KR102219843B1 - Estimating location method and apparatus for autonomous driving - Google Patents

Abstract

One embodiment of the present invention relates to a location estimation device and method for autonomous driving. The location estimation method for autonomous driving comprises the steps of: collecting an image photographed by one or more cameras mounted on an autonomous vehicle and vehicle location-related information obtained from a plurality of sensors; estimating a location of the autonomous vehicle based on the collected image and location-related information and a pre-stored laser scan-based three-dimensional map; and setting a driving route based on a preset target location and the estimated location of the autonomous vehicle. Therefore, the estimation of a precise location is possible.

Description

Location estimation device and method for autonomous driving {ESTIMATING LOCATION METHOD AND APPARATUS FOR AUTONOMOUS DRIVING}

An embodiment of the present invention relates to a location estimation apparatus and method for autonomous driving, and more particularly, a location estimation apparatus for autonomous driving capable of precise location estimation without installing a lidar using a pre-stored 3D map, and It's about how.

Autonomous driving refers to driving to a given destination by controlling the vehicle by recognizing the surrounding environment, determining the driving situation, and controlling the vehicle without driver intervention.

Recently, as vehicles and small mobility devices such as mobile robots are used for smart delivery or security, the demand for mobility devices, robots, and autonomous driving of vehicles has increased.

A key element in autonomous driving is estimating one's own position, and diversification and performance improvement have been made in estimating the position.

As a method of implementing autonomous driving in the past, use a device such as a LiDAR or a camera to create a map in advance and estimate the location on the map using sensor data measured during driving, or use a precise GPS The method of estimating the absolute position by receiving a signal is mainly used.

In the fields of mobility devices, last mile delivery, and security where robots and vehicles are used, the goal is to replace humans and reduce labor costs by using mobility devices, robots and vehicles capable of autonomous driving.

Therefore, it is necessary to reduce the operating cost of the 　mobility 　 device 　 autonomous driving system to less than labor cost, and in this case, the price of the autonomous driving sensor, which accounts for most of the cost of the device, becomes an important consideration. Therefore, for commercialization, there is a need to implement autonomous driving using inexpensive sensors rather than using expensive sensors.

However, since accuracy directly related to safety in autonomous driving is not a negotiable value, LiDAR or high-precision GPS has been used in conventional autonomous driving systems. These sensors are very expensive equipment.

Accordingly, there have been attempts to implement a relatively inexpensive autonomous driving system using only a relatively inexpensive camera. However, in the case of a camera, when estimating a location by extracting a feature point and creating a map, there may be a fatal problem that accuracy is degraded due to the extracted feature point that changes every time according to the amount of light.

The present invention was invented to solve the above problems, and the present invention is a method for estimating a location for autonomous driving of a mobility device, a robot, and a vehicle, and an accurate location estimation method at low cost by fusing data between various sensors on a camera basis, and I want to provide a device.

To this end, according to an embodiment of the present invention, a method for estimating a location for autonomous driving includes the steps of collecting an image photographed by one or more cameras mounted on an autonomous vehicle and location-related information of a vehicle obtained from a plurality of sensors. ; Estimating the location of the autonomous vehicle based on the collected image and location-related information and a pre-stored laser scan-based 3D (dimensional) map; And setting a driving route based on the preset target position and the estimated position of the autonomous vehicle.

In one embodiment, the location-related information may include angular velocity, acceleration, velocity, and GPS position coordinates of the vehicle.

In one embodiment, the 3D (dimensional) map is characterized in that it is a 3D feature point map obtained by extracting 3D feature points from a map generated by scanning a pre-stored laser.

In one embodiment, the step of estimating the position of the autonomous vehicle may include extracting feature points of a 2D image of an image of the camera; And estimating a pose and a direction by matching the feature points of the extracted 2D image to the 3D feature point map.

In one embodiment, the step of matching the feature points is characterized by employing a technique for minimizing a distance between the 3D feature points of the 3D feature point map and the feature points of the extracted 2D image.

In one embodiment, the collecting of the vehicle location related information includes synchronizing sensor periods of the plurality of sensors.

According to another embodiment of the present invention, an apparatus for estimating a location for autonomous driving includes: an information receiving unit that acquires vehicle location-related information from a plurality of sensors and images captured by one or more cameras mounted on an autonomous vehicle; A storage unit for storing a laser scan-based 3D (dimensional) map; And a control unit for estimating the position of the autonomous vehicle based on the 3D map, the collected image and location-related information, and setting a driving route based on a preset target position and the estimated position of the autonomous vehicle.

In an embodiment, the controller includes: a position estimation module for estimating a pose and a direction by extracting feature points of a 2D image of the camera image and matching feature points of the extracted 2D image to the 3D feature point map; And a route setting module for setting a driving route based on the preset target position and the estimated current position of the autonomous vehicle.

In one embodiment, the location estimation module is characterized by employing a technique of minimizing a distance between a 3D feature point of the 3D feature point map and a feature point of the extracted 2D image.

In an embodiment, the control unit may further include a sensor synchronization module for synchronizing sensor periods of the plurality of sensors.

According to the present invention, it is possible to accurately estimate a location without mounting expensive sensors such as LiDAR or precise GPS.

1 is a diagram schematically showing the configuration of a position estimation apparatus for autonomous driving according to an embodiment of the present invention.
2 is a view showing a plurality of sensors according to an embodiment of the present invention.
3 is a diagram schematically showing the configuration of a control unit of a position estimation apparatus for autonomous driving according to an embodiment of the present invention.
4 is a flowchart illustrating a method of estimating a location for autonomous driving according to an embodiment of the present invention.
5 is a partial flow chart illustrating a method of positioning an autonomous moving object according to an embodiment of the present invention.
6 is a block diagram illustrating a configuration of a mobile body equipped with a position estimation apparatus for autonomous driving according to an embodiment of the present invention.
7 is a view for explaining the driving of a moving object equipped with a position estimation device for autonomous driving according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. The configuration of the present invention and its effect will be clearly understood through the detailed description below. Note that prior to the detailed description of the present invention, the same components are denoted by the same reference numerals as possible even if they are displayed on different drawings, and a detailed description will be omitted when it is determined that the gist of the present invention may be obscure for known configurations. do.

1 is a diagram schematically showing a configuration of a position estimation apparatus for autonomous driving according to an embodiment of the present invention, FIG. 2 is a view showing a plurality of sensors according to an embodiment of the present invention, and FIG. A diagram schematically showing the configuration of a control unit of a position estimation apparatus for autonomous driving according to an embodiment of the present invention.

As shown in FIG. 1, the position estimation apparatus 100 for autonomous driving according to the present invention may include an input unit 110, an information receiving unit 120, a control unit 140, and a storage unit 160.

The input unit 110 converts a user's input operation into an input signal and transmits it to the control unit 140. The input unit 110 may be implemented as, for example, a keyboard, a mouse, a touch sensor on a touch screen, a touch pad, a keypad, a voice input, and other input processing devices that are possible in the present, in the past, or in the future. The input unit 110 receives destination information, for example, and transmits it to the control unit 140.

The information receiving unit 120 acquires an image photographed by one or more cameras 10 mounted on an autonomous vehicle and location-related information of a vehicle from a plurality of sensors 20. The information receiving unit 120 transmits it to the control unit 140.

Here, the vehicle location-related information includes an image of a camera, an angular velocity of the vehicle, acceleration, speed, and GPS position coordinates.

Here, at least one camera 10 may be provided with two to acquire a pair of stereo images, for example, a left image and a right image. Here, the term “two cameras” means acquiring images at two views using two different lenses, and includes a case of acquiring two images by having two lenses in one image acquisition device. Of course. In this case, the camera 10 may be mounted to have a preset region of interest. Here, the region of interest may be an area to be photographed using the camera 10 provided in the autonomous vehicle. For the ROI of the two cameras, a lens having a wide angle may be adopted to capture an image of a view angle or an omnidirectional image required during driving.

In another embodiment, one camera may be provided that generates an image of an angle of view or an omnidirectional image required when the one or more cameras 10 drive. Here, one camera can rotate the lens direction to generate an omnidirectional image.

In this case, the camera 10 may be mounted to have a preset region of interest. Here, the region of interest may be an area to be photographed using the camera 10 provided in the autonomous vehicle.

Referring to FIG. 2, the plurality of sensors 20 may include an IMU (Inertial Measurement Unit) 21, a GPS module 22, and a driving recorder 23.

The IMU 21 may measure inertial information of the autonomous driving robot, such as location and posture. The IMU 21 may include, for example, a gyro sensor and an acceleration sensor.

The gyro sensor collects angular velocity information of the mobile robot. It is preferable that the gyro sensor is a 3-axis sensor. That is, the gyro sensor may collect angular velocity information of three axes of the x-axis, y-axis, and z-axis.

The angular velocities of the three axes of the x-axis, y-axis, and z-axis detected by the 3-axis gyro sensor are called Roll, Pitch, and Yaw, respectively. Specifically, rotation about the x-axis is called roll, rotation about the y-axis is called pitch, and rotation about the z-axis is called yaw. The three-dimensional angle information to be obtained in the present invention means a roll, a pitch, and a yaw.

The acceleration sensor measures the acceleration due to gravity acceleration and the movement of the autonomous robot. Like the gyro sensor, the acceleration sensor is preferably a 3-axis sensor. That is, the acceleration sensor may collect acceleration information about three axes of the x-axis, y-axis, and z-axis.

The GPS module 22 is used to recognize the current position of the position estimating device for autonomous driving, and to obtain information on the position of the position estimating device.

In one embodiment, the location estimation apparatus 100 does not require high precision of the GPS module 22. For example, when used alone outdoors-a GPS module 22 having an error of about 10m or more can be adopted. Here, the error of 10 m is an example and is not limited thereto.

The driving recorder 23 stores information such as a driving date, time, speed, and engine speed.

In another modification, a communication module (not shown) may be further included. The communication module is a module for communicating with a base station for communication services such as 3G, 4G, 5G, or LTE, and may acquire approximate location information of itself based on a location on the base station.

Referring back to FIG. 1, the controller 140 controls each component of a position estimation apparatus for autonomous driving to generate position information for autonomous driving. Each component of the control unit 140 is composed of a set of hardware and software capable of performing the above functions.

Referring to FIG. 3, the controller 140 may include a position estimation module 142, a path setting module 144, and a sensor synchronization module 146 according to functions.

However, each component of the control unit 140 does not necessarily have to have a physically different configuration. Each component of the control unit 140 may be configured with one processor, and may be separated only by software or may be executed by separating execution routines. That is, each component of the control unit 140 is only functionally classified, and can be realized on the same hardware.

The location estimation module 142 extracts feature points of a 2D (dimensional) image of the camera image, and estimates a pose and orientation by matching feature points of the extracted 2D image to a 3D feature point map. Here, the 3D feature point map is a 3D feature point map obtained by extracting 3D feature points from a 3D map generated by scanning a pre-stored laser. In another modified example, it is possible to save memory by storing only the 3D feature point map without storing the 3D map.

The position estimation module 142 may employ a technique of minimizing a distance between a 3D feature point of the 3D feature point map and a feature point of the extracted 2D image in order to estimate a pose and an orientation. An example of a technique for minimizing distance is the least squares method. The least squares method is a widely known method, and a detailed description thereof is omitted.

The route setting module 144 sets a driving route based on a preset target position and an estimated current position of the autonomous vehicle. The driving route setting may be set according to a first predetermined period, but may be set according to a second period shorter than the first period when detecting a driving attention section, for example, a corner, a range, or an obstacle. Here, the driving attention section may be set by the position estimation module 142 of the control unit 140. The position estimation module 142 transmits the set period to the path setting module 144.

The sensor synchronization module 146 synchronizes the sensor cycles of the plurality of sensors before collecting measurement values from the plurality of sensors.

Referring back to FIG. 1, the storage unit 160 stores or converts information necessary for position estimation by a position estimation apparatus for autonomous driving into a database. For example, the storage unit 160 may store a Laser Scanned map and/or a 3D feature point map.

In addition, the storage unit 160 may transmit stored information to the controller 140 at the request of the controller 140.

Hereinafter, as another embodiment, a method for estimating a location for autonomous driving according to the present invention having the configuration as described above will be described in detail with reference to FIG. 4.

4 is a flowchart illustrating a method for estimating a location for autonomous driving according to an embodiment of the present invention, and FIG. 5 is a partial flow chart for explaining a method for positioning an autonomous moving object according to an embodiment of the present invention. .

Referring to FIG. 4, in step S110, the position estimation apparatus for autonomous driving collects an image photographed by one or more cameras mounted on an autonomous vehicle and location-related information of a vehicle obtained from a plurality of sensors. Here, the plurality of sensors may include an IMU, a GPS module, and a driving recorder, as described above. Discrete data input is received from the above camera and GPS module, and data input is continuously received from the driving recorder or IMU. Therefore, it is necessary to perform synchronization with the data acquisition period of the camera and the GPS module and data from the driving recorder or IMU. Sensor synchronization can adopt a widely known method.

In step S120, the position estimation apparatus for autonomous driving estimates the position of the autonomous moving object based on the collected image and position-related information and a pre-stored laser scan-based 3D (dimensional) map.

Hereinafter, a method of estimating a location will be described in detail with reference to FIG. 5.

In step S121, first, the initial position is estimated by fusing the respective position-related information measured from a plurality of sensors.

In one embodiment, GPS location information, that is, approximate longitude and latitude information may be obtained and corrected to precise coordinates using the attitude data of the IMU.

A filter-based algorithm such as a Kalman filter or a factor graph can be used to fuse (correct) each location-related information.

The Kalman filter is a method of obtaining a final estimate of a state variable in order to cancel the error of the measured value, and is divided into two steps: a prediction and a correction process.

Briefly, first, an initial value is selected based on the measured values of the sensors.

Then, the estimated value and error covariance are predicted for the selected initial value.

And Kalman gain is calculated using the predicted estimated value and error covariance value.

When the Kalman gain is calculated in this way, an actual estimated value of the predicted estimated value is calculated, and then an error covariance value is calculated. As disclosed in Application No. 10-2018-0097319, etc., the method of estimating the position by fusion of each of these sensors is a known technique in the related art, and thus a detailed description thereof will be omitted.

When using a factor graph, use the graph to define the problem. A node corresponds to the robot's pose, and an edge means a spatial relationship between nodes. In addition, the graph-based position estimation refers to a node setting that minimizes an error between relationships expressed through edges, that is, a process of finding a pose of a moving object. This process is divided into a front end stage called Graph Construction and a back end stage called Graph Optimization.

를 통해 얻은 현재 로봇의 상태와 측정된

와 실제 측정

사이의 오차

를 최소화하는 과정을 통해 로봇의 현재 상태 포즈를 추정하게 된다. 이러한 최소화과정은 하기의 수학식 1과 같이 나타낼 수 있다.

The current robot status and measured

And real measurements

Error between

The current state pose of the robot is estimated through the process of minimizing. This minimization process can be expressed as Equation 1 below.

[Equation 1]

Hereinafter, as another embodiment, a method for estimating a location for autonomous driving according to the present invention having the configuration as described above will be described in detail with reference to FIG. 4.

4 is a flowchart illustrating a method for estimating a location for autonomous driving according to an embodiment of the present invention, and FIG. 5 is a partial flow chart for explaining a method for positioning an autonomous moving object according to an embodiment of the present invention. .

Referring to FIG. 4, in step S110, the position estimation apparatus for autonomous driving collects an image photographed by one or more cameras mounted on an autonomous vehicle and location-related information of a vehicle obtained from a plurality of sensors. Here, the plurality of sensors may include an IMU, a GPS module, and a driving recorder, as described above. Discrete data input is received from the above camera and GPS module, and data input is continuously received from the driving recorder or IMU. Therefore, it is necessary to perform synchronization with the data acquisition period of the camera and the GPS module and data from the driving recorder or IMU. Sensor synchronization can adopt a widely known method.

In step S120, the position estimation apparatus for autonomous driving estimates the position of the autonomous moving object based on the collected image and position-related information and a pre-stored laser scan-based 3D (dimensional) map.

Hereinafter, a method of estimating a location will be described in detail with reference to FIG. 5.

In step S121, first, the initial position is estimated by fusing the respective position-related information measured from a plurality of sensors.

In one embodiment, GPS location information, that is, approximate longitude and latitude information may be obtained and corrected to precise coordinates using the attitude data of the IMU.

A filter-based algorithm such as a Kalman filter or a factor graph can be used to fuse (correct) each location-related information.

The Kalman filter is a method of obtaining a final estimate of a state variable in order to cancel the error of the measured value, and is divided into two steps: a prediction and a correction process.

Briefly, first, an initial value is selected based on the measured values of the sensors.

Then, the estimated value and error covariance are predicted for the selected initial value.

And Kalman gain is calculated using the predicted estimated value and error covariance value.

When the Kalman gain is calculated in this way, an actual estimated value of the predicted estimated value is calculated, and then an error covariance value is calculated. As disclosed in Application No. 10-2018-0097319, etc., the method of estimating the position by fusion of each of these sensors is a known technique in the related art, and thus a detailed description thereof will be omitted.

When using a factor graph, use the graph to define the problem. A node corresponds to the robot's pose, and an edge means a spatial relationship between nodes. In addition, the graph-based position estimation refers to a node setting that minimizes an error between relationships expressed through edges, that is, a process of finding a pose of a moving object. This process is divided into a front end stage called Graph Construction and a back end stage called Graph Optimization.

를 통해 얻은 현재 로봇의 상태와 측정된

와 실제 측정

사이의 오차

를 최소화하는 과정을 통해 로봇의 현재 상태 포즈를 추정하게 된다. 이러한 최소화과정은 하기의 수학식 1과 같이 나타낼 수 있다.

The current robot status and measured

And real measurements

Error between

The current state pose of the robot is estimated through the process of minimizing. This minimization process can be expressed as Equation 1 below.

[Equation 1]

를 찾는 과정으로 요약할 수 있다. 이러한 음의 로그 가능도 F(x)는 하기의 수학식 2와 같이 나타낼 수 있다. The error is expressed by the above equation, and in order to solve this problem with the maximum likelihood approach, a node that minimizes all negative log likelihood F(x) for all measurements.

It can be summarized as the process of finding. This negative log likelihood F(x) can be expressed as Equation 2 below.

[Equation 2]

To calculate the negative log likelihood F(x) for all measurements, use the above equation.

[Equation 3]

And through Equation 3, a node x that minimizes F(x) can be found.

In the process of minimization, the Gauss-Newton method or the Levenberg-Marquardt algorithm is mainly used.

In step S122, the image provided from the camera is preprocessed to correct distortion.

In step S123, 2D feature points are extracted from the 2D image of the camera. In this case, a 2D feature point plane connecting the 2D feature points or a 2D feature point line connecting the 2D feature points may be extracted. Here, the feature point may include a corner, that is, a vertex, but is not limited thereto.

In step S124, the 3D feature point planes connecting the 3D feature points extracted from the 3D map or the 3D feature point lines connecting the 3D feature points are matched with the 2D feature point planes or 2D feature point lines extracted in step S123. In another variation, matching may be performed by projecting the 2D feature point line onto the 3D feature point line.

In step S125, the posture and direction are estimated through the process of minimizing the distance between the matched 2D feature point and the 3D feature point based on the corrected coordinates.

In this case, an object that does not exist on the 3D feature point map but is recognized on the 2D feature point map may be estimated as an obstacle. When the obstacle is estimated as described above, the path setting period may be changed shortly or the appearance of the obstacle may be output as will be described later. Thereby, the connected moving object can respond to the appearance of an obstacle. The response to the appearance of an obstacle may include any one or more of ringing a clock, reducing speed, and stopping for a while.

Returning to FIG. 4 again, in step S130, a driving route is set based on the preset target position and the estimated position of the autonomously moving object.

6 is a block diagram for explaining the configuration of a moving object equipped with a position estimation device for autonomous driving according to an embodiment of the present invention according to an embodiment of the present invention, and FIG. This is a view for explaining the driving of a mobile object equipped with a position estimation device for autonomous driving.

6 and 7, a moving object (hereinafter referred to as a moving object) equipped with a position estimation device for autonomous driving includes one or more cameras 10, a plurality of sensors 20, a position estimation device 100, and a controller ( 200) and a driving device 300 may be included. As described in the above embodiment, at least one camera 10 and a plurality of sensors 20 transmit information to the position estimating device 100, and the position estimating device 100 and the moving object 200 transmit and receive information. I can. In addition, the controller 200 may transmit a control command to the driving device 300.

The camera 10, the plurality of sensors 20, and the position estimating device 100 are the same as the camera 10, the plurality of sensors 20, and the position estimating device 100 described in FIG. 1, and thus detailed descriptions thereof will be omitted. However, the location estimation apparatus 100 discretely sets a path until the moving object arrives at the destination and transmits it to the controller 200.

However, referring to FIG. 7, an example in which one camera is mounted is described. As described above, when only one camera is mounted, the position estimating apparatus 100 or the controller 200 rotates the direction of the camera lens to obtain an image of a required angle of view required during driving. Referring to FIG. 7, it can be confirmed that a 2D image indicated by a dotted line on a 3D map indicated by a solid line is matched.

The controller 200 controls each component of the moving object, and in particular, receives a set path from the position estimation device 100 and transmits a control command to the driving device 300 according to the set path.

The controller 200 may control and drive the driving device 300 until the moving object reaches the target position, and may terminate the driving when it reaches the destination.

The driving device 300 includes a power source, a wheel, a steering device, a brake, etc. according to a control command of the controller 200, and controls the steering and speed of the moving object.

Here, the power source may be variously adopted according to the size and function of a moving body, such as an electric-based motor or a fossil fuel-based engine.

According to an embodiment of the present invention, by adopting several sensors that are relatively inexpensive compared to sensors of a conventional autonomous driving system, position estimation through the fusion of measured values of the sensors and optimizing the position value through a pre-made Laser Scanned map. Through the process, it enables precise position estimation without using expensive sensors such as LiDAR or precise GPS directly on small mobility devices.

Since it uses 3D information (Point, Line, Depth) that can be used in position estimation using LiDAR using data from various sensors and laser scanned maps, it does not use a laser scanner sensor, but the advantage of using a laser scanner sensor Because (accurate depth information, 3D feature points) can be obtained, resources can be saved.

At this time, it will be appreciated that each block of the flowchart diagrams and combinations of the flowchart diagrams may be executed by computer program instructions. Since these computer program instructions can be mounted on the processor of a general purpose computer, special purpose computer or other programmable data processing equipment, the instructions executed by the processor of the computer or other programmable data processing equipment are described in the flowchart block(s). It creates a means to perform functions. These computer program instructions can also be stored in computer-usable or computer-readable memory that can be directed to a computer or other programmable data processing equipment to implement a function in a particular way, so that the computer-usable or computer-readable memory It is also possible to produce an article of manufacture containing instruction means for performing the functions described in the flowchart block(s). Computer program instructions can also be mounted on a computer or other programmable data processing equipment, so that a series of operating steps are performed on a computer or other programmable data processing equipment to create a computer-executable process to create a computer or other programmable data processing equipment. It is also possible for instructions to perform processing equipment to provide steps for executing the functions described in the flowchart block(s).

In addition, each block may represent a module, segment, or part of code that contains one or more executable instructions for executing the specified logical function(s). In addition, it should be noted that in some alternative execution examples, functions mentioned in blocks may occur out of order. For example, two blocks shown in succession may in fact be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order depending on the corresponding function.

In this case, the term'~ unit' used in the present embodiment means software or hardware components such as FPGA or ASIC, and'~ unit' performs certain roles. However,'~ part' is not limited to software or hardware. The'~ unit' may be configured to be in an addressable storage medium, or may be configured to reproduce one or more processors. Thus, as an example,'~ unit' refers to components such as software components, object-oriented software components, class components and task components, processes, functions, properties, and procedures. , Subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, database, data structures, tables, arrays, and variables. The components and functions provided in the'~ units' may be combined into a smaller number of elements and'~ units' or further separated into additional elements and'~ units'. In addition, components and'~ units' may be implemented to play one or more CPUs in a device or a security multimedia card.

Those of ordinary skill in the art to which the present specification pertains will appreciate that the present specification may be implemented in other specific forms without changing the technical spirit or essential features thereof. Therefore, it should be understood that the embodiments described above are illustrative and non-limiting in all respects. The scope of the present specification is indicated by the scope of the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning and scope of the claims and the concept of equivalents thereof are included in the scope of the present specification. It must be interpreted.

Meanwhile, in the present specification and drawings, preferred embodiments of the present specification have been disclosed, and although specific terms are used, these are merely used in a general meaning to easily describe the technical content of the present specification and to aid understanding of the invention. It is not intended to limit the scope of the specification. In addition to the embodiments disclosed herein, it is apparent to those of ordinary skill in the art that other modified examples based on the technical idea of the present specification may be implemented.

100: position estimation device
110: input unit
120: information receiver
140: control unit
142: position estimation module
144: path setting module
146: sensor synchronization module
160: storage unit

Claims (12)

Collecting an image photographed by one or more cameras mounted on an autonomous vehicle and location-related information of a vehicle obtained from a plurality of sensors;
Estimating the location of the autonomous vehicle based on the collected image and location-related information and a pre-stored laser scan-based 3D (dimensional) map; And
Setting a driving route based on the preset target position and the estimated position of the autonomous vehicle
Including,
The 3D (dimensional) map is a 3D feature point map obtained by extracting 3D feature points from a map generated by scanning a pre-stored laser,
Estimating the position of the autonomous vehicle,
Extracting feature points of a 2D image of the camera image; And
Estimating a pose and direction by matching the feature points of the extracted 2D image to the 3D feature point map
Including a location estimation method for autonomous driving. The method of claim 1,
The location-related information includes an angular velocity, acceleration, velocity, and GPS position coordinates of the vehicle. delete delete The method of claim 1,
The step of matching the feature points,
A method for estimating a location for autonomous driving, characterized in that a technique for minimizing the distance between the 3D feature point of the 3D feature point map and the feature point of the extracted 2D image is adopted. The method of claim 1,
The step of collecting the vehicle location related information,
And synchronizing sensor periods of the plurality of sensors. An information receiving unit that acquires an image photographed by one or more cameras mounted on the autonomous vehicle and location-related information of the vehicle from a plurality of sensors;
A storage unit for storing a laser scan-based 3D (dimensional) map; And
A control unit that estimates the location of the autonomous vehicle based on the 3D map and the collected image and location-related information, and sets a driving route based on a preset target location and the estimated location of the autonomous vehicle.
Including,
The 3D map is a 3D feature point map obtained by extracting 3D feature points from a map generated by scanning a pre-stored laser,
The control unit
A position estimation module for estimating a pose and a direction by extracting feature points of the 2D image of the camera image and matching the feature points of the extracted 2D image to the 3D feature point map; And
A position estimation device for autonomous driving, comprising a path setting module that sets a driving route based on a preset target position and an estimated current position of the autonomous vehicle. The method of claim 7,
The location-related information includes an angular velocity, acceleration, velocity, and GPS position coordinates of the vehicle. delete delete The method of claim 7,
The position estimation module,
A position estimation apparatus for autonomous driving, characterized in that it employs a technique for minimizing the distance between the 3D feature points of the 3D feature point map and the feature points of the extracted 2D image. The method of claim 7,
The control unit
Sensor synchronization module that synchronizes the sensor cycles of multiple sensors
Position estimation apparatus for autonomous driving, characterized in that it further comprises.

Images