KR20230009818A - Method for autonomous driving of stopping vehicle avoidance and apparatus using the same - Google Patents

Abstract

Disclosed are an autonomous driving method for stopping vehicle avoidance and an apparatus for the same, which improve the accuracy of avoidance determination. According to one embodiment of the present invention, the autonomous driving method for stopping vehicle avoidance comprises: a step in which an autonomous driving control apparatus provided in an autonomous vehicle acquires taillight recognition information of a stopping vehicle identified in front of the autonomous vehicle; a step in which the autonomous driving control apparatus considers the taillight recognition information to determines whether to avoid the stopping vehicle; a step in which the autonomous driving control apparatus considers return status determined based on an autonomous task to determine an avoidance method if the stopping vehicle is avoided; and a step in which the autonomous driving control apparatus sets avoidance time in accordance with the avoidance method and controls to avoid the stopping vehicle in accordance with an avoidance route generated to correspond to the avoidance time.

Description

Autonomous driving method for avoiding stopped vehicles and device therefor

The present invention relates to a stopped vehicle avoidance autonomous driving technology, and more particularly, to a technology in which an autonomous vehicle recognizes a stopped vehicle existing in a driving lane and avoids it.

For vehicle autonomous driving, most companies or institutions are using lidar sensors, camera sensors, and radar sensors. In general, when autonomous driving is performed in an urban environment, lane keeping, lane changing, traffic light recognition, inter-vehicle distance maintenance, emergency stop, etc. are the most basic necessary functions, and furthermore, avoidance functions are required. For example, autonomous driving to a destination is possible only when it is determined whether an autonomous vehicle should avoid a vehicle located in front of a driving lane, and an avoidance route is created and provided.

If the avoidance function is not provided, the self-driving vehicle continues to stop behind the stopped vehicle due to the inter-vehicle distance maintenance function with the preceding vehicle, or the driver is converted from the autonomous driving mode to the manual driving mode to perform avoidance. There may be a cumbersome situation in which the autonomous driving function must be operated again to drive.

Korean Patent Publication No. 10-2021-0011258, published on February 1, 2021 (Name: Autonomous vehicle driving priority determination method and device)

An object of the present invention is to provide a method for autonomous driving without switching from an autonomous driving mode to a manual driving mode to a destination by generating and providing an avoidance path for a stopped vehicle existing in a driving lane of an autonomous vehicle.

In addition, an object of the present invention is to provide an autonomous driving method that improves the accuracy of avoidance determination by recognizing the taillight of a stationary vehicle and performs an avoidance task.

In order to achieve the above object, an autonomous driving method for avoiding a stopped vehicle according to the present invention includes the steps of acquiring, by an autonomous driving control device provided in an autonomous vehicle, rear light recognition information of a stopped vehicle identified in front of the autonomous vehicle. ; determining whether to avoid the stopped vehicle in consideration of the rear light recognition information; determining an avoidance method in consideration of whether or not to return determined based on an autonomous driving task (TASK) when avoiding the stopped vehicle; and setting an avoidance time point according to the avoidance method, and controlling to avoid the stopped vehicle according to an avoidance path generated corresponding to the avoidance time point.

In this case, the avoidance method may correspond to one of a first avoidance method of returning to the original lane after avoiding the stopped vehicle and a second avoidance method of performing a lane change while avoiding the stopped vehicle.

At this time, the first avoidance method generates an avoidance area path at the avoiding time point determined by considering the collision time with the stopped vehicle to perform avoidance of the stopped vehicle, and the second avoidance method changes the lane at the lane change time point. It is possible to perform avoidance of the stationary vehicle by creating a local route.

In this case, the avoidance area path and the lane change area path may be generated in consideration of the avoidance free space and the avoidance direction measured using a high-precision map of a road level or a DRIVABLE SPACE.

At this time, if the avoidance method is determined to be the first avoidance method, a virtual self-driving vehicle having the same size as the self-driving vehicle is disposed in the avoidance space, and a virtual lane centerline is determined based on the centerline of the virtual self-driving vehicle. and the avoidance area path for moving from the lane centerline corresponding to the original lane to the virtual lane centerline.

At this time, if the avoidance method is determined to be the second avoidance method, the lane change area path for moving from the lane centerline corresponding to the original lane to the lane centerline corresponding to the target lane to be changed may be generated.

In this case, if the avoidance method is determined to be the first avoidance method, a return area path for moving from the virtual lane centerline to the lane centerline corresponding to the original lane after the avoidance of the stopped vehicle is completed is generated; , it is possible to perform the return according to the return local path.

At this time, the risk of collision with another vehicle driving in the lane located in the avoiding direction may be calculated, and avoidance may be performed when the risk of collision is equal to or less than a predetermined reference risk.

In this case, the taillight recognition information may include emergency light blinking information, taillight on/off information, direction indicator blinking information, and brake light on/off information.

In this case, the stationary vehicle may be identified based on obstacle information detected based on at least one of a camera sensor, lidar sensor, and radar sensor.

At this time, extracting a partial image corresponding to a rear light portion of the stationary vehicle by projecting obstacle information onto an image; and generating the rear light recognition information based on continuous data of the video scene (SCENE) corresponding to the partial video.

In addition, the autonomous driving control apparatus according to an embodiment of the present invention acquires taillight recognition information of a stopped vehicle identified in front of the autonomous vehicle, and determines whether to avoid the stopped vehicle in consideration of the taillight recognition information. In case of avoiding the stopped vehicle, an avoidance method is determined in consideration of whether or not to return determined based on the autonomous driving task (TASK), an avoidance time point is set according to the avoidance method, and an avoidance time point is generated corresponding to the avoidance time point. a processor controlling the self-driving vehicle to avoid the stopped vehicle according to the specified avoidance path; and a memory storing the taillight recognition information, the avoidance time point, and the avoidance path.

In this case, the avoidance method may correspond to one of a first avoidance method of returning to the original lane after avoiding the stopped vehicle and a second avoidance method of performing a lane change while avoiding the stopped vehicle.

At this time, the first avoidance method generates an avoidance area path at the avoiding time point determined by considering the collision time with the stopped vehicle to perform avoidance of the stopped vehicle, and the second avoidance method changes the lane at the lane change time point. It is possible to perform avoidance of the stationary vehicle by creating a local route.

In this case, the avoidance area path and the lane change area path may be generated in consideration of the avoidance free space and the avoidance direction measured using a high-precision map of a road level or a DRIVABLE SPACE.

At this time, if the avoidance method is determined to be the first avoidance method, the processor arranges a virtual autonomous vehicle having the same size as the self-driving vehicle in the avoidance space, and based on the center line of the virtual autonomous vehicle. A lane centerline may be created, and the avoidance area path for moving from the lane centerline corresponding to the original lane to the virtual lane centerline may be generated.

In this case, if the avoidance method is determined to be the second avoidance method, the processor may generate the lane change area path for moving from the lane centerline corresponding to the original lane to the lane centerline corresponding to the target lane to be changed. .

In this case, if the avoidance method is determined to be the first avoidance method, the processor sets a return area path for moving from the virtual lane centerline to the lane centerline corresponding to the original lane after the avoidance of the stopped vehicle is completed. and return can be performed according to the return local path.

In this case, the processor may calculate a risk of collision with another vehicle driving in a lane located in the avoiding direction, and perform avoidance when the risk of collision is equal to or less than a predetermined reference risk.

In this case, the taillight recognition information may include emergency light blinking information, taillight on/off information, direction indicator blinking information, and brake light on/off information.

In this case, the stationary vehicle may be identified based on obstacle information detected based on at least one of a camera sensor, lidar sensor, and radar sensor.

At this time, the processor projects the obstacle information onto an image to extract a partial image corresponding to the taillight portion of the stationary vehicle, and obtains the rearlight recognition information based on continuous data of a video scene (SCENE) corresponding to the partial image. can create

According to the present invention, it is possible to provide a method for autonomously driving to a destination without switching from an autonomous driving mode to a manual driving mode by generating and providing an avoidance path for a stopped vehicle existing in a driving lane of an autonomous vehicle.

In addition, the present invention can improve the accuracy of avoidance judgment by recognizing the taillight of a stationary vehicle and provide an autonomous driving method that performs an avoidance task.

1 is an operational flowchart illustrating an autonomous driving method for avoiding a stopped vehicle according to an embodiment of the present invention.
2 is a diagram illustrating an example of an avoidance task based on an avoidance area path and a lane change task based on a lane change area path according to the present invention.
3 to 6 are diagrams illustrating an example of determining an avoidance free space and an avoidance direction according to the present invention.
7 to 8 are diagrams illustrating examples of a first avoidance method and a second avoidance method according to the present invention.
9 is a diagram showing an example of determining an avoidance time point in consideration of a collision time with a stationary vehicle according to the present invention.
10 is a diagram showing an example of generating an avoidance area path according to the present invention.
11 and 12 are diagrams illustrating an example of performing avoidance by analyzing the risk of collision according to the present invention.
13 is an operation flowchart showing in detail an autonomous driving method for avoiding a stopped vehicle according to an embodiment of the present invention.
14 is a block diagram illustrating an autonomous driving control device for autonomous driving avoiding a stopped vehicle according to an embodiment of the present invention.

The present invention will be described in detail with reference to the accompanying drawings. Here, repeated descriptions, well-known functions that may unnecessarily obscure the subject matter of the present invention, and detailed descriptions of configurations are omitted. Embodiments of the present invention are provided to more completely explain the present invention to those skilled in the art. Accordingly, the shapes and sizes of elements in the drawings may be exaggerated for clarity.

Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the accompanying drawings.

1 is an operational flowchart illustrating an autonomous driving method for avoiding a stopped vehicle according to an embodiment of the present invention.

Referring to FIG. 1 , in an autonomous driving method for avoiding a stopped vehicle according to an embodiment of the present invention, an autonomous driving control device provided in an autonomous vehicle obtains tail light recognition information of a stationary vehicle identified in front of the autonomous vehicle. Do (S110).

In this case, the stationary vehicle may be identified based on obstacle information detected based on at least one of a camera sensor, lidar sensor, and radar sensor.

At this time, obstacle information may be projected onto an image to extract a partial image corresponding to the taillight of the stationary vehicle, and taillight recognition information may be generated based on continuous data of the video scene SCENE corresponding to the partial image.

For example, obstacles detected using a camera sensor, lidar sensor, or radar sensor provided in an autonomous vehicle are projected onto an image, and the image is cut only for the tail light of a stationary vehicle detected in the image to obtain a partial image. can be extracted. Thereafter, by inputting continuous data of the video scene (SCENE) corresponding to the partial video to the machine learning-based detection module, taillight recognition information indicating the taillight state of the stopped vehicle may be acquired through machine learning inference.

In this case, the taillight recognition information may include emergency light blinking information, taillight on/off information, direction indicator blinking information, and brake light on/off information.

For example, the taillight state according to the taillight recognition information may be obtained corresponding to {taillight ON/OFF, emergency light blinking, left/right direction indicator lights blinking, brake light ON/OFF}.

In addition, in the autonomous driving method for avoiding a stopped vehicle according to an embodiment of the present invention, the autonomous driving control device provided in the autonomous vehicle determines whether to avoid the stopped vehicle in consideration of the rear light recognition information (S120).

For example, as in CASE #1 shown in FIG. 3, a stopped vehicle 300 is located at the end of two one-way lanes or more, and the tail light recognition information of the stopped vehicle 300 is {tail light Off} or {hazard light flashing} or It can be assumed that it corresponds to {right turn signal flashing}. In this case, it is determined that the stopped vehicle 300 is in a parking situation or a temporary stop situation, and control is performed to perform avoidance in the left direction of the stopped vehicle 300 so that the autonomous vehicle does not continue to wait behind the stopped vehicle 300. can

As another example, as in CASE #2 shown in FIG. 4, a stopped vehicle 400 is located on a one-lane or more than two one-way lanes, and the tail light recognition information of the stopped vehicle 400 is {tail light OFF} or {hazard light flashing} can be assumed to correspond to Even in this case, it is determined that the stopped vehicle 400 is temporarily stopped, and the autonomous vehicle can be controlled to avoid the stopped vehicle 400 in the right direction.

As another example, as in CASE #3 shown in FIG. 5, a stopped vehicle 500 is located in a lane other than the first lane or the end lane of three one-way or more lanes, and the tail light recognition information of the stopped vehicle 500 {tail light OFF} or {flash of emergency lights}. In this case, since avoidance can be performed in both the left and right directions of the stationary vehicle 500, the avoidance direction can be determined in consideration of the TASK of the autonomous vehicle. For example, if the self-driving task after avoiding is 'turn right at the intersection' or 'arrive at the destination', it may be controlled to perform avoidance in the right direction of the stopped vehicle 500, and the autonomous driving task after avoiding may be 'turn left'. ' or 'U-turn', control to perform avoidance in the left direction of the stationary vehicle 500

As another example, as in CASE #4 shown in FIG. 6, a stopped vehicle 600 is located in one one-way lane, and the rear light recognition information of the stopped vehicle 600 is {tail light OFF}, {hazard light flashing}, or { It can be assumed that it corresponds to right turn signal flashing}. In this case, it is determined that the stopped vehicle 500 is parked or temporarily stopped, and control is performed to evade the stopped vehicle 600 in the left direction so that the autonomous vehicle does not continue to wait behind the stopped vehicle 600. can

In addition, in the autonomous driving method for avoiding a stopped vehicle according to an embodiment of the present invention, after the autonomous driving control device provided in the autonomous vehicle determines whether to avoid the stopped vehicle (S125), if it is determined that avoidance is necessary, the autonomous driving control device is autonomously driven. An avoidance method is determined in consideration of whether or not to return determined based on the driving task (TASK) (S130).

In this case, the avoidance method may correspond to one of a first avoidance method of returning to the original lane after avoiding a stopped vehicle and a second avoidance method of performing a lane change while avoiding a stopped vehicle.

That is, whether to perform the first avoidance method corresponding to the avoidance task TASK shown in FIG. 2 or the second avoidance method corresponding to the lane change task TASK shown in FIG. there is.

At this time, it can be considered that the autonomous driving task (NEXT TASK) is performed within a preset threshold distance (eg, 1Km) after avoiding a stopped vehicle, and the NEXT TASK corresponds to an intersection right turn, left turn, U-turn, destination arrival, etc. there is.

For example, in the situation of FIG. 3 , when the NEXT TASK is performed within a preset threshold distance and the NEXT TASK is a left turn or a U-turn at an intersection, the autonomous vehicle avoids the stopped vehicle 300 in the left direction and then proceeds to the original lane. A lane change task corresponding to the second avoidance scheme may be performed so as not to return to

For another example, in the situation of FIG. 4, when the NEXT TASK is performed within a preset threshold distance and the NEXT TASK is a right turn at an intersection or arrival at a destination, after the autonomous vehicle avoids the stopped vehicle 400 in the right direction A lane change task corresponding to the second avoidance scheme may be performed so as not to return to the original lane.

For another example, in the situation of FIG. 5, when the NEXT TASK is performed within a preset threshold distance and the NEXT TASK is a left turn or a U-turn at an intersection, after the autonomous vehicle avoids the stopped vehicle 500 in the left direction A lane change task corresponding to the second avoidance scheme may be performed so as not to return to the original lane. In addition, when the NEXT TASK is a right turn at an intersection or arrival at a destination, a lane change task corresponding to the second avoidance method may be performed so that the autonomous vehicle does not return to the original lane after avoiding to the right of the stopped vehicle 500. there is.

As another example, in the situation of FIG. 6, since the situation corresponds to a one-way one-lane, the first avoidance method is used to unconditionally return to the original lane after the self-driving vehicle avoids the stopped vehicle 600 regardless of NEXT TASK. A corresponding avoidance task can be performed.

Hereinafter, with reference to FIGS. 7 and 8 , a situation requiring return after avoiding a stopped vehicle and a situation requiring return will be described in more detail.

First, referring to FIG. 7 , it can be confirmed that the autonomous driving task 700 after the autonomous vehicle 710 avoids a stopped vehicle is 'go straight through an intersection'. In this case, since the lane moved to avoid the stopped vehicle is not a straight lane, it is possible to return to the original lane after avoiding the stopped vehicle according to the first avoidance scheme.

Referring to FIG. 8 , it can be confirmed that the autonomous driving task 800 after the autonomous vehicle 810 avoids a stopped vehicle is 'turn left at an intersection'. In this case, since the lane moved to avoid the stopped vehicle is the left turn lane, it may not return to the original lane after avoiding the stopped vehicle according to the second avoidance scheme.

At this time, the intersection traversal mission as shown in FIGS. 7 and 8 may be performed for each intersection that exists from the global path search result of the autonomous vehicle to the destination.

In addition, in the autonomous driving method for avoiding a stopped vehicle according to an embodiment of the present invention, an autonomous driving control device provided in an autonomous vehicle sets an avoidance time point according to an avoidance method, and sets an avoidance route corresponding to the avoidance time point. Avoid the stopped vehicle according to (S140).

At this time, the first avoidance method may perform avoidance of the stopped vehicle by generating an avoidance area path at an avoiding time point determined in consideration of a collision time with the stopped vehicle.

For example, the avoidance time point corresponding to the time point of generating the avoidance area path is before a preset threshold in consideration of the collision time (Time-to-break, TTB) between the self-driving vehicle and the stationary vehicle as shown in FIG. 9 . can be determined by If it is assumed that the preset threshold is 5 seconds, 5 seconds before the collision between the self-driving vehicle and the stationary vehicle may be determined as an avoidance time point and an avoidance area path may be created. At this time, the time-to-break (TTB) can be calculated using the speed v of the autonomous vehicle and the distance d from the stopped vehicle.

In this case, the second avoidance method may perform avoidance of the stopped vehicle by generating a lane change area path at the time of lane change.

That is, the lane change time point may correspond to a time point at which a lane change is required without considering a collision time with a stationary vehicle, etc., unlike the avoidance time point.

In this case, the avoidance area path and the lane change area path may be generated in consideration of the avoidance free space and the avoidance direction measured using a high-precision map of a road level or a driveable space (DRIVABLE SPACE).

For example, in the case of using a road level high-precision map, based on road-level high-precision map data for autonomous driving, as shown in FIGS. An avoidance direction may be determined by grasping a space, and an avoidance path (avoidance area path or a lane change area path) according to the determined avoidance direction may be created.

As another example, the DRIVABLE SPACE may be extracted through real-time inference by artificial intelligence learning using semantic segmentation. Based on this, as shown in FIGS. 3 to 6, by recognizing the drivable area from the center line to the curb or from the curb to the curb, and mapping the stopped vehicles 300, 400, 500, and 600 to the drivable area space, the corresponding A space in which the stopped vehicles 300, 400, 500, and 600 can be avoided can be grasped in the area. In this way, an avoidance direction can be determined so as to avoid in a wider direction by grasping the avoidance space. That is, in the case of FIG. 5, avoidance is performed according to the avoidance direction determined in consideration of the autonomous driving task (NEXT TASK) after avoidance, but an avoidance route can be generated by utilizing the drivable area recognition result at the time when avoidance is necessary. .

At this time, if the avoidance method is determined to be the first avoidance method, a virtual self-driving vehicle having the same size as the self-driving vehicle is placed in the avoidance free space, a virtual lane centerline is created based on the centerline of the virtual self-driving vehicle, and the original An avoidance area path for moving from a lane centerline corresponding to a lane to a virtual lane centerline may be created.

Hereinafter, with reference to FIG. 10, a process of generating an avoidance area path will be described in detail by dividing it into four steps.

First, for avoidance stability, a margin distance (MARGIN) may be set between an avoidance target vehicle (= stopped vehicle) and a virtual self-driving vehicle to be disposed in the avoidance space (S1010).

In this case, the margin distance may be set to such an extent that the autonomous vehicle does not collide with the stationary vehicle even when the autonomous vehicle is positioned side by side with the size of the stationary vehicle and the autonomous vehicle in consideration.

Thereafter, virtual self-driving vehicles may be arranged at set margin distance intervals (S1020), and a virtual lane centerline may be created based on the centerline of the virtual self-driving vehicle (S1030).

Thereafter, an avoidance area path for moving from the lane centerline corresponding to the original lane to the virtual lane centerline may be generated (S1040).

At this time, the avoidance area path may be generated as a followable path using a kinematics model of the autonomous vehicle.

At this time, if the avoidance method is determined to be the first avoidance method, a return local route for moving from the virtual lane centerline to the lane centerline corresponding to the original lane after the avoidance of the stopped vehicle is completed is generated, and the return local route Reversion can be performed according to

In this case, the process of generating the return local path may be the same as the process of generating the avoidance local path.

In this case, the avoidance task according to the first avoidance scheme may end when the self-driving vehicle is located on the lane centerline corresponding to the original lane.

At this time, if the avoidance method is determined to be the second avoidance method, a lane change area path for moving from the lane centerline corresponding to the original lane to the lane centerline corresponding to the target lane to be changed may be generated.

At this time, the risk of collision with another vehicle driving in the lane located in the avoidance direction may be calculated, and avoidance may be performed when the risk of collision is equal to or less than a predetermined reference risk.

For example, referring to FIG. 11 , when the autonomous vehicle avoids the stationary vehicle 1110 in the left direction and there is a left lane, the risk of collision with a rear-side vehicle driving in the left lane may be calculated. At this time, a virtual self-driving vehicle is placed on the left lane at the time the self-driving vehicle starts avoiding, calculates Time-To-Collision (TTC), and if the TTC is below the threshold, the risk of collision exceeds the preset standard risk. It can be judged that In this case, the driving of the autonomous vehicle is stopped to wait for avoidance, and avoidance can be started again when the TTC with the vehicle at the rear side is greater than or equal to the threshold.

At this time, Time-To-Collision (TTC) may be calculated as 'distance to other vehicle / (autonomous vehicle speed - other vehicle speed)'.

For another example, referring to FIG. 12 , when the autonomous vehicle crosses the center line and evades, a risk of collision with a front-side vehicle driving in a lane beyond the center line may be calculated. At this time, similarly to FIG. 11, a virtual self-driving vehicle is placed and placed in a lane beyond the center line at the time when the self-driving vehicle starts avoiding, calculates TTC (Time-To-Collision), and if TTC is less than the threshold, collision It may be determined that the risk level exceeds a preset standard risk level.

At this time, as a method of analyzing the risk of collision, in addition to the method using the TTC used in the example, a probability calculation method, an artificial intelligence deep learning method, and the like can be variously adopted and used.

Also, as a result of the determination in step S125, if avoidance is not necessary, avoidance ends.

In addition, although not shown in FIG. 1, in the autonomous vehicle avoidance method according to an embodiment of the present invention, when the autonomous vehicle is located side by side with the stationary vehicle, the avoidance task according to the first avoidance method is completed. and when the autonomous vehicle is located on the center line of the lane in the lane moved to avoid the stopped vehicle, it may be determined that the lane change task according to the second avoidance method is finished.

Through such a stopped vehicle avoidance autonomous driving method, autonomous driving can be performed without switching from autonomous driving mode to manual driving mode to the destination by creating and providing an avoidance route for the stopped vehicle existing in the driving lane of the autonomous vehicle. .

In addition, it is possible to improve the accuracy of avoidance determination by recognizing the taillight of a stationary vehicle, and perform autonomous driving to perform an avoidance task.

13 is an operation flowchart showing in detail an autonomous driving method for avoiding a stopped vehicle according to an embodiment of the present invention.

Referring to FIG. 13 , in the autonomous driving method for avoiding a stopped vehicle according to an embodiment of the present invention, first, when a destination of an autonomous vehicle is set, the autonomous driving control device searches for a global path to move to the destination and performs autonomous driving. It can start (S1310).

Thereafter, autonomous driving avoiding a stopped vehicle may be performed while continuously determining whether or not the vehicle has arrived at the destination (S1315).

As a result of the determination in step S1315, when the destination is reached, the autonomous vehicle avoidance driving may be terminated.

As a result of the determination in step S1315, if the destination has not been reached, the self-driving vehicle may be controlled by creating a local route according to the autonomous driving task for moving to the destination (S1320).

Thereafter, a stationary vehicle having a speed of '0' may be detected among detected front obstacles (vehicles or pedestrians) using a camera sensor, a lidar sensor, a radar sensor, etc. provided in the autonomous vehicle (S1330).

Thereafter, the rear light of the stopped vehicle is recognized to generate tail light recognition information (S1340), and it is possible to determine whether to avoid the stopped vehicle based on the tail light recognition information (S1345).

For example, when the tail light of a stopped vehicle is {tail light OFF}, {hazard light blinks}, {right turn indicator light blinks} at the end lane, etc., it can be determined that the stopped vehicle is avoided.

As a result of the determination in step S1345, when the stationary vehicle is not avoided, the autonomous vehicle may be controlled by generating a local route according to the autonomous driving task in accordance with step S1320 (S1320).

In addition, as a result of the determination in step S1345, if the control is performed to avoid the stopped vehicle, it may be determined whether it is necessary to return to the original lane after avoiding the stopped vehicle (S1355).

If it is necessary to return after avoiding as a result of the determination in step S1355, the self-driving vehicle can be controlled to avoid the stopped vehicle by creating an avoiding area path at the time of avoiding determined according to the first avoiding method (S1360).

In this case, the avoidance time point may be determined by considering a time-to-break calculated using the current location, speed, and distance of the self-driving vehicle. In addition, the avoidance area path may be created by setting the size of a stopped vehicle and a margin for avoiding it.

Thereafter, it is determined whether the avoidance has ended (S1365), and if the avoidance has not ended, the step S1360 may be repeatedly performed until the avoidance of the stationary vehicle is ended.

In addition, as a result of the determination in step S1354, if the avoidance has ended, the self-driving vehicle may be controlled to return to the original lane by generating an avoidance return area path (S1370).

In this case, it may be determined that the avoidance is finished when the autonomous vehicle is positioned alongside the stopped vehicle or passes the stopped vehicle.

Thereafter, it is determined whether the avoidance return is terminated (S1375), and if the avoidance return is not terminated, the step S1370 may be repeatedly performed until the avoidance return is terminated.

At this time, it may be determined that the avoidance return is completed when the autonomous vehicle is located on the lane center line of the original lane.

In addition, as a result of the determination in step S1355, if it is not necessary to return after avoiding, the self-driving vehicle can be controlled to avoid a stopped vehicle by creating a lane change area path at the time of lane change determined according to the second avoidance method (S1380). ).

Thereafter, it is determined whether the lane change has ended (S1385), and if the lane change has not ended, step S1380 may be repeatedly performed until the lane change is ended.

In this case, it may be determined that the lane change is complete when the autonomous vehicle is located on the lane center line of the changed lane.

14 is a block diagram illustrating an autonomous driving control device for autonomous driving avoiding a stopped vehicle according to an embodiment of the present invention.

Referring to FIG. 14 , an autonomous driving control device for autonomous driving avoiding a stopped vehicle according to an embodiment of the present invention may be implemented in a computer system such as a computer-readable recording medium. As shown in FIG. 14 , computer system 1400 includes one or more processors 1410, memory 1430, user input devices 1440, user output devices 1450, and storage that communicate with each other via a bus 1420. (1460). In addition, computer system 1400 may further include a network interface 1470 coupled to network 1480 . The processor 1410 may be a central processing unit or a semiconductor device that executes processing instructions stored in the memory 1430 or the storage 1460 . The memory 1430 and the storage 1460 may be various types of volatile or non-volatile storage media. For example, the memory may include ROM 1431 or RAM 1432.

Accordingly, embodiments of the present invention may be implemented as a computer-implemented method or a non-transitory computer-readable medium in which instructions executable by the computer are recorded. When executed by a processor, the computer readable instructions may perform a method according to at least one aspect of the present invention.

The processor 1410 obtains rear light recognition information of a stationary vehicle identified in front of the autonomous vehicle.

In this case, the taillight recognition information may include emergency light blinking information, taillight on/off information, direction indicator blinking information, and brake light on/off information.

In this case, the stationary vehicle may be identified based on obstacle information detected based on at least one of a camera sensor, lidar sensor, and radar sensor.

At this time, obstacle information may be projected onto an image to extract a partial image corresponding to the taillight of the stationary vehicle, and taillight recognition information may be generated based on continuous data of the video scene SCENE corresponding to the partial image.

In addition, the processor 1410 determines whether to avoid the stopped vehicle in consideration of the tail light recognition information.

In addition, when avoiding a stopped vehicle, the processor 1410 determines an avoidance method by considering whether or not to return determined based on the autonomous driving task TASK.

In this case, the avoidance method may correspond to one of a first avoidance method of returning to the original lane after avoiding a stopped vehicle and a second avoidance method of performing a lane change while avoiding a stopped vehicle.

In addition, the processor 1410 sets an avoidance time point according to an avoidance method, and avoids a stopped vehicle according to an avoidance path generated corresponding to the avoidance time point.

In this case, the first avoidance method may perform avoidance of the stopped vehicle by generating an avoidance area path at an avoidance time point determined in consideration of a collision time with the stopped vehicle.

In this case, the second avoidance method may perform avoidance of the stopped vehicle by generating a lane change area path at the time of lane change.

In this case, the avoidance area path and the lane change area path may be generated in consideration of the avoidance free space and the avoidance direction measured using a high-precision map of a road level or a driveable space (DRIVABLE SPACE).

At this time, if the avoidance method is determined to be the first avoidance method, a virtual self-driving vehicle having the same size as the self-driving vehicle is placed in the avoidance free space, a virtual lane centerline is created based on the centerline of the virtual self-driving vehicle, and the original An avoidance area path for moving from a lane centerline corresponding to a lane to a virtual lane centerline may be created.

At this time, if the avoidance method is determined to be the second avoidance method, a lane change area path for moving from the lane centerline corresponding to the original lane to the lane centerline corresponding to the target lane to be changed may be generated.

At this time, if the avoidance method is determined to be the first avoidance method, a return local route for moving from the virtual lane centerline to the lane centerline corresponding to the original lane after the avoidance of the stopped vehicle is completed is generated, and the return local route Reversion can be performed according to

At this time, the risk of collision with another vehicle driving in the lane located in the avoidance direction may be calculated, and avoidance may be performed when the risk of collision is equal to or less than a predetermined reference risk.

At this time, the details of performing autonomous driving avoiding a stopped vehicle by the processor 1410 are the same as those described with reference to FIG. 1 , so they will be omitted.

The memory 1430 stores taillight recognition information, an avoidance time point, and an avoidance route.

In addition, the memory 1430 stores various information generated in the autonomous driving control apparatus according to an embodiment of the present invention as described above.

According to an embodiment, the memory 1430 may be configured independently of the autonomous driving control device to support a function for autonomous driving avoiding a stopped vehicle. At this time, the memory 1430 may operate as a separate mass storage and may include a control function for performing an operation.

Meanwhile, the self-driving control device is equipped with a memory and can store information in the device. In one implementation, the memory is a computer readable medium. In one implementation, the memory may be a volatile memory unit, and in another implementation, the memory may be a non-volatile memory unit. In one implementation, the storage device is a computer readable medium. In various different implementations, the storage device may include, for example, a hard disk device, an optical disk device, or some other mass storage device.

By using such an autonomous driving control device, it is possible to perform autonomous driving to a destination without switching from an autonomous driving mode to a manual driving mode by generating and providing an avoidance route for a stopped vehicle existing in a driving lane of an autonomous vehicle.

In addition, it is possible to improve the accuracy of avoidance determination by recognizing the taillight of a stationary vehicle, and to perform autonomous driving in which an avoidance task is performed.

As described above, the method and apparatus for autonomous vehicle avoidance of a stopped vehicle according to the present invention are not limited to the configuration and method of the embodiments described above, but the embodiments can be modified in various ways. All or part of each embodiment may be configured by selectively combining them.

300, 400, 500, 600, 1110, 1210: stationary vehicles
700, 800: autonomous driving task path 710, 810: autonomous vehicle
1120, 1220: other vehicle 1400: computer system
1410: processor 1420: bus
1430: Memory 1431: ROM
1432: RAM 1440: User input device
1450: user output device 1460: storage
1470: network interface 1480: network

Claims (20)

An autonomous driving control device provided in an autonomous vehicle,
acquiring tail light recognition information of a stationary vehicle identified in front of the autonomous vehicle;
determining whether to avoid the stopped vehicle in consideration of the rear light recognition information;
determining an avoidance method in consideration of whether or not to return determined based on an autonomous driving task (TASK) when avoiding the stopped vehicle; and
Setting an avoidance time point according to the avoidance method, and controlling to avoid the stopped vehicle according to an avoidance path generated corresponding to the avoidance time point.
A method for autonomous driving avoiding a stopped vehicle, comprising: The method of claim 1,
The avoidance method
The autonomous driving method for avoiding a stopped vehicle, characterized in that it corresponds to any one of a first avoidance method of returning to the original lane after avoiding the stopped vehicle and a second avoidance method of performing a lane change while avoiding the stopped vehicle. . The method of claim 2,
The first avoidance method performs avoidance of the stopped vehicle by creating an avoidance area path at an avoiding time point determined in consideration of a collision time with the stopped vehicle,
The autonomous driving method for avoiding a stopped vehicle, characterized in that the second avoidance method performs avoidance of the stopped vehicle by generating a lane change area path at the time of lane change. The method of claim 3,
The avoidance area path and the lane change area path are
A method for autonomous driving avoiding a stopped vehicle, characterized in that it is generated in consideration of an avoidance free space and an avoidance direction measured using a high-precision map of a road level or a DRIVABLE SPACE. The method of claim 4,
When the avoidance method is determined as the first avoidance method,
A virtual self-driving vehicle having the same size as the self-driving vehicle is placed in the avoidance space, a virtual lane centerline is created based on the centerline of the virtual self-driving vehicle, and the virtual lane centerline corresponding to the original lane is defined as the virtual autonomous vehicle. A method for autonomous driving avoiding a stopped vehicle, characterized in that for generating the avoidance area path for moving to a lane centerline. The method of claim 4,
When the avoidance method is determined as the second avoidance method,
and generating the lane change area path for moving from the lane center line corresponding to the original lane to the lane center line corresponding to the target lane to be changed. The method of claim 5,
When the avoidance method is determined as the first avoidance method,
After the avoidance of the stopped vehicle is completed, a return area path for moving from the virtual lane center line to the lane center line corresponding to the original lane is generated, and the return is performed according to the return area path. Autonomous driving method to avoid stopped vehicles. The method of claim 4,
A method for autonomous driving avoiding a stopped vehicle, characterized in that calculating a risk of collision with another vehicle driving in a lane located in the avoiding direction, and performing avoidance when the risk of collision is less than or equal to a predetermined reference risk. The method of claim 1,
The tail light recognition information is
An autonomous driving method for avoiding a stopped vehicle, characterized in that it includes emergency light blinking information, tail light on/off information, direction indicator blinking information, and brake light on/off information. The method of claim 1,
The stopped vehicle
A method for autonomous driving avoiding a stopped vehicle, characterized in that identification is made based on obstacle information detected based on at least one of a camera sensor, lidar sensor, and radar sensor. The method of claim 2,
extracting a partial image corresponding to a rear light portion of the stationary vehicle by projecting the obstacle information onto an image; and
The autonomous driving method for avoiding a stopped vehicle further comprising generating the rear light recognition information based on continuous data of a video scene (SCENE) corresponding to the partial video. Acquiring taillight recognition information of a stopped vehicle identified in front of the autonomous vehicle, determining whether to avoid the stopped vehicle in consideration of the taillight recognition information, and in case of avoiding the stopped vehicle, an autonomous driving task (TASK) An avoidance method is determined in consideration of whether to return determined based on, an avoidance time point according to the avoidance method is set, and the autonomous vehicle is operated to avoid the stopped vehicle according to an avoidance path generated corresponding to the avoidance time point. a processor that controls; and
A memory for storing the taillight recognition information, the avoidance time point, and the avoidance path
Autonomous driving control device comprising a. The method of claim 12,
The avoidance method
The autonomous driving device for avoiding a stopped vehicle, characterized in that it corresponds to one of a first avoidance method of returning to the original lane after avoiding the stopped vehicle and a second avoidance method of performing a lane change while avoiding the stopped vehicle. . The method of claim 13,
The first avoidance method performs avoidance of the stopped vehicle by creating an avoidance area path at an avoiding time point determined in consideration of a collision time with the stopped vehicle,
The second avoidance method generates a lane change area path at a lane change time point to perform avoidance of the stopped vehicle. The method of claim 14,
The avoidance area path and the lane change area path are
An autonomous driving device for avoiding a stopped vehicle, characterized in that it is created in consideration of an avoidance free space and an avoidance direction measured using a high-precision map of a road level or a DRIVABLE SPACE. The method of claim 15
The processor
When the avoidance method is determined to be the first avoidance method, a virtual self-driving vehicle having the same size as the self-driving vehicle is disposed in the avoidance space, and a virtual lane centerline is created based on the centerline of the virtual self-driving vehicle. and generating the avoiding area path for moving from a lane centerline corresponding to the original lane to the virtual lane centerline. The method of claim 15
The processor
When the avoiding method is determined to be the second avoiding method, the lane change area path for moving from the lane center line corresponding to the original lane to the lane center line corresponding to the target lane to be changed is generated. autonomous driving device. The method of claim 16
The processor
When the avoidance method is determined to be the first avoidance method, a return area path for moving from the virtual lane centerline to a lane centerline corresponding to the original lane after the avoidance of the stopped vehicle is completed is generated; A stop vehicle avoidance autonomous driving device characterized in that for controlling the autonomous vehicle to perform a return according to a return area route. The method of claim 15
The processor
A vehicle avoidance autonomous driving device characterized in that for calculating a risk of collision with another vehicle driving in the lane located in the avoiding direction, and performing avoidance when the risk of collision is less than or equal to a predetermined reference risk. The method of claim 12,
The tail light recognition information is
A stop vehicle avoidance autonomous driving device comprising emergency light blinking information, taillight on/off information, direction indicator blinking information, and brake light on/off information.

Images