KR20200021688A - System and method for researching localization of autonomous vehicle - Google Patents

Abstract

The present invention relates to a position estimating system of an autonomous vehicle, which comprises: a step of constructing a precise map for driving an autonomous vehicle using road information; a navigation position estimating step of estimating a vehicle position using the constructed precise map and vehicle driving information; an error correcting step of correcting a position error in accordance with whether a GPS is short-circuited and a lane exists on the precise map; a step of re-estimating the vehicle position similar to a reference value of the precise map by applying an error-corrected correction value and the navigation position-estimated prediction value to an unsecented Kalman filter; and a step of finally estimating the vehicle position by applying a real-time GPS value and a re-estimated value in accordance with vehicle driving to a Kalman filter.

Description

Positioning system of autonomous vehicle and its location estimation method {SYSTEM AND METHOD FOR RESEARCHING LOCALIZATION OF AUTONOMOUS VEHICLE}

The present invention relates to a method for estimating the position of an autonomous vehicle.

More specifically, the step of constructing a precise map for driving the autonomous vehicle using the road information, the navigation position estimation step of estimating the vehicle position using the built precision map and the driving information of the vehicle, whether or not GPS short circuit and precision An error correction step of performing a position error correction according to whether a lane exists on a map, and applying the error corrected correction value and the navigation position estimation predicted value to an unsecented Kalman filter, a vehicle similar to the reference value of the precision map. And re-estimating the location of the vehicle, and finally estimating the location of the vehicle by applying the real-time GPS value and the re-estimation value to the Kalman Filter. It is about.

Self-driving cars are a collection of intelligent car technologies that allow drivers to get in the car and designate their desired destinations so that they can drive by creating an optimal route to their current location or destination without any special manipulation.

In addition, autonomous vehicles can proactively cope with accidents by recognizing road traffic signs and signs, maintaining appropriate speeds according to traffic flows, and aware of dangerous situations. You can drive to your desired destination with proper steering to avoid overtaking and obstacles.

Self-driving cars have become faster than expected due to the development competition between governments and IT companies since the DARPA Challenge, led by the US Defense Agency for Advanced Defense Research in the 2000s.

Various automakers, as well as automakers developing autonomous cars, are planning to launch fully autonomous cars around 2020. According to data from the market research firm IHS, the global annual sales of fully autonomous vehicles is expected to reach 11.8 million units by 2035, up from 230,000 units by 2025.

According to a WHO report, 90% of traffic accidents worldwide are inadvertent and mistaken, with more than 50 million casualties and about $ 3 trillion in social costs each year. The introduction of autonomous cars can reduce accidents caused by the driver's carelessness in addition to the driver's convenience, as well as the reduction of carbon dioxide emissions and the efficient use of energy.

With the development of autonomous vehicles, many studies have been conducted on the technology of positioning autonomous vehicles. The location of autonomous vehicles is generally used by GNSS (global navigation satellite system).

In this case, the system for controlling autonomous driving may include sensor equipment such as a scanning device, a camera, a radar, and the like to recognize and determine in real time the driving environment of a moving object moving at high speed.

As described above, the present invention relates to a system capable of performing autonomous driving. For example, Patent No. 10-1116033 discloses a location receiver configured to receive location data of a main body from a GPS satellite, and a GPS satellite detected by the main body. Disclosed is a technology including a satellite tracking unit configured to track the number of tracks and a driving method determination unit configured to determine a driving method according to the number of tracked GPSs.

In autonomous cars, positioning of cars can be an important prerequisite, but as GPS is used as a basic sensor that occupies a high proportion of the entire navigation system, problems due to satellite navigation errors are occurring.

In addition, the position accuracy of the GNSS is not continuous because the GNSS is affected by signal waiting conditions, satellite distribution, and wireless signal noise. In addition, they may be inaccurate or unavailable in shaded areas such as high-rise buildings and tunnels.

In general, dead reckoning is used to compensate for an unstable position estimation of the GNSS. The dead reckoning is a method of estimating its own position by using the encoder and the inertial navigation device to estimate the moving distance and the direction of the vehicle. However, this method has a problem that accurate position estimation is difficult due to accumulation of measurement errors depending on the driving environment.

Accordingly, the applicant continuously estimates the position of the autonomous vehicle using the inertial navigation, GNSS, and Unscented Kalman Filter (UKF), but the information of the points measured through the road information feature map and the rider (Lidar) The present invention aims to provide a position estimation system and a position estimation method capable of accurate position estimation by applying a measurement value corrected for position error through matching.

1. Korean Registered Patent No. 10-1116033 (announced on March 13, 2012)

An object of the present invention, using the inertial navigation, GNSS and Unscented Kalman Filter (UKF) to estimate the continuous position of the autonomous vehicle, the road information feature map and the point of the point measured through the Ridar The present invention provides a position estimation system and a method for estimating a position thereof, which can accurately estimate a position by applying a measurement value of correcting a position error through matching.

According to an embodiment of the present invention, a system for estimating a position of an autonomous vehicle according to an exemplary embodiment of the present invention provides a navigation position estimator for estimating a vehicle position using a precision map constructed using road information and driving information of a vehicle. According to the government, the presence of GPS short-circuit, and the presence of lanes on the precision map, the error correction unit performing position error correction, the error corrected correction value and the navigation position estimation predicted value are applied to the Unsecented Kalman Filter. And re-estimating the position of the vehicle similar to the reference value of the precision map, and applying the real-time GPS value and the reestimation value to the Kalman Filter according to the driving of the vehicle. have.

In addition, the error correction unit maps the road information measured using the Lidar at the position where there is no lane information or the lane information in the driving route and where the GPS measurement is short-circuited and maps the precision map with a matching value. Azimuth angle for calculating azimuth correction value of a vehicle using a lane when there is a lane detected on the precision map or a lane measured through the lidar when the GPS match is short-circuited in the calculated map matching unit and the driving route. It may include a correction unit.

In addition, the error correction unit is not driven when lane information and GPS are sensed. When the error correction unit is not driven, the re-estimation unit outputs the prediction value as a reestimated value, and the final position estimating unit is the reestimated value when GPS is shorted. Can be output as a final estimate.

In addition, the navigation position estimating may estimate the vehicle position by using the rear wheel movement amount of the vehicle.

In addition, the precision map, using the rider to classify the objects from the ground based on the reference reflectivity, clusters, generates a cumulative map, divides the map to a predetermined size, and divides the divided map into unit cells in each unit cell If the number of included points is less than or equal to the threshold value, it is determined as noise, and each unit cell may be constructed by simplifying the average value.

On the other hand, the position estimation method of the autonomous vehicle according to an embodiment of the present invention comprises the steps of constructing a precise map for the driving of the autonomous vehicle using the road information, the vehicle using the constructed precision map and the driving information of the vehicle A navigation point estimation step for estimating a position, an error correction step for performing a position error correction according to whether there is a GPS short circuit, and whether there is a lane on the precision map, and a dispersion point Kalman filter for calculating the error corrected correction value and the predicted navigation position estimate value. Re-estimating the position of the vehicle similar to the reference value of the precision map by applying to the Unsecented Kalman Filter, and finally estimating the position of the vehicle by applying the real-time GPS value and the reestimated value to the Kalman Filter according to the driving of the vehicle. It may include a step.

In addition, the error correction step, the matching value by map matching the road map and the precision map measured using the Lidar at the position where there is no lane information in the driving route or where there is no lane information and the GPS measurement is short-circuited Computing the azimuth angle correction value of the vehicle using the lane, if there is a lane detected in the precision map or the lane measured through the lidar when the GPS measurement is short-circuited in the map matching step and the driving route It may include a step.

In addition, the error correction step is not performed when the lane information and GPS is sensed, when the error correction step is not performed, the re-estimation step outputs the prediction value as a re-estimation value, and the final position estimation step is a GPS short circuit. The reestimated value may be output as a final estimated value.

In the navigation position estimating step, the vehicle position may be estimated using the rear wheel movement amount of the vehicle.

The building of the precision map may include classifying and clustering objects from the ground based on a reference reflectivity using a lidar, generating a cumulative map, dividing the generated cumulative map into a predetermined size, and dividing. The divided map may be divided into unit cells, and if the number of points included in each divided unit cell is equal to or less than a threshold, noise may be determined, and each unit cell may be simplified to an average value.

As described above, the position estimation system of the autonomous vehicle of the present invention intermittently estimates the position by using inertial navigation, UKF and GPS while intermittently having no lane information (including intersections), at the GPS short position. Accurate position estimation can be achieved by correcting the position error through map matching.

In addition, by simplifying the construction of the precision map, the map accuracy can be increased while reducing the computational load and system load due to redundant data (object points).

In addition, accurate road maps can be constructed by classifying objects by acquiring feature vectors from road markings such as lanes and crosswalks recognized from road information through ANN learning. By using such a precision map, it is possible to further increase the accuracy when tracking the location.

1 is a block diagram illustrating a schematic configuration of a system for estimating a position of an autonomous vehicle according to an embodiment of the present invention.
2 is a view for explaining a measuring point of the lidar (Lidar) according to an embodiment of the present invention.
3 is a flowchart illustrating a method for estimating a location of an autonomous vehicle according to an embodiment of the present invention.
FIG. 4 is a flowchart illustrating a method of constructing an accurate map of FIG. 3.
5 is a view for explaining a position estimation using the rear wheel movement amount according to an embodiment of the present invention.
6 is a diagram for describing an azimuth correction according to an embodiment of the present invention.
FIG. 7 illustrates a test run result using the position estimation method of FIG. 3.
Figure 8 shows the inertial navigation estimation results by applying the GNSS paragraph and error correction.
FIG. 9 is an enlarged graph for each section of FIG. 7.
10A to 10C illustrate RMS graphs of respective sections of FIG. 9.
11A to 11C illustrate azimuth errors for each section of FIG. 9.

The terms or words used in this specification and claims are not to be construed as limiting in their usual or dictionary meanings, and the inventors may appropriately define the concepts of terms in order to best describe their invention. It should be interpreted as meanings and concepts corresponding to the technical idea of the present invention based on the principle that the present invention.

Therefore, the embodiments described in the specification and the configuration shown in the drawings are only the most preferred embodiments of the present invention and do not represent all of the technical idea of the present invention, various equivalents that may be substituted for them at the time of the present application It should be understood that there may be water and variations.

Before describing the present invention with reference to the drawings, it is not shown or specifically described for the matters that are not necessary to reveal the gist of the present invention, that is, can be obviously added by those skilled in the art. Make a note.

1 is a block diagram illustrating a schematic configuration of a system for estimating a position of an autonomous vehicle according to an embodiment of the present invention. Position estimation system (hereinafter, referred to as position estimation system) of the autonomous vehicle according to an embodiment of the present invention is the sensor unit 110, navigation position estimation unit 120, error correction unit 130, re-estimation unit 140 ) And the final estimator 150. In addition, the error correction unit 130 may include a map matching unit 131 and an azimuth correction unit 132.

Meanwhile, in the description of FIG. 1, schematic functions and operations of the components 110 to 150 will be described, and specific operations will be described in detail with reference to FIGS. 3 to 11C.

Sensor unit 110 is a sensor provided in the autonomous vehicle for position estimation, the exercise state measuring sensor 111 for sensing the driving information, such as speed, angular velocity, heading value, moving distance that can be obtained from the vehicle, Lidar 112 and GNSS sensor 113 may be included. Here, the movement state measuring sensor 110 may be a wheel sensor, a steering sensor, an inertial measurement unit (IMU), or the like. The inertial measurement sensor may be composed of a gyroscope, an accelerometer, and a geomagnetic sensor.

Referring to FIG. 2, a lidar 112 is provided at an upper center of an autonomous vehicle to sense road information (object) on a driving route, and an inertial measurement sensor (IMS) and a GNSS sensor 113 are configured to rear axle shafts. Can be installed as a standard.

The sensing values sensed by the sensor unit 110 are stored in real time in a memory (not shown), and the navigation position estimation unit 120, the map matching unit 131, the azimuth correction unit 132, and the final estimation unit 150 are provided. It may be provided as.

The navigation position estimating unit 120 may calculate a predicted value for estimating the vehicle position using the driving information of the vehicle sensed through the precision map and the motion state measuring sensor 111 constructed using the road information. At this time, the navigation position estimation unit 120 may estimate the predicted value by applying the driving information (speed, angular velocity, heading value, moving distance, etc.) of the vehicle to the motion model according to an embodiment of the present invention.

In addition, the position estimation system according to an embodiment of the present invention was applied to build a precise map for accurate position estimation. The precision map according to an embodiment of the present invention is constructed using a lidar, an inertial navigation device and a GNSS sensor, and GPS values sensed by the GNSS sensor may be set as reference position values when constructing the precision map.

Here, the reference position value of the precision map may be used as a reference for position correction in the re-estimation unit 140 and the final estimator 150.

The precision map of the present invention classifies and clusters objects from the ground based on the reference reflectivity using a lidar, generates a cumulative map, divides the map into preset sizes, and divides the divided map into unit cells. If the number of included points is less than or equal to the threshold value, it is determined as noise, and each unit cell may be constructed by simplifying the average value. Meanwhile, the prediction value estimation of the navigation position estimator 120 using the precision map construction and the motion model will be described in detail with reference to FIGS. 2 and 3.

The error correction unit 130 may perform error correction through the map matching unit 131 or the azimuth correction unit 132 according to whether a GPS short circuit exists and whether a lane exists on the precision map.

The map matching unit 131 may calculate a matching value through an error correction that map-matches the road information and the precision map measured using a lidar while driving. In this case, the map matching unit 131 may be driven at a position where there is no lane information, such as an intersection in a driving route, and at a position where GPS measurement is shorted without lane information.

That is, the map matching unit 131 may be driven intermittently to correct an error caused by inertial navigation. Thus, when the map matching unit 131 is not driven, the position estimation may be continuously performed through the navigation position estimating unit 120.

In the present invention, the map matching unit 131 may calculate a matching value using a point-to-point matching method according to an iterative closet point (ICP) algorithm.

Although the GNSS sensor 113 is short-circuited, the azimuth correction unit 132 may calculate an azimuth correction value by correcting the azimuth of the vehicle using the lane information when lane information is detected. The azimuth correction unit 132 calculates the azimuth correction value of the vehicle by using the vehicle proximity lane information detected in the driving route of the vehicle or the lane information measured in the lidar 112 from the precision map, The azimuth included in the input measured value may be changed to the calculated azimuth correction value.

The re-estimation unit 140 may estimate, as the vehicle position, a value close to the reference value among the predicted and measured values of the UKF input by using the unsecented Kalman filter. In this case, the prediction value estimated by the navigation position estimator 120 may be applied as the prediction value input of the UKF, and the matched value or the azimuth correction value calculated by the error correction unit 130 may be applied as the measurement value input of the UKF.

That is, by applying the matching value (which becomes the measured value of UKF) calculated by the map matching unit 131 to the unscented Kalman filter, a value similar to the reference position value on the precision map is used as the vehicle position. Can be reestimated.

Alternatively, when there is an azimuth corrected by the azimuth correction unit 132, the x and y values of the predicted values estimated by the navigation position estimating unit 120 may be applied as they are and may be applied as the azimuth corrected only by the azimuth angle θ. . That is, the UKF measurement value (X value, Y value, and azimuth correction value of the predicted value estimated by the navigation position estimation unit 120) may be applied.

Dispersion point Kalman filter (UKF) uses a deterministic sampling technique known as Unscented Transform (UT) to obtain the minimum set of sample points (called sigma points) around the mean. These sigma points are passed through a nonlinear function and can take the form of finding the mean and covariance for the transformed points.

On the other hand, the error correction unit 130 is not driven when the lane information and GPS is sensed, when the error correction unit 130 is not driven, the recalibration unit 140, the recalibration unit 140 is the navigation position estimation unit 120 By reestimating the predicted value estimated by using the measured value of UKF, the predicted value may be output as the reestimated value.

The final estimator 150 may apply a GPS value measured by the GNSS sensor 113 to make a more stable and accurate position estimation. The final estimator 150 calculates a real-time GPS value (which is a measured value of the Kalman filter) and the reestimated value estimated by the position re-estimation 140 (which is an estimated value of the Kalman filter) according to the driving of the vehicle. It can be applied to the final estimation of the error correction based on the reference position value.

In this case, in order to continuously estimate the Kalman filter at the time of the GPS short, the final position estimation unit 150 may use the reestimation value estimated by the reestimation unit 140 as the measurement value of the Kalman filter to make the final estimation. That is, since both the predicted value and the measured value of the Kalman filter at the time of GPS short-circuit become the reestimated value estimated by the reestimation unit 140, it may be output as the final estimated value.

3 is a flowchart illustrating a method for estimating a location of an autonomous vehicle according to an embodiment of the present invention. FIG. 4 is a flowchart illustrating a method of constructing an accurate map of FIG. 3. In order to understand the description, it may be described with reference to FIGS. 1 and 2.

In the location estimation method according to an embodiment of the present invention, accurate location estimation may be performed by constructing a precision map using road information (S301) and using the location estimation. Referring to FIG. 4, an object may be classified from the ground based on a reference reflectance from points detected through the Lidar measurement (S410) (S420).

,

,

]로 변환될 수 있다. 여기서, α는 라이다의 분해능 각도가될 수 있다.Referring to FIG. 2, the point R measured in the lidar 112 may be expressed as in Equation 1 below.

,

,

Can be converted to]. Where α may be the resolution angle of the lidar.

In general, the curb height is legally 15 cm from the ground. Among the points converted to the vehicle reference, the reflectivity of the points 15 cm or less from the ground becomes 20 or less. Objects may be separated from the ground using the reflectivity and the height H from the ground (S420).

Next, the point data acquired from the LiDAR 112 may be clustered using a Radius Based Nearest Neighbor (RBNN) algorithm (S430). The RBNN calculates the distance between the point and the proximity point using distance as a parameter, and sets it as the new cluster if it is below the threshold and if it is above. When the size of the clustering cluster is 1 or less, it may be determined as noise.

Next, a cumulative map is generated using the clustered point data (S440), the map is divided into a predetermined size (S450), and the divided map is divided into unit cells (S460). The number of points included in each unit cell is shown. If it is equal to or less than the threshold value (S470), it is determined as noise (S480), and each unit cell can be simplified to an average value (S490) to build a precise map (S500).

Meanwhile, in an embodiment of the present invention, the object is classified through map learning using an ANN learning algorithm for the accuracy of object recognition of the constructed precision map. At this time, the lane and road mark information (crosswalk, etc.) using the eigenvectors of the x, y, and z dimensions of the object, the average reflectivity of the object, and the z-axis position of the object through Principal Component Analysis (PCA) Learning was applied.

Using the navigation information of the vehicle sensed through the precision map and the motion state measuring sensor 111 constructed as shown in FIG. 4, the predicted value for estimating the vehicle position may be calculated through navigation position estimation (S302).

At this time, the navigation position estimation unit 120 calculates the movement amount of the rear wheel of the vehicle by applying the driving information (speed, angular velocity, heading value, moving distance, etc.) of the vehicle to the motion model according to an embodiment of the present invention. It is possible to estimate the position estimation predicted by.

에서 차량의 위치는

로 나타낼 수 있다.

는 차량의

좌표 및 방향을 의미하며, 수학식2의

는 차량의 속도

, 각속도

이다. 속도 기반 모델은 다음 수학식3과 같다.5 is a view for explaining a position estimation using the rear wheel movement amount according to an embodiment of the present invention. Referring to Figure 5, the vehicle location

The location of the vehicle in

It can be represented as.

Of the vehicle

It means the coordinates and the direction,

Speed of the vehicle

, Angular velocity

to be. The speed-based model is shown in Equation 3 below.

를 구하는 방법으로 본 발명의 일 실시예에서는 차량의 후륜바퀴의 이동량을 사용하여 속도

를 측정했다. 속도

, 각속도

를 구하는 식은 다음 수학식 4와 같다.The positional movement of the vehicle may be represented as shown in FIG. 5, and the speed of the vehicle

In an embodiment of the present invention to obtain a speed using the amount of movement of the rear wheel of the vehicle

Was measured. speed

, Angular velocity

The equation to obtain is as shown in Equation 4.

는 차량을 위치 이동을 나타내기 위한 함수이고,

는 차량의 위치의 차이로, 직선이동거리 변화량이 된다. 또한,

는 차량의 헤딩(차량의 방향성) 변화량이며, R은 차량의 뒷 차축으로부터 차량 중심점간의 거리이다.Referring to Figure 5, Equation 3 and Equation 4,

Is a function for indicating the positional movement of the vehicle,

Is the difference in the position of the vehicle, and is the linear movement distance change amount. Also,

Is the amount of change in the heading of the vehicle, and R is the distance between the vehicle center point and the rear axle of the vehicle.

)에서 현재 위치(

)로 이동을 하였을 경우, 이전 위치에서 현재위치까지의 이동량을 알면 현재 위치를 예측할 수 있다. 차량의

이동거리를 알기 위해서 차량의 후륜바퀴의 이동량을 통해 속도를 구하였다. 시간은 차량의 위치 이동시 소요된 사간을 사용하였다. 수학식 4를 통해, 각속도를 구하였고 각속도에 대한 수학식 3의

에 대한 공식처럼 시간변화당 각속도의 합을 통해 헤딩을 구하였다.The vehicle is in its previous location (

At your current location (

If you move to), you can predict the current position by knowing the amount of movement from the previous position to the current position. Vehicle

In order to know the moving distance, the speed was calculated from the amount of movement of the rear wheel of the vehicle. The time used was the time spent in moving the vehicle. Through Equation 4, the angular velocity was obtained and

The heading is obtained from the sum of the angular velocities per time change, as in the formula for.

As described above, the navigation position estimated prediction value may be calculated through the motion model. At this time, in the present invention for the continuous position estimation can be estimated using the UKF and KF. When GPS is measured (S303: N) and lane information is detected from the precision map (S305: Y), the predicted value estimated through the motion model may be input as a predicted value of the UKF.

On the other hand, the error correction may be performed through the map matching unit 131 or the azimuth correction unit 132 according to whether the GPS short circuit and whether there is a lane on the precision map. Specifically, if GPS is short-circuited (S303: Y) and lane information is not detected from the precision map or the rider (S305: N), the point group set to be a reference value from the precision map and the point group set detected from the lidar 112 are selected. An error-corrected match value may be calculated by map matching using the ICP algorithm (S306). On the other hand, step S306 is performed when there is no lane information, so it may be performed even when No in step S305.

On the other hand, if the GPS is shorted (S303: Y) and lane information is detected from the precision map or the rider (S305: Y), the azimuth angle may be corrected using the detected lane information and the center of the vehicle (S307). In this case, the azimuth angle is corrected by using lane information of a map without an obstacle, and in the case of a path that does not exist on a precise map, the azimuth angle may be corrected by using point data of the nearest lane measured from LiDAR of the vehicle.

6 is a diagram for describing an azimuth correction according to an embodiment of the present invention. Referring to FIG. 6, the azimuth angle of the vehicle may be corrected using an angle difference between the center line of the two lanes and the center line of the vehicle from the detected lane. In this case, when only one side of the lane is detected, the azimuth angle of the vehicle may be corrected from a point 1.75 m away from the detected lane in consideration of a legal lane area of 3.5 m.

Next, the predicted value calculated by the navigation position estimator (motion model: 120) is input as a UKF predicted value according to whether a GPS short circuit exists and whether there is a lane on the precision map, and the map matching unit 131 or the azimuth correction unit 132 The calculated value can be entered as a UKF measurement. Using the inputted prediction value and the measured value, a value close to a reference value preset according to the UKF algorithm may be calculated as a reestimated value (S308).

On the other hand, the error correction unit 130 is not driven when the lane information and GPS is sensed, and when the error correction unit 130 is not driven, by re-estimating the estimated value estimated in step S302 using the measured value of the UKF, the predicted value Can be output as a reestimation value.

Next, a real-time GPS value (which becomes a measured value of the Kalman filter) and the reestimated value estimated by the position re-estimation 140 (which becomes an estimated value of the Kalman filter) according to the driving of the vehicle are applied to the Kalman Filter. An error-corrected final estimate can be made based on the position value (b309).

In this case, in order to estimate the position of the Kalman filter at the time of GPS short, the re-estimation value estimated in step S308 may be used as the measured value of the Kalman filter. That is, since both the predicted value and the measured value of the Kalman filter at the time of GPS short-circuit become the reestimated value estimated in step S308, it can be output as the final estimated value.

FIG. 7 illustrates a test run result using the position estimation method of FIG. 3. Here, RTK_GNSS is a reference value, (DR) heading_ICP is the error correction of the map matching unit 131, GNSS_DR is estimated only by GNSS and navigation position estimation, Noise_GNSS is estimated only by GNSS measurement Indicates. On the other hand, since the general GNSS cannot simultaneously measure RTK_GNSS with respect to the reference value, Gaussian noise is added to the reference value and used as Noise-GNSS.

At this time, the crosswalk was used to correct the error of the intersection portion where the lane and the lane do not exist to correct the error of the dead reckoning. Therefore, as the vehicle travels, the error varies depending on the distance between the zone where the lane exists and the zone that does not exist, and the dead reckoning characteristics are analyzed according to the four cases as shown in FIG. As shown in Fig. 8, GNSS short-circuited position tracking result was confirmed.

8 shows the results of inertial navigation estimation by GNSS paragraph and no error correction. FIG. 8 relates to a dead reckoning without error update using a distance between a wheel sensor and a rear wheel of a vehicle, and confirms that a path is distorted due to a cumulative error.

FIG. 9 is an enlarged graph for each section of FIG. 7.

Section 1 (upper left) of FIG. 9 is a graph relating to a straight section immediately after the vehicle starts. In the case of (DR) heading_ICP with error correction, it is confirmed that the error of tracking the position of the vehicle is small.

Section 2 (upper right) of FIG. 9 is a curved section where a lane does not exist as a first intersection after the vehicle is driven. It was confirmed that an error occurred in following the location of the vehicle because there was no lane, and it was confirmed that following the normal driving path of the vehicle by correcting the error using the crosswalk through the intersection.

Section 3 (lower left) of FIG. 9 is a curved section without a lane. It can be seen that the error is aggravated by the section with a more severe curve than the section 2, and that the error correction through the crosswalk is slower in convergence than the section 2.

Section 4 (lower right side) of FIG. 9 is a straight section close to the driving end section of the vehicle. It can be seen that the error of position tracking is greatly widened, and that oscillation occurs to follow the position through the lane. When the GNSS was measured, it was confirmed that the position of the driving route based on the entire section was followed.

10A to 10C illustrate RMS graphs of respective sections of FIG. 9. 10A is a dead-measurement RMS graph using lanes when the GNSS is a short circuit. The distance RMS of the position tracking using the lane is 2.11537m, and the distance error about 2.11537 is considered to be the error in the longitudinal direction because the error in the longitudinal direction is not corrected.

Figure 10b is the position estimation RMS when measuring the GNSS and the position tracking RMS of 0.088 confirmed the position estimation results within 8cm.

10C is an RMS graph of the noise GNSS. GNSS data with strong Gaussian noise showed 0.23471, 20cm error, and compared with the RMS result of Fig. 4.10.

11A to 11C illustrate azimuth errors for each section of FIG. 9. FIG. 11A is an azimuth error in the case of GNSS short circuit, FIG. 11B is an azimuth error in GNSS measurement, and FIG. 11C is an azimuth error of noise GNSS data. In the case of GNSS short-circuit, it was confirmed that the error was increased after section 3 through position correction. 11B and 11C, it was confirmed that the azimuth error slightly occurred in the GNSS measurement.

In addition, what was described above using FIG. 1 thru | or 11C was described only the main matter of this invention, and this invention is not limited to the structure of FIGS. 1-11 as many designs are possible within the technical scope. Is self-explanatory.

110: sensor unit 120: navigation position estimation
130: error correction unit 131: map matching unit
132: azimuth correction unit 140: re-estimation
150: final estimator

Claims (10)

A navigation position estimating unit for estimating a vehicle position using a precision map constructed using road information and driving information of the vehicle;
An error correction unit for performing position error correction according to whether a GPS short circuit exists and whether a lane exists on the precision map;
A re-estimation unit for re-estimating the position of the vehicle similar to the reference value of the precision map by applying the error-corrected correction value and the navigation position estimation prediction value to an unsecented Kalman filter; And
And a final estimator for finally estimating the position of the vehicle by applying a real-time GPS value and the reestimation value of the reestimation according to the driving of the vehicle to a Kalman filter.
The method of claim 1,
The error correction part,
A map matching unit configured to calculate a matching value by map-matching the road map and the precision map measured using a Lidar at a position where there is no lane information on a driving route or where GPS information is shorted at a position where GPS measurement is short-circuited; And
And azimuth correction unit that calculates an azimuth correction value of the vehicle using a lane when a lane is detected in the precision map or a lane measured through the lidar when the GPS measurement is shorted in the driving route. Location tracking system of autonomous vehicles.
The method of claim 2,
The error correction unit is not driven when lane information and GPS are sensed.
And when the error correction unit is not driven, the re-estimation unit outputs the prediction value as a re-estimation value, and the final estimation unit outputs the re-estimation value as a final estimated value at the time of GPS short-circuit.
The method of claim 1,
The navigation position estimation unit,
Position tracking system of an autonomous vehicle, characterized in that for estimating the vehicle position using the amount of rear wheel movement of the vehicle.
The method of claim 1,
The precision map,
Classify and cluster objects from the ground based on the standard reflectivity using a rider, create a cumulative map, divide the map into preset sizes, divide the divided map into unit cells, and count the number of points included in each unit cell. A system for tracking the position of an autonomous vehicle, wherein the unit cell is judged as noise, and each unit cell is simplified to an average value.
Constructing a precise map for driving of the autonomous vehicle using road information;
A navigation position estimating step of estimating a vehicle position using the constructed precision map and driving information of the vehicle;
An error correction step of performing a position error correction according to whether a GPS short circuit exists and whether a lane exists on the precision map;
Re-estimating the position of the vehicle that is similar to the reference value of the precision map by applying the error corrected correction value and the navigation position estimation prediction value to an unsecented Kalman filter; And
And a final estimation of the position of the vehicle by applying the real-time GPS value and the reestimated reestimation value according to the driving of the vehicle to the Kalman Filter.

The method of claim 6,
The error correction step,
A map matching step of calculating a matching value by map-matching the road map and the precision map measured using a Lidar at a position where there is no lane information or a lane information in a driving route and where GPS measurement is short-circuited; And
Calculating a azimuth correction value of the vehicle by using a lane when a lane is detected on the precision map or a lane measured through the lidar when the GPS measurement is shorted in the driving route. Position tracking method of autonomous vehicles
The method of claim 7, wherein
The error correction step is not performed if the lane information and GPS is sensed,
When the error correction step is not performed, the re-estimating step outputs the prediction value as a re-estimation value, and the final estimating step outputs the re-estimation value as a final estimated value when GPS is shorted. How to follow.
The method of claim 6,
The navigation position estimation step,
A method for tracking the position of an autonomous vehicle, comprising estimating a vehicle position using a rear wheel movement amount of a vehicle.
The method of claim 6,
The step of building the precision map,
Generating a cumulative map by classifying and clustering objects from the ground based on a reference reflectivity using a lidar;
Dividing the generated accumulated map into a preset size;
Dividing the divided division map into unit cells; And
And determining that the number of points included in each divided unit cell is equal to or less than a threshold value, and simplifying the unit cell to an average value.

Images