57 that is not yet within the recognition area 52 of the recognition unit 23.

In this way, when only the front portion of the parking space line 12 can be recognized, the initial parking space 37(A) is not set to the position where the vehicle 1 will be finally parked, but is set to a position closer to the final position as shown in FIG. 8. Similarly, the initially generated parking path 38 also ends up (insufficient) just before the final position of the parking space 37(B) in the direction of movement 46.

Therefore, at the beginning of automatic parking or parking assistance, the vehicle 1 is moved (backed up) along the initial (insufficient) parking path 38 generated for this initial incorrect parking space 37(A) so as to park in the initial parking space 37(A). The collision prediction unit 32 also predicts a collision with the initial parking space 37(A) using the initial (insufficient) parking path 38 as a collision prediction range 53.

As the vehicle 1 moves (backs up) due to parking assistance, the recognition area 52 of the recognition unit 23 expands in the movement direction 46 as shown in FIG. 8, and the unrecognized area 57 (FIG. 6) also gradually decreases. In the figure, the recognition area 52 of the recognition unit 23 expands to most of the rear side of the parking space line 12.

As a result, the parking space recognition unit 26 updates the position of the latest parking space 37 in the movement direction 46 relative to the initial parking space 37 (A), and the unrecognized area 57 is almost completely eliminated, thereby obtaining the final parking space 37 (B).

In this case, as shown in the comparative example of FIG. 10, if the collision prediction range 53 of the collision prediction unit 32 remains within the range of the original parking path 38, a problem occurs in which collision prediction is no longer performed when the vehicle 1 passes beyond the original parking path 38 (an unpredictable range 58 occurs) even if the vehicle 1 is moving in the movement direction 46 toward the updated (latest) parking space 37 or the parking position 51 of the final parking space 37 (B).

Therefore, in this embodiment, as shown in FIG. 9, the collision prediction unit 32 can correct the collision prediction range 53 in the movement direction 46 in accordance with the (latest) parking space 37 updated in the movement direction 46 by the parking space recognition unit 26 and the final parking position 51 of the parking space 37 (B) (a corrected collision prediction range 53' is obtained).

As a result, even if the position of the initial parking space 37 (A) is updated in the movement direction 46, there is no longer a range (unpredictable range 58) in which collision prediction is not performed, and the collision prediction unit 32 will perform collision prediction over the entire range from when the vehicle 1 moves toward the (latest) parking space 37 updated in the movement direction 46 by the parking space recognition unit 26 until the vehicle 1 is parked (entered) in the final parking space 37 (B).

(2) When the end position 55a of the parking path 38 is updated in the moving direction 46 of the vehicle 1, the collision prediction unit 32 may perform a collision prediction for the parking space 37 corresponding to the updated end position 55a of the parking path 38.

Here, in Figures 9 and 10, as described above, the parking path 38 is represented by a series of multiple points 55. These multiple points 55 indicate the positions that the host vehicle position 16 must pass through as the host vehicle 1 moves over time. The end position 55a is the last point 55 of the multiple points 55 that indicate the parking path 38. The original parking path 38 is held independently inside the collision prediction unit 32. The points 55 that make up the original parking path 38 and the end position 55a are also held inside the collision prediction unit 32 as known positions.

When the parking space 37 is updated, information such as the updated (latest) parking space 37 and the final parking position 51 of the parking space 37 (B) can be obtained from each part of the parking assistance device 3. The updated parking position 51 can be obtained, for example, from the recognition result of the recognition unit 23, the recognition result of the parking space recognition unit 26, or the correction result of the parking position correction unit 31. Either of these may be used, but in this embodiment, the parking position 51 is obtained directly from the recognition result of the recognition unit 23 or the recognition result of the parking space recognition unit 26. For example, the recognition result of the recognition unit 23 and the recognition result of the parking space recognition unit 26 are converted into image data (peripheral image 18) such as an overhead image, and the situation around the vehicle 1 is recognized by performing image recognition on this image data.

For example, by image recognition, the parking space line 12 is recognized as a "line segment." The vehicle 1 is recognized as a "rectangle." The position of the vehicle 1 (vehicle position 16) is recognized as a "point" indicating the center position of the rear wheel axle. The parking space 37 is recognized as a "rectangle" large enough to accommodate the vehicle 1. The position of the parking space 37 (parking position 51) is recognized as a "point" indicating a position equivalent to the center position of the rear wheel axle of the vehicle 1 parked in the parking space 37. Then, when the "point" indicating the position of the vehicle 1 (vehicle position 16) reaches the position of the "point" indicating the parking position 51 of the parking space 37 in the moving direction 46 (when it overlaps or almost coincides), it is determined that the vehicle 1 is correctly parked in the parking space 37 (B).

When the parking space 37 is updated to the moving direction 46 of the vehicle 1, the collision prediction unit 32 may update the end position 55a to the moving direction 46 by moving the end position 55a of the original parking path 38 to the parking position 51 of the parking space 37 updated to the moving direction 46. By updating the end position 55a of the parking path 38 to the moving direction 46 of the vehicle 1, the collision prediction unit 32 becomes able to predict a collision with the parking space 37 corresponding to the end position 55a of the updated parking path 38 as described above.

Moving the end position 55a of the original parking path 38 to the parking position 51 of the updated (latest) parking space 37 or the final parking space 37(B) means displacing (shifting or offsetting) the end position 55a of the original parking path 38 to the parking position 51 of the updated (latest) parking space 37 or the final parking space 37(B) or its vicinity (offset position 55b). For this purpose, the collision prediction unit 32 performs a process of replacing the coordinates of the end position 55a of the original parking path 38 with the coordinates of the parking position 51 of the updated (latest) parking space 37 or the final parking space 37(B). Note that the offset position 55b may be slightly shifted from the parking position 51.

In this embodiment, the distance between the two points from the end point position 55a of the original parking path 38 to the parking position 51 of the updated (latest) parking space 37 or the final parking space 37(B) is calculated, and the position obtained by linearly extending the original parking path 38 in the moving direction 46 by the calculated distance is set as the offset position 55b (relative to the updated latest parking space 37 or the final parking space 37(B)), thereby correcting the collision prediction range 53 (or obtaining a corrected collision prediction range 53'). In this way, by offsetting the end point position 55a of the original parking path 38 to a position on the linear extension of the parking path 38, it is possible to further simplify and simplify the collision prediction process.

Specifically, as shown in FIG. 11 , when the coordinates of the end point position 55a of the original parking path 38 are (xt, yt), the attitude of the vehicle 1 at that time is (at), the coordinates of the updated (latest) parking space 37 or the parking position 51 of the final parking space 37(B) are (xr, yr), and the attitude of the vehicle 1 at that time is (ar), the distance L between the two points from the end point position 55a of the original parking path 38 to the updated (latest) parking space 37 or the parking position 51 of the final parking space 37(B) is expressed as follows:
It is calculated using the formula L = √((xt - xr)² + (yt - yr)²).

Next, the length Z of an extension line obtained by linearly extending the original parking path 38 in the moving direction 46 by an amount corresponding to the calculated distance L is calculated as follows:
It is calculated using the formula Z=L×cos(at-ar).

Then, the coordinates (xf, yf) of a position that is a length Z away from the end position 55a (xt, yt) of the parking path 38 are obtained by decomposing a vector with a length of Z and an orientation of (at) into X and Y components on the reference X-Y coordinate system.

By doing this, the collision prediction unit 32 can obtain the same effect as essentially correcting the collision prediction range 53 to the parking position 51 of the updated (latest) parking space 37 (or its vicinity) while maintaining the basic setting of the collision prediction that sets the original parking path 38 as the collision prediction range 53, by simply moving (replacing) the end position 55a of the original parking path 38 to the parking position 51 of the updated (latest) parking space 37 or the final parking space 37 (B) (making collision prediction possible up to the corrected collision prediction range 53'), and collision prediction can be continued without interruption up to the position of the coordinates (xf, yf) of the offset position 55b.

In this embodiment, the collision prediction unit 32 independently modifies the collision prediction range 53 by moving (offsetting) the end position 55a in the movement direction 46.

(3) As already described above in FIG. 4, the parking assistance device 3 includes a parking position correction unit 31 that calculates a correction amount 48 of the parking position 51 relative to the parking space 37 based on the parking space 37 updated in the moving direction 46 of the vehicle 1 by the parking space recognition unit 26;
The parking system includes a vehicle control unit 29 that controls the movement of the vehicle 1 using the parking path 38 and the correction amount 48 of the parking position 51 determined by the parking position correction unit 31.
Then, when the vehicle control unit 29 corrects the control of the vehicle 1 using the correction amount 48 of the parking position 51, the collision prediction unit 32 may perform collision prediction in accordance with the updated parking space 37.

Here, as described above, the parking position correction unit 31 is configured to send to the vehicle control unit 29 a correction amount 48 (target parking position correction amount) of the parking position 51 in the moving direction 46 that matches the updated (latest) parking space 37 when the position of the parking space 37 is updated in the moving direction 46 of the vehicle 1. The correction of the parking position 51 in the moving direction 46 by the parking position correction unit 31 is performed independently by the parking position correction unit 31. In this embodiment, the parking position correction unit 31 determines the correction amount 48 of the parking position 51 based on the initial parking path 38, etc.

The parking position correction unit 31 may be configured to correct the parking position 51 in the moving direction 46 using data (corrected collision prediction range 53') obtained by correcting the initial collision prediction range 53 by the collision prediction unit 32 in the moving direction 46, and may be configured to do so. Alternatively, conversely to the above, the collision prediction unit 32 may be configured to correct the collision prediction range 53 in the moving direction 46 using the correction amount 48 of the parking position 51 in the moving direction 46 by the parking position correction unit 31, and may be configured to do so. However, since the parking position correction unit 31 and the collision prediction range 53 have different purposes, it is preferable that the correction process of the parking position 51 in the moving direction 46 by the parking position correction unit 31 and the correction process of the collision prediction range 53 in the moving direction 46 by the collision prediction unit 32 are performed independently, as described above.

As described above, the vehicle control unit 29 creates a control signal (corrected control signal 49, FIG. 4) in which the original parking path 38 is corrected by the correction amount 48 of the parking position 51 in the moving direction 46 based on the parking path 38 from the path generation unit 28 and the correction amount 48 of the parking position 51 in the moving direction 46 from the parking position correction unit 31, and sends the corrected control signal (corrected control signal 49) to the vehicle control device unit 4, causing the vehicle control device unit 4 to correct the control of the movement of the vehicle 1 to match the updated (latest) parking frame 37 and the final parking frame 37 (B). As a result, the vehicle 1 can move in the moving direction 46 (rear of the vehicle) beyond the parking path 38 with respect to the parking position 51 of the updated (latest) parking frame 37 in the moving direction 46 and the final parking frame 37 (B), and the vehicle 1 is parked in the correct parking frame 37.

At this time, the vehicle control unit 29 may, for example, use a predetermined threshold value as a reference and, if the correction amount 48 of the parking position 51 in the movement direction 46 is smaller than the threshold value, output the control signal 15 corresponding to the original parking path 38 to the vehicle control equipment unit 4 without correcting the control signal 15, and may correct the control signal 15 and output the corrected control signal 49 to the vehicle control equipment unit 4 only when the correction amount 48 of the parking position 51 becomes larger than the threshold value.

As a result, even if the parking position correction unit 31 generates a correction amount 48 of the parking position 51 in the movement direction 46 by updating the position of the parking space 37 in the movement direction 46, for example, if the displacement amount 47 (FIG. 8) of the parking position 51 in the movement direction 46 is slight and the correction amount 48 is small enough to be negligible, the vehicle control unit 29 can avoid particularly correcting the control signal 15. In this case, the threshold value is appropriately set, for example, to be a boundary value between a value at which the magnitude of the displacement amount 47 can be ignored and a value at which it cannot be ignored.

In accordance with this, the vehicle control unit 29 may send information (correction information 72) on whether the control signal 15 has actually been corrected to the collision prediction unit 32, and the collision prediction unit 32 may receive the correction information 72 from the vehicle control unit 29. If the correction information 72 from the vehicle control unit 29 has not corrected the control signal 15, the collision prediction unit 32 determines that there is no substantial effect on the collision prediction, and leaves the collision prediction range 53 as the original parking path 38 and does not correct the collision prediction range 53. As a result, the collision prediction range 53 is corrected only when the control signal 15 has actually been corrected. Note that the correction information 72 may be generated only when the control signal 15 is corrected, for example, or may be generated periodically as either "corrected" or "not corrected".

(4) As described above, the parking assistance device 3 includes the obstacle map generator 25 that generates the obstacle map 34 using the recognition result of the recognition unit 23.
The collision prediction unit 32 performs collision prediction using the parking path 38 and the obstacle map 34 generated by the obstacle map generation unit 25 .
Then, the obstacle map generating unit 25 may update the obstacle map 34, and the collision predicting unit 32 may use the updated obstacle map 34 to perform collision prediction (for the corrected collision prediction range 53').

Here, successively updating the obstacle map 34 means continually replacing the obstacle map 34 with the latest one, thereby keeping the obstacle map 34 always up to date.

Figure 6 shows the state in which the vehicle 1 is parked in the parking space 13. As described above, when automatic parking or parking assistance is initially started, if the recognition area 52 of the recognition unit 23 extends to the near side of the parking space line 12 (or if only the near side can be recognized), the initial parking space 37 (A) is set closer to the actual parking space. Therefore, when the parking space 37 is updated in the moving direction 46, a collision prediction cannot be made for the obstacle 14 behind the final parking space 37 (B) where the vehicle 1 will ultimately be parked due to the insufficient collision prediction range 53.

Similarly, when automatic parking or parking assistance is first started, the periphery detection result acquisition unit 22 may not be able to detect the obstacle 14 behind the final parking space 37 (B) because the detection area of the periphery detection unit 6 does not extend there.

The obstacle map generating unit 25 may be configured to continuously update the obstacle map 34 based on the detection results of the surrounding detection result acquiring unit 22. This allows the collision prediction unit 32 to always use the latest obstacle map 34. The collision prediction unit 32 can then more accurately predict a collision by using the latest obstacle map 34 to perform a collision prediction range 53' corrected in the movement direction 46, thereby achieving a state in which collisions can be appropriately avoided, as shown in FIG. 2.

For example, if the obstacle map 34 remains the same as it was initially, and is not updated, there is a possibility that the obstacle 14 at the back of the parking space 37 may not be detected as described above. As a result, as shown in FIG. 3, it is not possible to predict a collision with the obstacle 14 at the back of the parking space 37(B) in which the vehicle 1 will ultimately be parked, and this may cause the external emergency brake device 9 to function and apply the emergency brakes.

In order to be able to use the latest obstacle map 34 as described above, in the obstacle map generation process 41 of FIG. 7, the obstacle map generation unit 25 generates the obstacle map 34 in step Sa1, and then repeatedly updates the obstacle map 34 at predetermined time intervals in step Sa2.

(5) A parking assistance method using the parking assistance device 3 of this embodiment will now be described.
In the parking assistance method, the recognition unit 23 recognizes the situation around the host vehicle 1 .
In addition, the parking space recognition unit 26 sets a parking space 37 using the recognition result of the recognition unit 23.
In addition, the route generation unit 28 generates a parking route 38 from the position of the vehicle 1 (vehicle position 16) to the parking space 37 (parking position 51) using the situation around the vehicle 1 recognized by the recognition unit 23 and the parking space 37 set by the parking space recognition unit 26.
In addition, the collision prediction unit 32 uses the parking path 38 generated by the path generation unit 28 and the recognition result of the recognition unit 23 to predict whether the vehicle 1 will collide with an obstacle 14 when the vehicle 1 moves along the parking path 38.
Then, when the recognition area 52 of the recognition unit 23 expands in the moving direction 46 of the vehicle 1 as the vehicle 1 moves, the parking space recognition unit 26 updates the position of the parking space 37 in the moving direction 46.
The collision prediction unit 32 performs collision prediction according to the (latest) parking space 37 updated in the moving direction 46 of the host vehicle 1 by the parking space recognition unit 26 .

In the parking assistance process of FIG. 7, the collision prediction process 45 by the collision prediction unit 32 is performed as shown in FIG. 12.

First, in step Se1, information on the initial parking route 38 is obtained from the output of the route generation unit 28.

Next, in step Se2, the start of traveling of the vehicle 1 is confirmed based on information (vehicle control information) from the vehicle control unit 29. Note that steps Se1 and Se2 may be performed in reverse order or simultaneously.

In addition, in step Se3, information on new and updated parking spaces 37 (parking space information) is acquired from the parking space recognition unit 26, and vehicle control correction information 72 is acquired from the vehicle control unit 29. Note that acquisition of the parking space 37 information and acquisition of the vehicle control correction information 72 in step Se3 may be performed separately.

Then, in step Se4, it is determined whether or not correction processing of the collision prediction range 53 is required based on the correction information 72. If the answer is Yes in step Se4, correction processing of the collision prediction range 53 is performed in step Se5. If the answer is No in step Se4, correction processing of the collision prediction range 53 is not performed and the process proceeds directly to step Se6.

The process of correcting the collision prediction range 53 in step Se5 is performed as described in FIG. 11.

In step Se6, a collision determination is made using the obstacle map 34 from the obstacle map generation unit 25, and if the result is No, parking is continued by normal driving. If the result is Yes, the collision prediction unit 32 sends a signal to the vehicle control unit 29, which activates the collision avoidance function to slow down or stop the vehicle 1 in front of the obstacle 14. Thereafter, the above process is repeated until the vehicle 1 is parked in the final parking space 37(B).

<Effects> This embodiment provides the following effects:

(Effect 1) When parking using automatic parking or parking assistance in the parking assistance device 3, the recognition area 52 of the recognition unit 23 gradually expands in the movement direction 46 of the vehicle 1 (not recognized at first) as the vehicle 1 moves. Then, if the position of the initial parking space 37 (A) obtained by the parking space recognition unit 26 based on the initial recognition area 52 of the recognition unit 23 is closer to the vehicle than the final parking space 37 (B), the parking space recognition unit 26 will sequentially update the position of the parking space 37 in the movement direction 46 according to the expansion of the recognition area 52 until the parking space 37 becomes the correct position (the parking space 37 becomes the final parking space 37 (B)). Accordingly, the collision prediction unit 32 of this embodiment is capable of predicting a collision according to the parking space 37 updated by the parking space recognition unit 26.

As a result, in the conventional case where the collision prediction unit 32 limited the collision prediction range 53 to the initial parking path 38 set for the initial parking space 37 (A) to perform collision prediction, it is now possible to continue performing collision prediction for areas further ahead in the moving direction 46 where collision prediction was not possible. Therefore, it is possible to eliminate areas where collision prediction is not performed until the vehicle 1 is parked (entered) in the final parking space 37 (B).

As a result, the collision prediction unit 32 can continue to function properly until parking into the final parking space 37(B) is actually completed. As shown in FIG. 2, a collision can be predicted in advance even for an obstacle 14 located at the back (inside or nearby) of the final parking space 37(B), and gentle deceleration or stopping can be performed at the appropriate timing with minimal load on the occupants to avoid a collision.

In contrast, for example, if the collision prediction unit 32 performs collision prediction by limiting the collision prediction range 53 within the range of the initial parking path 38, even if the parking space 37 is updated in the moving direction 46, the collision prediction cannot be performed once the range of the initial parking path 38 is exceeded, and the collision prediction unit 32 cannot avoid the collision as described above from the middle of the process. As a result, as shown in FIG. 3, the emergency brake device 9 provided on the vehicle 1 separately from the parking assistance device 3 is activated instead for collision avoidance against the obstacle 14 at the back of the final parking space 37 (B). However, with the emergency brake device 9, the emergency brake is applied at the very last moment of the collision, so that the vehicle is suddenly decelerated or stopped, which places a heavy burden on the occupants, and the ride comfort during parking with parking assistance is deteriorated, causing great anxiety to the occupants.

Therefore, in this embodiment, the collision prediction unit 32 continues to function until parking of the vehicle 1 into the final parking space 37(B) is completed, and the vehicle is gradually decelerated or stopped at the appropriate timing with respect to the obstacle 14 at the back of the final parking space 37(B), thereby avoiding a collision before the emergency brake is applied by the emergency brake device 9. This improves the ride comfort and sense of security when parking with parking assistance, and has a significant physical and psychological effect on the occupants.

(Effect 2) When the end position 55a of the parking path 38 is updated in the moving direction 46 of the vehicle 1, the collision prediction unit 32 may perform a collision prediction for the parking space 37 corresponding to the end position 55a of the updated parking path 38. This makes it possible to perform a collision prediction for the latest parking space 37 with a simple process.

In this case, when the parking space 37 is updated to the moving direction 46 of the vehicle 1, the collision prediction unit 32 may update the end position 55a to the moving direction 46 by moving the end position 55a of the original parking path 38 to the parking position 51 of the parking space 37 updated to the moving direction 46. In this way, the same effect as when the parking path 38 is extended in the moving direction 46 can be obtained simply by adding a process to replace the end position 55a of the original parking path 38 with the parking position 51 (or its vicinity) of the latest parking space 37, so that the processing is simple. In addition, when the parking space 37 is updated to the moving direction 46 of the parking space 37, the collision prediction range 53 can be corrected by simply changing a part of the original parking path 38 (the coordinates of the end position 55a) while keeping it almost the same, without performing a large-scale process such as recalculating the entire parking path.

In this case, the parking position 51 of the latest parking space 37 can be easily obtained as coordinates on image data (peripheral image 18) such as an overhead image, so the process of moving the end position 55a of the original parking path 38 to the parking position 51 of the latest parking space 37 (or its vicinity) requires only simple coordinate calculation, which reduces the processing load on the collision prediction unit 32.

(Effect 3) The parking position correction unit 31 may determine a correction amount 48 of the parking position 51 relative to the parking space 37 based on the parking space 37 updated to the moving direction 46 of the vehicle 1 by the parking space recognition unit 26. The vehicle control unit 29 may then control the movement of the vehicle 1 using the original parking path 38 and the correction amount 48 of the parking position 51 determined by the parking position correction unit 31.

As a result, even if the parking space 37 is updated to a moving direction 46 from the original parking space 37 (A), the vehicle 1 can be moved in the moving direction 46 from the original parking path 38 toward the latest parking space 37 by using the control signal 15 for the original parking path 38 and the correction amount 48 for the vehicle position 51, and can be correctly parked in the final parking space 37 (B). At this time, by using the original parking path 38 and the correction amount 48 for the parking position 51 for the original parking path 38, the movement of the vehicle 1 in the moving direction 46 can be controlled and corrected for the updated parking space 37 while basically retaining the control signal 15 for the original parking path 38, without performing large-scale processing such as recalculating to obtain a new control signal for the new parking path.

The collision prediction unit 32 may then perform a collision prediction based on the updated parking space 37 when the vehicle control unit 29 (actually) corrects the control of the vehicle 1 using the correction amount 48 of the parking position 51.

As a result, the collision prediction unit 32 corrects the collision prediction range 53 only when the vehicle control unit 29 receives correction information 72 indicating that the correction amount 48 of the parking position 51 has actually been used to correct the control of the movement of the vehicle 1, and when the correction amount 48 of the parking position 51 has not been used (when the correction information 72 has not been received, or when correction information 72 indicating that the correction amount 48 has not been used (information indicating "no correction") has been received), the collision prediction unit 32 can avoid correcting the collision prediction range 53. Therefore, for example, when the amount of update of the parking space 37 in the movement direction 46 is small and does not substantially affect the collision prediction, it is possible to avoid unnecessary correction of the collision prediction range 53 in the movement direction 46.

(Effect 4) The obstacle map generating unit 25 may generate the obstacle map 34 using the recognition result of the recognition unit 23. In addition, the collision prediction unit 32 may perform collision prediction using the parking path 38 and the obstacle map 34 generated by the obstacle map generating unit 25.

At this time, the obstacle map generating unit 25 may update the obstacle map 34. The collision prediction unit 32 may perform collision prediction (for the corrected collision prediction range 53') using the updated obstacle map 34.

If there is no obstacle 14 around the updated parking space 37 in the direction of movement 46, the parking assistance may be used to park the vehicle 1 directly in the final parking space 37 (B), and if there is an obstacle 14 around the updated parking space 37 in the direction of movement 46, the collision prevention function may be used to slow down or stop the vehicle 1 just before the final parking space 37 (B) to avoid a collision (to be able to control the vehicle control unit 29).

This allows the obstacle map generation unit 25 to always obtain the latest obstacle map 34, and the collision prediction unit 32 can perform collision prediction (for the collision prediction range 53' corrected in the movement direction 46) for the latest parking space 37 based on the latest obstacle map 34, which is updated sequentially, thereby improving both the accuracy of the obstacle map 34 and the accuracy of the collision prediction.

(Effect 5) According to the parking assistance method described above, the same effect as that of the parking assistance device 3 can be obtained.

REFERENCE SIGNS LIST 1 Vehicle 3 Parking assistance device 14 Obstacle 16 Vehicle position 18 Surrounding image 23 Recognition unit 25 Obstacle map generation unit 26 Parking space recognition unit 28 Path generation unit 29 Vehicle control unit 31 Parking position correction unit 32 Collision prediction unit 34 Obstacle map 37 Parking space 37(A) Initial parking space 37(B) Final parking space 38 Parking path 46 Movement direction 48 Correction amount 51 Parking position 52 Recognition area 53 Collision prediction range 53' Corrected collision prediction range 55a End position 72 Correction information

Claims (5)

A recognition unit that recognizes a situation around the vehicle;
A parking space recognition unit that sets a parking space using the recognition result of the recognition unit;
A route generating unit that generates a parking route from the position of the vehicle to the parking space using the surrounding situation of the vehicle recognized by the recognition unit and the parking space set by the parking space recognition unit;
a vehicle control unit that performs vehicle control for parking assistance based on the parking path generated by the path generation unit,
a collision prediction unit that predicts whether the host vehicle will collide with an obstacle when the host vehicle moves along the parking path, using the parking path generated by the path generation unit and the recognition result of the recognition unit, during a period from the start to the end of the vehicle control by the vehicle control unit,
When the recognition area of the recognition unit expands in the moving direction of the vehicle as the vehicle moves, the parking space recognition unit updates the parking space in the moving direction.
The parking assistance device according to claim 1, wherein the collision prediction unit performs a collision prediction in accordance with the parking space updated by the parking space recognition unit in the moving direction of the host vehicle. 2. The parking assistance device according to claim 1,
The parking assistance device is characterized in that, when the parking route is updated in the moving direction of the vehicle, the collision prediction unit moves an end position of the parking route before the update to an end position of the updated parking route, thereby correcting a collision prediction range for the parking space corresponding to the updated parking route by a distance by which the end position has moved . 3. The parking assistance device according to claim 1,
A parking position correction unit that calculates a correction amount of a parking position relative to the parking space based on the parking space updated in the moving direction of the vehicle by the parking space recognition unit;
a vehicle control unit that controls a movement of the vehicle using the parking path and the correction amount of the parking position obtained by the parking position correction unit,
the vehicle control unit corrects control of the host vehicle only when the correction amount of the parking position becomes greater than a predetermined threshold value;
The parking assistance device is characterized in that the collision prediction unit performs a collision prediction according to the updated parking space when the vehicle control unit corrects control of the host vehicle using the correction amount of the parking position. A parking assistance device according to any one of claims 1 to 3,
an obstacle map generation unit that generates an obstacle map using the recognition result of the recognition unit;
the collision prediction unit performs collision prediction using the parking path and the obstacle map generated by the obstacle map generation unit,
the obstacle map generation unit updates the obstacle map, and the collision prediction unit performs collision prediction using the updated obstacle map;
A parking assistance device , wherein the obstacle map is a superposition map in which a surrounding image is superimposed on point cloud data of reflection points of a search wave . The recognition unit recognizes the situation around the vehicle,
A parking space recognition unit sets a parking frame using the recognition result of the recognition unit,
A route generating unit generates a parking route from the position of the vehicle to the parking space by using the surrounding situation of the vehicle recognized by the recognition unit and the parking space set by the parking space recognition unit,
A vehicle control unit performs vehicle control for parking assistance based on the parking path generated by the path generation unit, and
In addition to the vehicle control by the vehicle control unit, a collision prediction unit predicts whether the host vehicle will collide with an obstacle when the host vehicle moves along the parking path by using the parking path generated by the path generation unit and the recognition result of the recognition unit during a period from the start to the end of the parking assistance,
When the recognition area of the recognition unit expands in the moving direction of the vehicle as the vehicle moves, the parking space recognition unit updates the parking space in the moving direction.
The parking assistance method according to the present invention, wherein the collision prediction unit performs a collision prediction according to the parking space updated by the parking space recognition unit in the moving direction of the host vehicle.