阅读说明：本技术 在自动代客泊车系统中用于估计位置的设备及方法 (Apparatus and method for estimating position in automated valet parking system ) 是由 金桐旭 于 2020-03-16 设计创作，主要内容包括：一种在自动代客泊车系统中用于估计位置的设备及方法,该设备包括前置相机处理器,处理车辆的前方图像；环绕视图监视器(SVM)处理器,通过处理车辆的环绕视图图像来识别短距离车道和停止线；地图数据单元,存储高清晰度地图；以及控制器,当识别出的车辆进入停车场时,从地图数据单元下载包括被设置为停车区的区域的地图,并且当基于所识别的短距离车道和停止线来识别自动代客停车开始位置时,基于前置相机处理器和SVM处理器的识别和处理结果以及地图数据单元的停车场地图,通过执行地图匹配来校正车辆的位置测量值。(An apparatus and method for estimating a position in an automated valet parking system, the apparatus including a front camera processor processing a front image of a vehicle; a Surround View Monitor (SVM) processor recognizing a short-distance lane and a stop line by processing a surround view image of a vehicle; a map data unit storing a high definition map; and a controller that downloads a map including an area set as a parking area from the map data unit when the recognized vehicle enters the parking lot, and corrects a position measurement value of the vehicle by performing map matching based on recognition and processing results of the front camera processor and the SVM processor and a parking lot map of the map data unit when the automatic valet parking start position is recognized based on the recognized short-distance lane and the stop line.)

1. An apparatus for estimating a location in an automated valet parking system, the apparatus comprising:

a front camera processor configured to process a front image of the vehicle;

a look-around monitor processor configured to identify short-range lanes and stop-lines by processing a look-around image of the vehicle;

a map data unit configured to store a high definition map; and

a controller configured to download a map including an area set as a parking area from the map data unit when it is recognized that the vehicle enters a parking lot, and correct a position measurement value of the vehicle by performing map matching based on recognition and processing results of the front camera processor and the look-around monitor processor and a parking lot map of the map data unit when an automatic valet parking start position is recognized based on a short-distance lane and a stop line recognized by the look-around monitor processor.

2. The device of claim 1, wherein the controller is configured to:

predicting the behavior of the vehicle by dead reckoning when the automatic valet parking start position is recognized, and

estimating an automated valet parking initial position of the vehicle by fusing the position measurement value of the vehicle corrected by the map matching with a predicted behavior of the vehicle.

3. The apparatus of claim 1, wherein the controller includes a vehicle behavior prediction unit configured to predict behavior of the vehicle by dead reckoning based on the GPS information received from the GPS receiver and the vehicle steering wheel angle, yaw rate, and wheel speed received from the vehicle sensor unit.

4. The apparatus of claim 1, wherein the controller includes a map matching unit configured to perform the map matching based on at least one of lane fusion data in which a long-distance lane identified by the front camera processor and a short-distance lane and a stop line identified by the look-around monitor processor have been fused, parking lot map data from the map data unit, and vehicle behavior data for each time predicted by dead reckoning.

5. The device of claim 4, wherein the map matching unit is configured to: where the iterative closest point logic is used to minimize the distance error between the sensor data and the map data to calculate the position and rotation corrections.

6. The apparatus according to claim 1, wherein the controller includes a position fusion unit configured to fuse the vehicle attitude output as a result of the map matching with GPS information of the vehicle position predicted by dead reckoning.

7. The apparatus of claim 6, wherein:

the controller includes a fail-safe diagnosis unit configured to receive the vehicle position and the flag output by the position fusion unit and perform fail-safe diagnosis, an

The fail-safe diagnosis unit is configured to perform the fail-safe diagnosis using a distribution map configured with estimated positioning results in which positioning results at past times have been projected to a current time and positioning results are input at the current time.

8. The apparatus of claim 6, wherein the vehicle pose comprises one or more of: longitude, latitude, heading, covariance, warning/trouble/safety, signs, and lane offset.

9. A method of estimating a location in an automated valet parking system, the method comprising the steps of:

downloading, by the controller, a map including an area set as a parking area from a map data unit for storing a high definition map when it is recognized that the vehicle enters the parking lot;

identifying, by the controller, an automated valet parking start location based on the short-range lane and the stop line identified by the look-around monitor processor; and

correcting, by the controller, the position measurement value of the vehicle by performing map matching based on the results of the recognition and processing by the front camera processor and the look-around monitor processor and the parking lot map of the map data unit.

10. The method of claim 9, further comprising:

predicting, by the controller, a behavior of the vehicle by dead reckoning when the automatic valet parking start position is recognized, and

estimating, by the controller, an auto valet parking initial position of the vehicle by fusing the position measurement of the vehicle corrected by the map matching with a predicted behavior of the vehicle.

11. The method of claim 10, wherein in predicting the behavior of the vehicle, the controller predicts the behavior of the vehicle by dead reckoning based on GPS information received from a GPS receiver and a vehicle steering wheel angle, a yaw rate, and a wheel speed received from a vehicle sensor unit.

12. The method of claim 9, wherein in correcting the position measurement, the controller performs the map matching based on at least one of lane fusion data in which a long-distance lane identified by the front camera processor and a short-distance lane and a stop line identified by the look-around monitor processor have been fused, parking lot map data from the map data unit, and vehicle behavior data for each time predicted by dead reckoning.

13. The method of claim 12, wherein in correcting the position measurements, the controller calculates position and rotation corrections therein using iterative closest point logic to minimize distance errors between sensor data and map data.

14. The method according to claim 9, wherein in estimating the automated valet parking initial position, the controller fuses a vehicle attitude output as a result of map matching with GPS information of a vehicle position predicted by dead reckoning.

15. The method of claim 14, further comprising receiving, by the controller, the vehicle position and the flag output as a result of the position fusion, and performing a fail-safe diagnosis,

wherein, in performing the fail-safe diagnosis, the controller performs the fail-safe diagnosis using a distribution map configured with estimated positioning results in which positioning results at past times have been projected to a current time and positioning results are input at the current time.

Technical Field

Embodiments of the present disclosure relate to an apparatus and method for estimating a location in an automated valet parking system, and more particularly, to an apparatus and method for estimating a location in an automated valet parking system, which can estimate an initial location in an Automated Valet Parking (AVP) system using a Surround View Monitor (SVM).

Background

In general, an autonomous vehicle refers to a vehicle that autonomously determines a driving path by recognizing a surrounding environment using a function for detecting and processing external information while driving and independently travels using its own power.

Positioning methods applied to automotive vehicles include Global Navigation Satellite System (GNSS) based satellite positioning methods such as Global Positioning System (GPS), differential GPS (dgps), or carrier-phase differential (RTK); vehicle-based behavior dead reckoning for correcting satellite positioning using vehicle sensors and Inertial Measurement Units (IMU) (e.g., vehicle speed, steering angle, and wheel odometer/yaw rate/acceleration); and a map matching method of relatively estimating a position of the vehicle by comparing with an accurate map for autonomous driving having data from various sensors (e.g., a camera, a stereo camera, an SVM camera, and a radar).

Recently, Automated Valet Parking (AVP) has been developed to facilitate parking. An autonomous vehicle on which the AVP system is mounted can autonomously travel, search for a parking space, and perform parking or exit from a parking lot without a driver. Furthermore, even a function for performing parking by extending a target parking space to a surrounding parking lot in a traffic-crowded area has been developed.

Therefore, a positioning method for estimating a position becomes important. However, the conventional satellite positioning method has a problem in that the method is very expensive because it requires a high-definition GPS, a high-definition radar, and a high-resolution camera; this method has low processing speed and accuracy because it is configured with a complex algorithm; and the method cannot continuously maintain its performance because it is affected by the characteristics of the lane and the characteristics of the surrounding geographical features.

Related art of the present disclosure is disclosed in U.S. patent application publication No.2018-0023961(2018, month 1, day 25) entitled "system and method for aligning crowd-sourced sparse map data".

Disclosure of Invention

Various embodiments are directed to an apparatus and method for estimating a location in an automated valet parking system, which can estimate an initial location of the Automated Valet Parking (AVP) system using a Surround View Monitor (SVM) without expensive equipment.

In one embodiment, an apparatus for estimating a location in an automated valet parking system includes a front camera processor configured to process a front image of a vehicle; a Surround View Monitor (SVM) processor configured to recognize a short-distance lane and a stop-line by processing a surround view image of a vehicle; a map data unit configured to store a high definition map; and a controller configured to download a map including an area set as a parking area from the map data unit when it is recognized that the vehicle enters the parking lot, and correct a position measurement value of the vehicle by performing map matching based on recognition and processing results of the front camera processor and the SVM processor and a parking lot map of the map data unit when an auto-valet parking (AVP) start position is recognized based on the short-distance lane and the stop line recognized by the SVM processor.

In one embodiment, the controller is configured to predict the behavior of the vehicle by law when the AVP start position is identified, and estimate the AVP initial position of the vehicle by fusing the position measurement value of the vehicle corrected by map matching with the predicted behavior of the vehicle.

In one embodiment, the controller includes a vehicle behavior prediction unit configured to predict the behavior of the vehicle by dead reckoning based on the GPS information received from the GPS receiver and the vehicle steering wheel angle, yaw rate, and wheel speed received from the vehicle sensor unit.

In one embodiment, the controller includes a map matching unit configured to perform map matching based on at least one of lane fusion data in which a long-distance lane recognized by the front camera processor and a short-distance lane recognized by the SVM processor and a stop line have been fused, parking lot map data from the map data unit, and vehicle behavior data for each time predicted by dead reckoning.

In one embodiment, the map matching unit is configured to: in which a distance error between sensor data and map data is minimized using Iterative Closest Point (ICP) logic to calculate a position and rotation correction amount.

In an embodiment, the controller includes a position fusion unit configured to fuse GPS information of a vehicle attitude output as a result of map matching and a vehicle position predicted by dead reckoning.

In an embodiment, the controller includes a fail-safe diagnostic unit configured to receive the vehicle position and the flag output by the position fusion unit and perform a fail-safe diagnosis. The fail-safe diagnosis unit is configured to perform the fail-safe diagnosis using a distribution map configured with estimated positioning results in which positioning results at past times have been projected to a current time and positioning results are input at the current time.

In one embodiment, the vehicle attitude comprises one or more of: longitude, latitude, heading, covariance, warning/trouble/safety, signs, and lane offset.

In one embodiment, a method of estimating a location in an automated valet parking system, when it is recognized that a vehicle enters a parking lot, downloads, by a controller, a map including an area set as a parking area from a map data unit for storing a high definition map; identifying, by the controller, an Automated Valet Parking (AVP) start location based on the short-range lane and stop line identified by the Surround View Monitor (SVM) processor; and correcting, by the controller, the position measurement value of the vehicle by performing map matching based on the results of the recognition and processing by the front camera processor and the SVM processor and the parking lot map of the map data unit.

In one embodiment, the method further includes predicting, by the controller, a behavior of the vehicle by dead reckoning when the AVP start position is identified, and estimating, by the controller, an AVP initial position of the vehicle by fusing the position measurement of the vehicle corrected by map matching and the predicted behavior of the vehicle.

In one embodiment, in predicting the behavior of the vehicle, the controller predicts the behavior of the vehicle by dead reckoning based on the GPS information received from the GPS receiver and the vehicle steering wheel angle, yaw rate, and wheel speed received from the vehicle sensor unit.

In one embodiment, in correcting the position measurement value, the controller performs map matching based on at least one of lane fusion data in which the long-distance lane identified by the front camera processor and the short-distance lane and the stop line identified by the SVM processor have been fused, parking lot map data from a map data unit, and vehicle behavior data for each time predicted by dead reckoning.

In one embodiment, in correcting the position measurement, the controller calculates the position and rotation correction amounts therein using Iterative Closest Point (ICP) logic to minimize a distance error between the sensor data and the map data.

In one embodiment, in estimating the AVP initial position, the controller fuses the vehicle attitude output as a result of map matching with GPS information of the vehicle position predicted by dead reckoning.

In one embodiment, the method further includes receiving, by the controller, the vehicle position and the flag output as a result of the position fusion, and performing a fail-safe diagnosis. In performing the fail-safe diagnosis, the controller performs the fail-safe diagnosis using a distribution map configured with estimated positioning results in which positioning results at past times have been projected to the current time and positioning results are input at the current time.

Drawings

Fig. 1 is a block diagram illustrating an apparatus for estimating a location in an automated valet parking system according to an embodiment of the present disclosure.

Fig. 2 is a diagram more specifically describing an apparatus for estimating a location in an automated valet parking system according to an embodiment of the present disclosure.

Fig. 3 is a flowchart for describing a method of estimating a location in an automated valet parking system according to an embodiment of the present disclosure.

Fig. 4 is an exemplary diagram of an apparatus and method for estimating a location in an automated valet parking system according to an embodiment of the present disclosure.

Fig. 5 is a diagram for describing a map matching logic of an apparatus and method for estimating a location in an automated valet parking system according to an embodiment of the present disclosure.

Detailed Description

Hereinafter, an apparatus and a method for estimating a location in an automated valet parking system according to an embodiment of the present disclosure are described with reference to the accompanying drawings. Thicknesses of lines, sizes of constituent elements, and the like are exaggeratedly illustrated in the drawings for clarity and convenience in the description.

Further, terms to be described below are defined by taking functions in the present disclosure into consideration, and the definition may be different according to the intention of a user, an operator, or practice. Therefore, each term should be defined based on the contents of the entire specification.

Furthermore, the implementations described in this specification may be implemented, for example, as a method or process, an apparatus, a software program, a data stream, or a signal. While the disclosure has been discussed in the context of only a single form of implementation (e.g., discussed only as a method), an implementation having the features discussed may also be implemented in another form (e.g., a device or program). The apparatus may be implemented as suitable hardware, software or firmware. The method may be implemented in an apparatus such as a processor, which refers generally to a processing device including, for example, a computer, microprocessor, integrated circuit, or programmable logic device. Processors include communication devices, such as computers, cellular telephones, mobile/palmtop computers ("PDAs"), and other devices that facilitate the communication of information between end-users.

Fig. 1 is a block diagram illustrating an apparatus for estimating a location in an automated valet parking system according to an embodiment of the present disclosure. Fig. 2 is a diagram more specifically describing an apparatus for estimating a location in an automated valet parking system according to an embodiment of the present disclosure. Fig. 5 is a diagram for describing a map matching logic of an apparatus and method for estimating a location in an automated valet parking system according to an embodiment of the present disclosure. An apparatus for estimating a location in an automated valet parking system is described below with reference to fig. 1, 2, and 5.

As shown in fig. 1, an apparatus for estimating a location in an automated valet parking system according to an embodiment of the present disclosure includes a front camera processor 10, a Surround View Monitor (SVM) processor 20, a map data unit 30, a GPS receiver 40, a vehicle sensor unit 50, a controller 60, and an output unit 70.

Some exemplary embodiments may be shown in the figures from the perspective of functional blocks, units, portions and/or modules, as is well known in the art. Those skilled in the art will appreciate that such blocks, units, and/or modules are physically implemented by electronic (or optical) circuits such as logic circuits, discrete components, processors, hard-wired circuits, memory devices, and wired connections. When the blocks, units and/or modules are implemented by a processor or other similar hardware, the blocks, units and modules may be programmed and controlled by software (e.g., code) to perform the various functions discussed in this specification. Further, each of the blocks, units and/or modules may be implemented by dedicated hardware or a combination of dedicated hardware for performing some functions and a processor for performing another function (e.g., one or more programmed processors and associated circuitry). In some example embodiments, each of the blocks, units and/or modules may be physically divided into two or more blocks, units and/or modules that are interactive and discrete without departing from the scope of the present disclosure. Furthermore, the blocks, units and/or modules in some example embodiments may be physically coupled into more complex blocks, units and/or modules without departing from the scope of the present disclosure.

First, the present embodiment is for estimating an initial position in an Automated Valet Parking (AVP) system using an SVM that assists in parking a vehicle by allowing surrounding parking spaces of the vehicle to be visible within the vehicle through cameras attached to the front, rear, and sides of the vehicle. That is, the present embodiment relates to a vehicle positioning apparatus, and does not require expensive equipment, stop lines, or the like, and can measure the position of a vehicle using an image captured by a camera and improve the accuracy of map matching.

The front camera processor 10 may receive a front image of the vehicle from a front camera of the vehicle, and may recognize a long-distance lane and a traffic sign by processing the front image of the vehicle.

Further, the front camera processor 10 may include an inside lane recognition device for recognizing an inside lane in the front image, a lane tracking device for tracking a lane having the same characteristics as the recognized lane, and a reliability calculation device.

The inside lane recognition device may recognize a lane in the form of a solid line or a dotted line having a specific color (e.g., white or yellow) in the front image.

Although the components (e.g., color, thickness, and shape) of the recognized lane do not partially maintain the same characteristics (e.g., the same line color, the same line thickness, and the same line interval), the lane tracking apparatus may track the lane having the same characteristics within a pre-specified range by considering the traffic flow (or direction) of the recognized lane.

Further, the reliability calculation means may calculate a ratio of components (e.g., color, thickness, and shape) of the tracked lane to be the same as a pre-specified reference value of each component (i.e., a lane component matching ratio). When the calculated lane component matching ratio is closer to 100%, this means high reliability. Conversely, when the calculated lane component matching ratio is closer to 0%, this means low reliability. Further, the reliability calculation means may predict the current lane (i.e., the predicted lane) using the recognition result of the previous lane and the motion information of the vehicle, and may calculate the reliability in such a manner that the predicted lane is compared with the current lane recognized in the front image and the reliability count (or reliability score) is increased when the difference between the predicted lane and the current lane is a preset threshold or less. When the reliability count is greater than the preset threshold, the reliability calculation means may determine that the corresponding lane identification is valid (i.e., the corresponding lane is a valid lane).

The SVM processor 20 may identify short-range lanes and stop-lines by processing a surround view image of the vehicle.

In addition, the SVM processor 20 serves to recognize a lane in the surround view image (or the surround view composition image). The surround view image refers to an image obtained by synthesizing surrounding images (e.g., front, side, and rear images) of the vehicle in the form of a top view or a surround view captured by one or more cameras. Accordingly, the SVM processor 20 may identify lanes in the area proximate to the vehicle (i.e., short distance lanes).

In this case, the cameras are arranged at the front, rear, left, and right sides of the vehicle. In order to increase the finish of the top view or surround view image and prevent the occurrence of a photographing blind spot, additional cameras may be further provided at upper sides of the front and rear of the vehicle, i.e., at positions relatively higher than those of the cameras provided at the front, rear, left and right sides.

Further, as with the front camera processor 10, the SVM processor 20 may include an inside lane recognition device, a lane tracking device, and a reliability calculation device.

That is, the inner lane recognition device may recognize the inner lane in the surround view image, and may recognize the lane in the form of a solid line or a dotted line having a specific color (e.g., white or yellow) in the surround view image. In the present embodiment, specifically, the inside lane recognition device may recognize the stop line.

Although the components (e.g., color, thickness, and shape) of the recognized lane do not partially maintain the same characteristics (e.g., the same line color, the same line thickness, and the same line interval), the lane tracking apparatus may track the lane having the same characteristics within a pre-specified range by considering the traffic flow (or direction) of the recognized lane.

Further, the reliability calculation means may calculate a ratio of components (e.g., color, thickness, and shape) of the tracked lane to be the same as a pre-specified reference value of each component (i.e., a lane component matching ratio). When the calculated lane component matching ratio is closer to 100%, this means high reliability. Conversely, when the calculated lane component matching ratio is closer to 0%, this means low reliability. Further, the reliability calculation means may predict the current lane (i.e., the predicted lane) using the recognition result of the previous lane and the motion information of the vehicle, and may calculate the reliability in such a manner that the predicted lane is compared with the current lane recognized in the surround view image and the reliability count (or reliability score) is increased when the difference between the predicted lane and the current lane is a preset threshold or less. When the reliability count is greater than the preset threshold, the reliability calculation means may determine that the corresponding lane identification is valid (i.e., the corresponding lane is a valid lane).

The short-distance lane may mean a lane in an area that can be recognized in the surround view image. The long-distance lane may mean a lane in a long-distance area that can be recognized in the front image.

The map data unit 30 stores a high-definition map in which information on roads and surrounding topography is constructed with high accuracy, and provides the high-definition map in response to a request from the controller 60. In the present embodiment, specifically, the map data unit 30 may store a High Definition (HD) map for a parking lot (i.e., an area set as a parking area).

The GPS receiver 40 receives GPS signals from satellites and provides the GPS signals to the controller 60 so that the position of the vehicle can be set based on the current position.

The vehicle sensor unit 50 refers to various sensors in the vehicle. In the present embodiment, specifically, the vehicle sensor unit 50 may include a vehicle steering wheel angle sensor, a yaw rate sensor, and a wheel speed sensor for vehicle behavior prediction.

The controller 60 recognizes that the vehicle enters an area set as a parking lot or a parking area, and downloads a map of the corresponding area. That is, when it is recognized that the vehicle enters the parking lot, the controller 60 may download a map including an area set as a parking area from the map data unit 30.

Further, the controller 60 may identify the AVP start position based on the short-distance lane and stop line identified by the SVM processor 20.

At this time, the controller 60 may generate a single fused lane (i.e., a single lane that is not divided into a short-distance lane and a long-distance lane) by fusing the lane recognized by the SVM processor 20 with the lane recognized by the front camera processor 10.

That is, the controller 60 may fuse lanes through comparison of lane errors, may determine a valid lane, and may generate a fused lane.

The controller 60 calculates (or determines) a position error (e.g., an interval between the end of the lane and an angle of each lane in the vehicle reference coordinate system) by comparing the lane (i.e., the short-distance lane) recognized by the SVM processor 20 and the lane (i.e., the long-distance lane) recognized by the front camera processor 10. In this case, the vehicle reference coordinate system means a coordinate system indicating a lateral coordinate X, a longitudinal coordinate Y, and a vehicle movement direction θ corresponding to a movement distance and direction of the vehicle with respect to the center of the vehicle.

When the position error as a result of the comparison between the two lanes (i.e., the long-distance lane and the short-distance lane) is within the preset allowable range, the controller 60 generates a single merged lane (i.e., a single lane that is not divided into the short-distance lane and the long-distance lane) by merging the two lanes (i.e., the long-distance lane and the short-distance lane). Further, when the position error as a result of the comparison between the two lanes (i.e., the long-distance lane and the short-distance lane) exceeds the preset allowable range, the controller 60 does not merge the two lanes (i.e., the long-distance lane and the short-distance lane) and determines the lane having relatively high reliability as the valid lane.

Accordingly, the controller 60 may determine that the two lanes (i.e., the long-distance lane and the short-distance lane) are not valid lanes when the reliability of each lane is less than a preset threshold. When the reliability of each of the two lanes is a preset threshold or more and the position error between the two lanes is within a preset allowable range, the controller 60 may generate a single fused lane (i.e., a single lane that is not divided into a short-distance lane and a long-distance lane) by fusing the two lanes. When the position error between the two lanes exceeds the preset allowable range, the controller 60 may determine a lane having relatively high reliability of the two lanes as a valid lane.

Further, the controller 60 includes a vehicle behavior prediction unit 62, a map matching unit 64, a location fusion unit 66, and a fail-safe diagnosis unit 68. The controller 60 may predict the behavior of the vehicle through dead reckoning, may correct the position measurement value of the vehicle through map matching based on the recognition and processing results of the front camera processor 10 and the SVM processor 20 and the parking lot map of the map data unit 30, and may finally estimate the AVP initial position of the vehicle by fusing the predicted behavior of the vehicle and the corrected vehicle position measurement value.

The vehicle behavior prediction unit 62 may predict the behavior of the vehicle through dead reckoning based on the GPS information received from the GPS receiver 40 and the vehicle steering wheel angle, the yaw rate, and the wheel speed received from the vehicle sensor unit 50.

Further, the map matching unit 64 may perform map matching based on at least one of lane fusion data in which the long-distance lane identified by the front camera processor 10 and the short-distance lane and the stop line identified by the SVM processor 20 have been fused, parking lot map data from the map data unit 30, and each time vehicle behavior data predicted by dead reckoning.

At this time, the map matching unit 64 may calculate the position and rotation correction amounts therein using Iterative Closest Point (ICP) logic to minimize the distance error between the sensor data and the map data. ICP logic is a method of registering current data with an existing data set, and is a method of finding an association based on the closest point of data, moving and rotating the current data based on the association, and adding the current data to the existing data set.

For example, the position T and the rotation (R) correction amount may be calculated with reference to the following formula and fig. 5.

Minimum size

Further, the position fusion unit 66 may fuse the vehicle attitude output as a result of the map matching and the GPS information of the vehicle position predicted by dead reckoning.

In this case, the position fusion unit 66 may be implemented as a method of fusing the lanes recognized by the SVM processor 20 and the front camera processor 10, but another method may be used to fuse the position measurement values.

The controller 60 includes a fail-safe diagnostic unit 68 for receiving the vehicle position and the flag output by the position fusion unit 66 and performing fail-safe diagnostics. The fail-safe diagnosis unit 68 may perform fail-safe diagnosis using a distribution map configured with estimated positioning results in which positioning results at past times have been projected to the current time and positioning results are input at the current time.

In this embodiment, the vehicle attitude may include one or more of: longitude, latitude, heading, covariance, warning/trouble/safety, signs, and lane offset.

That is, in the present embodiment, the output unit 70 may output the diagnosis result of the fail-safe diagnosis unit 68. In this case, the output unit 70 may output the fail-safe diagnosis result of the vehicle posture information.

In the present embodiment, the autonomous driving system can perform sensor fusion localization based on map matching and can perform fail-safe diagnosis to improve the reliability of the system and make the localization information calculated (or calculated or estimated) and stable robustly in the process of performing sensor fusion localization. Further, in the present embodiment, the fail-safe diagnosis does not require additional hardware because it is a fail-diagnosis based on redundancy analysis, but the present disclosure is not limited thereto.

Referring to fig. 2, the present embodiment basically may include a performance core for performing fusion processing of a position measurement value corrected by map matching and a prediction result of the behavior of the vehicle, and a safety core for performing fail-safe diagnosis of the vehicle position fused in the performance core.

In the performance core, the front camera processor 10 and the SVM processor 20 may perform sensor value processing, and the map data unit 30 may download and manage a map. Further, in the security kernel, the GPS receiver 40 may perform GPS signal processing.

Further, in the controller 60, the map matching unit 64 and the location fusion unit 66 may be included in a performance core, and the vehicle behavior prediction unit 62 and the fail-safe diagnosis unit 68 may be included in a safety core, but the present disclosure is not limited thereto.

In other words, the performance core receives the lane and stop line from the SVM and the lane and traffic signs from the front camera. Furthermore, the performance core may process recognition data, i.e. sensor values received from the SVM and the front camera. In other words, the controller 60 may fuse the recognition data from the SVM and the front camera, may download the HD map of the parking area, and may perform map matching. In this case, the controller 60 may perform map matching using the GPS signal, the vehicle trajectory information predicted by dead reckoning, and the GPS information. Further, the controller 60 may fuse the vehicle attitude (or position) corrected by map matching with the position value based on the GPS information, and may finally estimate an initial position in the automated valet parking system by performing a fail-safe diagnosis on the result of the fusion.

Fig. 3 is a flowchart for describing a method of estimating a location in an automated valet parking system according to an embodiment of the present disclosure. Fig. 4 is an exemplary diagram of an apparatus and method for estimating a location in an automated valet parking system according to an embodiment of the present disclosure. A method of estimating a location in an automated valet parking system is described below with reference to fig. 3 and 4.

As shown in fig. 3, in the method of estimating a position in an automated valet parking system according to an embodiment of the present disclosure, first, the controller 60 recognizes that a vehicle enters a parking lot (S10).

In this case, the controller 60 may recognize that the vehicle enters the parking lot by receiving the vehicle position from the GPS receiver 40, but the present invention is not limited to such a method.

Further, when it is recognized that the vehicle enters the parking lot, the controller 60 downloads a map including an area set as a parking area from the map data unit 30 in which a High Definition (HD) map is stored (S20).

Further, if it is recognized that the vehicle has been parked at the AVP start position (S30), the controller 60 determines whether an AVP start region has been recognized by the SVM processor 20 (S40).

That is, as shown in (a) of fig. 4, when the vehicle stops at the AVP start position, as shown in (b) of fig. 4, the controller 60 may identify the AVP start position based on the short-distance lane and the stop line identified by the SVM processor 20.

If the AVP start region is not recognized by the SVM processor 20, the controller 60 may return to step S30 and perform AVP start position parking. The AVP start position parking may be performed by a user. The controller 60 may recognize that the vehicle has been parked at the AVP start position by receiving GPS information or a signal indicating that the vehicle has been parked from the AVP infrastructure installed in the parking lot.

Further, the controller 60 sets the position to the initial value of the AVP start position (S50).

That is, as shown in (c) of fig. 4, the controller 60 may set the position to an initial value of the AVP start position based on the AVP start region recognized by the SVM processor 20.

Further, as shown in fig. 4 (d), the controller 60 corrects the AVP start position (S60).

At this time, the controller 60 may predict the behavior of the vehicle through dead reckoning, and may correct the position measurement value of the vehicle through map matching based on the recognition and processing results of the front camera processor 10 and the SVM processor 20 for recognizing the long-distance lane and the traffic sign and the parking lot map of the map data unit 30.

Further, the controller 60 may ultimately estimate the AVP initial position of the vehicle by fusing the corrected position measurement of the vehicle with the predicted behavior of the vehicle.

In this case, the controller 60 may predict the behavior of the vehicle through dead reckoning based on the GPS information received from the GPS receiver 40 and the vehicle steering wheel angle, yaw rate, and wheel speed received from the vehicle sensor unit 50. Further, the controller 60 may perform map matching based on at least one of lane fusion data in which the long-distance lane recognized by the front camera processor 10 and the short-distance lane and the stop line recognized by the SVM processor 20 have been fused, parking lot map data from the map data unit 30, and vehicle behavior data for each time predicted by dead reckoning.

In the present embodiment, the controller 60 may calculate the position and rotation correction amounts therein using Iterative Closest Point (ICP) logic to minimize a distance error between the sensor data and the map data.

Further, the controller 60 may fuse the vehicle attitude output as a result of the map matching with the GPS information of the vehicle position predicted by dead reckoning.

Finally, the controller 60 may receive the vehicle position and the flag output as a result of the position fusion, and may perform a fail-safe diagnosis (S70).

In this case, the controller 60 may perform the safety failure diagnosis using a distribution map configured with an estimated localization result in which the localization result at the past time has been projected to the current time and the localization result is input at the current time. In this case, the vehicle attitude may include one or more of: longitude, latitude, heading, covariance, warning/trouble/safety, signs, and lane offset.

As described above, the apparatus and method for estimating a location in an automated valet parking system according to an embodiment of the present disclosure may perform map matching without expensive equipment, and estimate initial locations regardless of interior and exterior by estimating initial locations in an Automated Valet Parking (AVP) system using a Surround View Monitor (SVM), and may improve map matching accuracy by increasing cognitive distance accuracy (i.e., correction accuracy) by performing measurements in the vicinity of a geographic feature.

Although the preferred embodiments of the present disclosure have been disclosed for illustrative purposes, those skilled in the art will appreciate that various modifications, additions and substitutions are possible, without departing from the scope of the disclosure as defined in the accompanying claims.

Therefore, the true technical scope of the present disclosure should be defined by the appended claims.