KR101947478B1 - Method and Apparatus for Determining Placement for LiDAR Sensor - Google Patents

Abstract

Disclosed are a method and an apparatus for determining the placement for a light detection and ranging (LiDAR) sensor. The present invention relates to a method and an apparatus for determining the placement for a LiDAR sensor, which place a sampling board, perform optimization based on distance information between the sampling board and the LiDAR sensor, and determine the placement of a plurality of LiDAR sensors.

Description

Field of the Invention [0001] The present invention relates to a method and an apparatus for determining a position of a LiDa sensor,

The present embodiment relates to a method and an apparatus for determining an arrangement position of a lidar sensor.

The contents described in this section merely provide background information on the present embodiment and do not constitute the prior art.

In order to drive an autonomous vehicle, various types of sensors are needed to monitor the surroundings on behalf of the driver. In the autonomous navigation technology, a LiDAR (Light Detection and Ranging) sensor is typically used to obtain distance information from an automobile to surrounding objects (obstacles).

Lidar sensor is a technology that can detect the distance, direction, speed, temperature, material distribution and concentration characteristic to an object by irradiating the laser to the object. The accuracy of the distance information is about ± 3 cm, which is higher than other distance sensors such as stereo camera and ultrasonic sensor, and is widely used in autonomous vehicles.

In the past, researches are being conducted on how to maximize the coverage of the sensor in the top view of the vehicle by expressing the surrounding environment in 2D. Therefore, a new definition for expressing the coverage of the 3D render sensor is required, and it is necessary to arrange the render sensor in consideration of the difference of the vertical axis distances between the channels of the render sensor in the 3D environment.

In general, the Lidar sensor can be classified into a high-resolution Lidar sensor of 32 channels or more and a low-resolution Lidar sensor of 16 channels or less. The high-resolution lidar sensor provides a large amount of information in one sensing, but it is ten times more expensive than the low-resolution ladar sensor. Therefore, there is a need for a method of arranging the Lidar sensor for improving the accuracy using a low-resolution Lidar sensor.

In this embodiment, a layout position determination method of LRITA sensors and a device therefor are used to determine placement positions of a plurality of LRD sensors by arranging a sampling board and performing optimization based on distance information between the sampling board and LRD sensors There is a main purpose in providing.

According to an aspect of the present invention, there is provided an apparatus for determining a placement position of a plurality of Ridasensors, the apparatus comprising: Sampling board setting unit; An occupied grid forming unit for forming an occupied grid on the sampling board based on the target information; An optimization processing unit for fixing an arrangement position of the first Rather sensor based on a position of the first Rather sensor and adjusting an arrangement position of the Rather than the first Rather sensor among the plurality of Rather sensors to estimate an optimized placement position; And a placement position determiner for determining a final placement position of the plurality of row sensors based on the estimated placement positions.

According to another aspect of the present invention, there is provided a method for determining an arrangement position of a plurality of Lattice sensors by a placement position determination apparatus, the method comprising the steps of: A sampling board setting step of setting target information about the target board; Forming an occupancy grid on the sampling board based on the target information; An optimization process for estimating an optimized placement position by adjusting placement positions of the first Rather sensors except for the first Rather sensor among the plurality of Rather sensors, ; And a placement position determination step of determining placement positions of the plurality of row sensors based on the estimated placement positions.

As described above, according to the present embodiment, the occupancy grid is formed by using a plurality of sampling boards (square boards), whereby the coverage according to the arrangement of the Lidar sensors can be quantitatively measured.

In addition, according to this embodiment, by changing the size of the cell constituting the occupation grid according to the distance between the sampling board and the Lidar sensor, even if the interval between the channels according to the vertical axis resolution is changed, it is possible to quickly respond to the change .

According to the present embodiment, there is an effect that the weight of some sampling boards is given to calculate the occupancy of the grid, thereby optimizing the user's area of interest.

In addition, according to the present embodiment, an optimal algorithm is used to change the position of the Lidar sensor and find an optimal placement position, thereby improving the number of channels. According to the present embodiment, there is an effect that it is possible to perform arrangement for improving the precision of two or more Lidar sensors by repeatedly applying the arrangement method for improving the accuracy of the two Lidar sensors.

Figs. 1A to 1C are diagrams for explaining the difference between the prior art and the present invention.
FIG. 2 is a block diagram schematically showing an apparatus for determining the arrangement position of a 3D Laydioc sensor according to the present embodiment.
3 is a flowchart for explaining a method of determining the placement position of the 3D Raidas sensor according to the present embodiment.
4A and 4B are diagrams for explaining a procedure of setting a sampling board according to the present embodiment.
5A and 5B are diagrams for explaining the process of forming the occupation grid according to the present embodiment.
6 is an exemplary diagram for explaining an optimization process according to the present embodiment.
FIG. 7 is an exemplary view showing an arrangement position of the sampling board according to the present embodiment.
FIG. 8 is an exemplary diagram illustrating a sampling board of an optimized 3D RAY sensor according to the present embodiment.

Hereinafter, the present embodiment will be described in detail with reference to the accompanying drawings. In explaining the present invention, '... Quot ;, " module ", and " module " refer to a unit that processes at least one function or operation, and may be implemented by hardware or software or a combination of hardware and software.

The present embodiment describes a LiDAR (Light Detection and Ranging) sensor disposed in an autonomous vehicle. However, this embodiment is applicable to various fields using two or more Lidar sensors. It should be noted that although the present embodiment describes that the placement position is determined based on two Lidar sensors, the present invention is not limited to this and is applicable to a method of arranging a plurality of Lidar sensors.

Figs. 1A to 1C are diagrams for explaining the difference between the prior art and the present invention.

In the autonomous navigation technology, a LiDAR (Light Detection and Ranging) sensor is typically used to obtain distance information from an automobile to surrounding objects (obstacles). Lidar sensor is a technology that can detect the distance, direction, speed, temperature, material distribution and concentration characteristic to an object by irradiating the laser to the object. The accuracy of the distance information is about ± 3 cm, which is higher than other distance sensors such as stereo camera and ultrasonic sensor, and is widely used in autonomous vehicles.

In general, the Lidar sensor can be classified into a high-resolution Lidar sensor of 32 channels or more and a low-resolution Lidar sensor of 16 channels or less. The high-resolution lidar sensor provides a large amount of information in one sensing, but it is ten times more expensive than the low-resolution ladar sensor. Therefore, in the present invention, a method of determining the layout position of the LIDAR sensor will be described, in which a plurality of low-resolution Lattice sensors are arranged to derive a high-precision result of the high-resolution Lattice sensor.

In the present invention, the arrangement of the two 3D Lidar sensors is determined, and two vertical 3D Lidar sensors of low channel (low resolution) based on the arrangement position are used to improve the vertical axis resolution in the near region, The same result as that using the sensor can be derived.

FIGS. 1A through 1C compare techniques using a high resolution single row sensor 110 and techniques using a plurality of low resolution LR sensors 210 and 212. FIG.

FIGS. 1A to 1C are views showing an autonomous vehicle using a single LR sensor 110. FIGS. 1A to 1C are diagrams showing autonomous traveling vehicles using a plurality of LR sensors 210 and 212, Fig.

As shown in FIG. 1A, by using a plurality of Lidar sensors 210 and 212, a distance range for acquiring information on the surrounding environment of the autonomous vehicle is widened. That is, by distributing at least one or more Lidar sensors to the autonomous vehicle at a certain distance, a wider range of distance information can be obtained.

Further, as shown in FIG. 1B, the use of the plurality of Lidar sensors 210 and 212 widens the range of the viewing angle that can sense the surroundings of the autonomous vehicle. That is, it is possible to reduce the blind spot of the autonomous vehicle caused by the mounting of the single Lada sensor 110 by using the plurality of Lada sensors 210 and 212.

1C, by using a plurality of LRID sensors 210 and 212, the vertical axis resolution of the measurement information can be improved in the near region where the overlap of the distance information with respect to the target object occurs.

FIG. 2 is a block diagram schematically showing an apparatus for determining the arrangement position of a 3D Laydioc sensor according to the present embodiment.

In the 3D RL sensor system 200 according to the present embodiment, a method and apparatus for positioning RL diagonal sensors for improving the accuracy of the two RL sensors 210 and 212 are proposed. The 3D ray sensor system 200 determines the positions of the two Lidar sensors 210 and 212 so as to improve the vertical axis resolution of the near region. The 3D Lidar sensor system 200 can derive the same result as using a high resolution Lidar sensor using a plurality of low resolution Lidar sensors depending on the arrangement position.

The 3D Lidar sensor measures the distance to an object (obstacle) around it at a certain angle (angle). The constant angle is divided into horizontal and vertical angular resolution according to the axis. The horizontal angle / vertical angle resolution allows two targets to be identified individually by angle measurement, and two targets within the same distance resolution cell of the same intensity can be identified and separated.

The distance information measured by the 3D lidar sensor appears as a set of points (set of points) in 3D space. In an actual environment, an example of measuring the distance to an object using a 3D Ridar sensor is shown in (a) of FIG. 4A. 4A shows the horizontal resolution, and FIG. 4C shows the vertical resolution.

As shown in FIG. 4A, the horizontal angle resolution of the 3D ray sensor is defined as θ _hor , and the vertical angle resolution of the 3D ray sensor is defined as θ _ver , as shown in FIG. 4A (c) do.

In measuring the distance information of the 3D render sensor, one light source is defined as a channel, and the 3D render sensor has a plurality of channels. 4 (d) shows a part of distance information (Distance Information) of the same channel by a dashed line box.

Assuming that the set of points for the distance information measured by the 3D render sensor is defined by PC and the number of horizontal measurements is N and the number of channels is M, ] Can be expressed as follows.

In addition, any point p _{i, j} (i ∈ {1, ..., N}, j ∈ {1, ..., M}) constituting the PC can be expressed as: .

Here, x _{i, j} , y _{i, j} , z _{i, j} represent the 3D coordinates of the point p _{i, j} , and r _{i, j} represents the reflectivity. Reflectivity refers to the energy ratio of Reflected Light to Incident Light, which depends on the material properties of the object.

The 3D lidar sensor measures the distance information using the laser beam, so that the reflectivity can be measured.

The 3D RL sensor system 200 according to the present embodiment includes a first RL sensor 210, a second RL sensor 212, a sampling board 220, and a positioning device 230. The placement positioning apparatus 230 includes a sampling board setting unit 240, an occupancy grid forming unit 250, an optimization processing unit 260, and a placement position determination unit 270 to determine a placement position of the 3D ray sensor. . The 3D ray sensor system 200 shown in FIG. 2 is according to one embodiment, and not all of the blocks shown in FIG. 2 are required, and in another embodiment, the 3D ray sensor system 200 shown in FIG. Some blocks may be added, changed or deleted.

The first RLayer sensor 210 and the second RLayer sensor 212 scan the surrounding environment using a laser beam (laser pulse) having properties close to the radio wave and sense an object (obstacle).

The first Lidar sensor 210 and the second Lidar sensor 212 may be provided in an autonomous vehicle for monitoring the four sides of the autonomous vehicle. The 3D Lidar sensor system 200 uses two Lada sensors such as the first Lada sensor 210 and the second Lada sensor 212. However, the present invention is not limited thereto.

The first RLayer sensor 210 and the second RLayer sensor 212 may be low-resolution RL sensors of 16 channels or less. Here, the first RL sensor 210 means a RL sensor fixed at a predetermined position as a reference sensor, and the second RL sensor 212 means a first RL sensor 210 (Adjusts) the position based on the position of the LIDAR sensor. If a plurality of Lidar sensors are used in the 3D Lidar sensor system 200, the number of Lidar sensors which move the position is increased like the second Lidar sensor 212.

The sampling board 220 is a board disposed for optimization of the Lidar sensors 210 and 212 and means a board on which distance information measured by the Lidar sensors 210 and 212 is projected. The sampling board 220 is preferably a rectangular board, and may be implemented in a solid color.

The size of the sampling board 220 can be changed in consideration of the vertical axis resolution, the distance to the first Lada sensor 210, and the like. The sampling board 220 is disposed within the measurement range of the Lidar sensors 210 and 212, and the number, the distance, and the like of the square boards can be determined by a user's operation.

The placement position determination device 230 determines the placement position of the Lidar sensors 210 and 212 in order to improve the accuracy of the Lidar sensors 210 and 212. [ The placement position determination device 230 can determine the placement positions of the LRID sensors 210 and 212 based on the distance information between the Lidar sensors 210 and 212 and the sampling board and the optimization algorithm. The placement positioning apparatus 230 according to the present embodiment includes a sampling board setting unit 240, an occupancy grid forming unit 250, an optimization processing unit 260, and a placement position determination unit 270.

The sampling board setting unit 240 sets target information related to the sampling board 220 arranged to improve the accuracy of the Lidar sensors 210 and 212. Here, the sampling board setting unit 240 sets various information about the sampling board 240 disposed around the sensor 210, which is the first one of the plurality of the Latha sensors 210 and 212.

The sampling board setting unit 240 may set the sampling board 220 to a predetermined size based on the size of the sampling board 220, the distance between the sampling board 220 and the first Lada sensor 210, the number of channels of the first Lada sensor 210, The vertical axis resolution of the display unit 210, and the like. The sampling board setting unit 240 may set all information related to the sampling board 220, but is not limited thereto. For example, the sampling board setting unit 240 may acquire some information among all the information related to the sampling board 220 from a user or an external device, and may calculate and set remaining information based on some information. Specifically, the sampling board setting unit 240 sets the sampling board 220 based on the distance between the sampling board 220 and the first Lada sensor 210, the number of channels of the first Lada sensor 210, Resolution and the resolution of the sampling board 220 to calculate the height and width of the sampling board 220 to set the target information.

Since the first RL sensor 210 has a fixed vertical axis resolution, the channel spacing of the vertical axis varies depending on the distance between the first RL sensor 210 and the sampling board 220. The operation of varying the channel spacing of the vertical axis according to the distance between the first RLayer sensor 210 and the sampling board 220 is shown in (c) of FIG. 4b. For example, when the distance between the first RLayer sensor 210 and the sampling board 220 is shortened, the channel spacing on the vertical axis becomes narrow and the distance between the first RLayer sensor 210 and the sampling board 220 becomes long The channel spacing of the vertical axis is widened.

The sampling board setting unit 240 sets the size of the sampling board 220 in consideration of the vertical axis resolution, the distance between the first Lidar sensor 210 and the sampling board 220, and the like. Here, the first RL sensor 210 is a sensor for setting a set value, and the second RL sensor 212 may be changed to a sensor according to the user's setting. The distance between the first RLayer sensor 210 and the sampling board 220 means a distance between a representative position of each of the sampling boards 220 and a sending unit (not shown) of the first RLayer sensor 210, The representative position of the sampling board 220 is preferably the center of the sampling board 220.

The sampling board setting unit 240 can set the height and the width of the sampling board 220 and the height and width of the sampling board 220 can be set using Equation 3 and Equation 4 .

, 임의의 샘플링 보드(rb_i)에 투영할 채널(Channel) 수: N, 제1 라이다 센서(210)의 수직축 해상도: θ_ver, θ_ver에서 임의의 샘플링 보드(rb_i)의 높이(Height):

, θ_ver에서 임의의 샘플링 보드(rb_i)의 너비(Width):

)(Arbitrary sampling board: rb _i , distance between any sampling board (rb _i ) and first lidar sensor 210:

, The number of channels (Channel) to be projected on any sampling board (rb _i): N, a first referred to the vertical axis of the sensor 210, the resolution is the height (Height of θ _ver, θ _ver random sampling board (rb _i) in ):

, the width of an arbitrary sampling board (rb _i ) at θ _ver :

)

The sampling board setting unit 240 sets the sampling rate of the first RL sensor 210 and the first RL sensor 210 and the tangent of the vertical resolution of the first RL sensor 210, The height and width of the sampling board 220 can be set to less than the result obtained by multiplying the tangent value. The sampling board setting unit 240 sets the number of channels by subtracting a predetermined number of channels (e.g., one channel) from the number of channels of the first RLayer sensor 210, 210 and the tangent of the vertical axis resolution of the first LR sensor 210. The height of the sampling board 220 can be set to exceed the resultant value.

The sampling board setting unit 240 limits the positions at which the sampling boards 220 are disposed within the measurement ranges of the Lidar sensors 210 and 212. The number and the distance of the sampling boards 220 within the measurement range It can be changed according to the setting of the user.

Occupying grid forming unit 250 performs the operation of forming an occupancy grid on sampling board 220. Occupying grid forming unit 250 forms an occupancy grid on sampling board 220 based on the target information.

The occupancy grid forming unit 250 forms a plurality of cells on the sampling board 220 based on the target information, and calculates a height and a width for each of the plurality of cells. The occupancy grid forming unit 250 may include a height of the cell for forming the occupancy grid based on the distance between the first RL sensor 210 and the sampling board 220 and the vertical resolution of the first RL sensor 210, And width.

Occupying grids formed in the occupancy grid forming portion 250 are shown in Figures 5A and 5B. The occupancy grid includes a plurality of cells, and any cell (j-th cell) is defined as rb _{i, j} .

로 정의되고, 임의의 셀의 너비는

로 정의된다. 점유 그리드 형성부(250)는 라이다 센서(210, 212)의 수직축 해상도가 θ_ver 일 때, [수학식 5]를 이용하여 임의의 셀의 높이 및 너비를 산출할 수 있다.Occupying grid forming section 250 calculates the height and width of any cell contained in the occupation grid. Here, the height of an arbitrary cell is

, And the width of any cell is defined as

. Occupying grid forming unit 250 determines that the vertical axis resolution of Lidar sensors 210 and 212 is θ _ver , The height and the width of an arbitrary cell can be calculated using [Equation (5)].

, 임의의 셀의 너비:

, 임의의 샘플링 보드(rb_i)와 제1 라이다 센서(210) 간의 거리:

, 제1 라이다 센서(210)의 수직축 해상도: θ_ver)(Height of arbitrary cell:

, The width of any cell:

, The distance between any sampling board (rb _i ) and first lidar sensor 210:

, The vertical axis resolution of the first RI sensor 210: [theta] _ver )

), 제1 라이다 센서(210)의 수직축 해상도(θ_ver) 등에 따라 점유 그리드에 포함된 셀의 높이 및 너비가 변화되므로, 점유 그리드 형성부(250)는 각각의 샘플링 보드(220)마다 서로 다른 점유 그리드를 형성할 수 있다. 한편, 점유 그리드 형성부(250)는 사용자의 입력신호에 근거하여 각각의 샘플링 보드(220)에 형성된 점유 그리드의 셀 크기를 조정할 수 있다. The distance between the first RLayer sensor 210 and the sampling board 220

The occupancy grid forming unit 250 may calculate the height of each of the cells included in the occupied grid and the width of the occupied grid according to the vertical axis resolution θ _ver of the first Rather sensor 210, Other occupancy grids can be formed. The occupancy grid forming unit 250 may adjust the cell size of the occupancy grid formed on each sampling board 220 based on the input signal of the user.

When the distance information of the second LR sensor 212 is projected between the channels of the first LR sensor 210, the occupancy grid forming unit 250 generates a projection of the first LR sensor 210 And the position of the projection point of the second RLayer sensor 212 to determine whether the occupied grid is occupied. That is, the occupancy grid forming unit 250 checks whether the first RL sensor 210 and the second RL sensor 212 occupy different grids.

The occupancy grid forming unit 250 determines that the cell is occupied if the distance information of at least one of the first RL sensor 210 and the second RL sensor 212 occupies the cell do. Occupying grid forming unit 250 calculates the occupancy state value of the cell according to the occupation state of the cell using Equation (6).

The optimization processing unit 260 estimates the placement position of the Lidar sensors 210 and 212 by applying an optimization algorithm to improve the accuracy of the Lidar sensors 210 and 212. Here, the optimization processing unit 260 can optimize the placement position of the Lidar sensors 210 and 212 using various optimization algorithms such as a genetic algorithm, a quenching technique, and the like.

The optimization processing unit 260 can fix the placement position of the first RL sensor 210 as a reference and estimate the optimal placement position by moving the second RL sensor 212 through the optimization process. Here, the second RLayer sensor 212 may move to a unit distance using the occupancy grid.

The optimization processing unit 260 can set a constraint condition for the second LR sensor 212. [ When the constraint condition for the second Lada sensor 212 is set, the amount of movement of the second Lada sensor 212 may be limited according to the constraint condition.

The optimization processing unit 260 according to the present embodiment includes a preprocessing unit 262, a weight setting unit 264, and a placement position estimating unit 266. An operation for optimizing the placement positions of the Lidar sensors 210 and 212 using the genetic algorithm in the optimization processing unit 260 according to the present embodiment will be described. Other algorithms or techniques may be applied for optimization in other embodiments.

The preprocessing unit 262 removes the first RL sensor 210. The first RL sensor 262 detects a chromosome based on the position information and the attitude of the sensor (the second RL sensor 212) We define populations including a plurality of chromosomes, and calculate an occupancy evaluation function value for the occupancy of the occupancy grid through an optimization algorithm.

The preprocessing unit 262 applies a genetic algorithm to improve the accuracy of the Ridar sensors 210 and 212. Here, Genetic Algorithm is a probabilistic search method for solving the optimization problem, and is a computational model based on the evolutionary process of the organism. The genetic algorithm defines a population consisting of multiple chromosomes and Population is an evolutionary process such as Reproduction, Selection, Mutation and Crossover. ).

In the genetic algorithm, the probability that a population with a selected chromosome will pass on its own chromosome to a next generation will be increased by comparing preset conditions among a plurality of populations (Population) . Therefore, the next generation of genetic algorithms is closer to the optimal solution.

The preprocessing unit 262 according to the present embodiment can perform the following optimization process.

Part 1: the size of the populations (Population _size) Pop size _', crossing probability (crossover probability) P _c, mutation probability (mutation probability) P _m, up to three (maximum generation) initializes the parameters (parameters), such as Gene _max ≪ / RTI > The index of each household is set to 1.

Step 2: Initialize the chromosomes by generating Pop _{size 's} .

Step 3: Calculate evaluation function values for all chromosomes

Step 4: Select the chromosomes using the Remainder selection method (Reproduction or selection).

Step 5: Generate chromosomes that will form the next generation through crossover and mutation.

Step 6: If the generation index and the maximum generation Gene _max are the same, the optimal value is returned after genetic algorithm termination. On the other hand, if the generation index and the maximum generation Gene _max are not equal, index = index + 1 is set, and then the procedure moves to process 3 and repeats.

The preprocessor 262 according to the present embodiment is configured to determine the position information of the Lada sensors 210 and 212 and the attitude information including the Roll information and pitch information of the Lada sensors 210 and 212 To form a chromosome. Here, the chromosome is the first RL except for the sensor 210. The location information and the attitude information about the sensor (meaning the second RL sensor 212 in this embodiment) ), And the configuration of the chromosome can be represented as shown in FIG. 6 (b).

In the chromosome configuration, ch means chromosome and ch _{a and b} means b-th Chromosome of a-th Generation. Also, X _c is a bit constituting ch _{a, b} , which means a binary value.

For the generation of chromosomes, a calculation formula for the number of bits constituting the variables x, y, z,? And? Is as shown in Equation (7). Here, the constraint condition for each variable is expressed by Equation (8).

Here, x _bit, y _bit, z _bit, φ _bit and θ _bit indicates the number of bit required to express the variables to the chromosome _{(Chromosome), x n, y} n, z n, φ n and θ _n are each Indicates the number of decimal places allowed in the variable.

Also, x _lower , y _lower , z _lower , φ _lower , θ _lower and x _higher , y _higher , z _higher , _higher and _higher are constraints on the lower and upper limits of the variable, And limitations on the change of the position information and the attitude of the second RLayer sensor 212 can be set.

The preprocessor 262 generates a chromosome including the position information and the attitude of the second RL sensor 212 and generates a chromosome including chromosomes constituting a population of each generation, (Evaluation Function Value) for the occupancy of the occupied grid according to the Chromosome.

Weight setting unit 264 may in calculating the occupancy weighting (w _rbi) for sampling the board _(i rb). That is, the weight setting unit 264 assigns different weights to the sampling boards 220 to determine the occupancy. The weight setting unit 264 can derive optimized results with specific gravity on the region of interest or the direction of interest of the user by assigning different weights to the sampling boards 220. Here, the evaluation function value of the weighted chromosome ch _{a, b} is calculated using Equation (9).

_Rbi where w refers to the weight for the sampling board 220, and is rb _{i, j} refers to a cell (Cell) of the sampling board 220. An example of this is shown in Figs. 7 (a) and 7 (b). (A) a plurality of rb _1, rb _2, sampling board _{(220) rb 3, rb 4} , rb 5, rb 6, rb 7, shows a layout view of rb _8, (b) of Fig. 7 in Fig. 7 And represents the distance information projected on the plurality of sampling boards 220.

As shown in FIG. 7 (b), the red and blue dots are different from each other by the first Lada sensor 210 and the second Lada sensor 212 on the sampling board 220 Projected distance information. In addition, the difference between the sizes of the boards rb ₁ , rb ₈ , and rb ₂ and the cell size occurs according to the distance from the first rider sensor 210.

The placement position estimator 266 estimates the optimized placement position of the Ridas sensors 210 and 212 based on the optimization algorithm. The placement position estimating unit 266 detects a chromosome having the maximum occupancy based on the occupancy evaluation function value and detects the placement position of the row sensors 210 and 212 using the detected chromosome .

According to the optimization algorithm, the chromosome evolves through several generations. Therefore, the optimum result varies with the generation. The objective function for finding the optimal result, that is, the optimized placement position Fitness _obj is expressed by Equation (10).

ch _{a, b} means the b-th Chromosome of the a-th Generation and eval (ch _{a, b} ) means the value of the evaluation function for the chromosome ch _a, to be. That is, a chromosome that maximizes the occupancy is searched and the placement position of the Lidar sensors 210 and 212 is estimated from the chromosome that is searched.

The placement position determination unit 270 determines the final placement position of the Ridas sensor 210, 212 based on the estimated placement position.

The placement position determination unit 270 further checks the occupancy of the predetermined peripheral region based on the estimated placement position, and determines the placement position determined to have the highest occupancy as the final placement position.

The placement position determination unit 270 can determine the placement position of the Lidar sensors 210 and 212 using Local Search based on the estimated placement position. Here, it is preferable to use the hill-climbing algorithm for the local search algorithm, but it is not limited thereto. Simulated Annealing, Local Beam Search, and Genetic Algorithms may be used.

3 is a flowchart for explaining a method of determining the placement position of the 3D Raidas sensor according to the present embodiment.

The placement position determination apparatus 230 sets information on the plurality of sampling boards 220 (S310). The placement position determination device 230 sets target information on a plurality of sampling boards 220 disposed around the first row sensor 210 among a plurality of row sensors.

The arrangement position determining apparatus 230 may determine at least one of the distance between the sampling board 220 and the first Lidar sensor 210, the number of channels of the first Lidar sensor 210, and the vertical resolution of the first Lidar sensor 210 Target information is set by calculating the height and width of the sampling board 220 using one piece of condition information.

The positioning unit 230 forms an occupation grid on the sampling board 220 (S320). The placement position determination device 230 forms a plurality of cells on the sampling board 220 based on the target information and calculates a height and a width of each of the plurality of cells to form an occupation grid on the sampling board 220 . The placement position determining device 230 determines the height of the cell for forming the occupation grid based on the distance between the first RL sensor 210 and the sampling board 220 and the vertical resolution of the first RL sensor 210 And width.

The placement position determination apparatus 230 calculates an occupancy of the RI sensors 210 and 212 by applying an optimization algorithm (S330) (S340). The placement position determination device 230 estimates the placement position of the Lidar sensors 210 and 212 by applying an optimization algorithm to improve the accuracy of the Lidar sensors 210 and 212. [ Here, the positioning apparatus 230 can optimize the placement of the Lidar sensors 210 and 212 using various optimization algorithms such as a Genetic Algorithm, a Simulated Annealing, and the like.

The placement position determination device 230 can fix the placement position of the first RL sensor 210 as a reference and estimate the optimal placement position by moving the second RL sensor 212 through the optimization process. The placement position determination device 230 determines the location of the chromosome 210 based on the location information and the attitude of the sensor 210 (second RID sensor 212) ), Populations including a plurality of chromosomes are defined, and an occupancy evaluation function value for the occupancy of the occupancy grid is calculated through an optimization algorithm.

When the arrangement condition determining device 230 satisfies the preset termination condition (S350), the arrangement position determining device 230 determines the arrangement position of the Lidar sensors 210 and 212 (S360). Here, the termination condition refers to a condition for further checking the occupancy for the predetermined peripheral region based on the estimated placement position, and determining whether the occupancy is the highest. That is, the placement position determination device 230 determines the arrangement position of the sensors 210 and 212 when there is no placement position having a higher degree of occupancy in the peripheral area.

3, it is described that steps S310 to S360 are sequentially executed. However, this is merely an exemplary description of the technical idea of the present embodiment, and it will be understood by those skilled in the art that, It will be understood that various changes and modifications may be made to the invention without departing from the essential characteristics thereof, or alternatively, by executing one or more of the steps S310 through S360 in parallel, But is not limited thereto.

As described above, the operation of the positioning device 230 according to the present embodiment described in FIG. 3 can be implemented by a program and recorded on a computer-readable recording medium. A program for implementing the operation of the positioning device 230 according to the present embodiment is recorded, and the computer-readable recording medium includes all kinds of recording devices for storing data that can be read by the computer system. Examples of such computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage, etc., and also implemented in the form of a carrier wave (e.g., transmission over the Internet) . The computer readable recording medium may also be distributed over a networked computer system so that computer readable code is stored and executed in a distributed manner. In addition, functional programs, codes, and code segments for implementing the present embodiment can be easily inferred by programmers in the technical field to which the present embodiment belongs.

4A and 4B are diagrams for explaining a procedure of setting a sampling board according to the present embodiment.

4 (b) and 4 (b) show a sampling board 220 disposed around the Lidar sensors 210 and 212, respectively. 4B shows a plurality of sampling boards 220, and each of the sampling boards 220 may have different sizes.

FIG. 4C shows that the channel spacing of the vertical axis varies depending on the distance between the RIDAR sensors 210 and 212 and the sampling board 220. FIG. 4B, when the distance between the first RL sensor 210 and the sampling board 220 is shortened, the channel spacing of the vertical axis of the first RL sensor 210 is narrow Loses. On the other hand, if the distance between the second RL sensor 212 and the sampling board 220 is increased, the channel spacing of the second RL sensor 212 on the vertical axis is widened.

5A and 5B are diagrams for explaining the process of forming the occupation grid according to the present embodiment.

As shown in FIG. 5A, the sampling board 220 may be a rectangular board. In addition, the sampling board 220 may be implemented as a solid color. The size of the sampling board 220 can be changed in consideration of the vertical axis resolution, the distance to the first Lada sensor 210, and the like.

) 및 너비(

)는 수직축 해상도, 제1 라이다 센서(210)와 샘플링 보드(220) 간의 거리 등을 고려하여 결정된다. The height of the sampling board 220

) And width

Is determined in consideration of the vertical axis resolution, the distance between the first Lada sensor 210 and the sampling board 220, and the like.

) 및 너비(

)는 제1 라이다 센서(210)와 샘플링 보드(220) 간의 거리 및 제1 라이다 센서(210)의 수직축 해상도 등을 고려하여 결정된다. The height of the cell for forming the occupancy grid on the sampling board 220 (

) And width

Is determined in consideration of the distance between the first RL sensor 210 and the sampling board 220 and the vertical axis resolution of the first RLayer sensor 210. [

FIG. 5B shows a sampling board 220 that generates an occupancy grid of 0.5 times the channel spacing of the Lidar sensors 210 and 212.

5B shows a first RL sensor 210 and a second RL sensor 212. The sensor 210 is a first RL sensor and the second RL sensor 212 is a first RL sensor, Are simultaneously projected.

As shown in FIG. 5B, when the placement position determination apparatus 230 has projected the distance information of the second LR sensor 212 between the channels of the first LR sensor 210 , It is determined whether or not the occupation grid is occupied based on the projection point of the first Rather sensor 210 and the projection point of the second Rather sensor 212. That is, the positioning device 230 can check whether the first RL sensor 210 and the second RL sensor 212 occupy different grids.

When the positioning information on the at least one of the first RL sensor 210 and the second RL sensor 212 occupies a cell, the arrangement position determining device 230 determines that the cell is in an occupied state .

6 is an exemplary diagram for explaining an optimization process according to the present embodiment.

6 (a) shows the arrangement positions of the first RL sensor 210 and the second RL sensor 212. In FIG. As shown in FIG. 6A, the first RL sensor 210 is a RL sensor fixed at a predetermined position as a reference sensor, and the second RL sensor 212 performs placement optimization (Adjusts) the position based on the position of the first RL sensor 210. [

FIG. 6 (b) shows a configuration of a chromosome for a genetic algorithm. As shown in FIG. 6 (b), ch denotes chromosome in the chromosome structure, ch _{a and b} are the b-th Chromosome in the a-th Generation, . Also, X _c is a bit constituting ch _{a, b} , which means a binary value.

FIG. 8 is an exemplary diagram illustrating a sampling board of an optimized 3D RAY sensor according to the present embodiment. 8 (a) shows a sampling board 220 located in front of an autonomous traveling vehicle, and FIG. 8 (b) shows a sampling board 220 located behind an autonomous traveling vehicle.

The foregoing description is merely illustrative of the technical idea of the present embodiment, and various modifications and changes may be made to those skilled in the art without departing from the essential characteristics of the embodiments. Therefore, the present embodiments are to be construed as illustrative rather than restrictive, and the scope of the technical idea of the present embodiment is not limited by these embodiments. The scope of protection of the present embodiment should be construed according to the following claims, and all technical ideas within the scope of equivalents thereof should be construed as being included in the scope of the present invention.

110: Conventional single row sensor
200: 3D Lidar sensor system 210: First Lidar sensor
212: second Lada sensor 220: sampling board
230: Positioning device for positioning 240: Sampling board setting part
250: Occupied grid forming unit 260: Optimization processing unit
262: preprocessing unit 264: weight setting unit
266: placement position estimation unit 270: placement position determination unit

Claims (18)

An apparatus for determining an arrangement position of a plurality of Lidar sensors,
A sampling board setting unit for setting target information about a sampling board disposed in the periphery of the first row sensor of the plurality of row sensors;
An occupied grid forming unit for forming an occupied grid on the sampling board based on the target information;
An optimization processing unit for fixing an arrangement position of the first Rather sensor based on a position of the first Rather sensor and adjusting an arrangement position of the Rather than the first Rather sensor among the plurality of Rather sensors to estimate an optimized placement position; And
And a placement position determining unit for determining a final placement position of the plurality of row sensors based on the estimated placement position,
And a position sensor for detecting the position of the lidar sensor. The method according to claim 1,
Wherein the sampling board setting unit sets,
Calculating a height and a width of the sampling board using at least one condition information among a distance between the sampling board and the first Rather sensor, a number of channels of the first Rather sensor, and a vertical axis resolution of the first Rather sensor And sets the target information based on the position information. 3. The method of claim 2,
Wherein the sampling board setting unit sets,
The height of the sampling board is less than the result of multiplying the number of channels of the first RLayer sensor, the distance between the sampling board and the first RLayer sensor, and the tangent of the vertical axis resolution of the first RLayer sensor, Width of the lidar sensor. The method of claim 3,
Wherein the sampling board setting unit sets,
The number of channels obtained by subtracting a predetermined number of channels from the number of channels of the first RL sensor, the distance between the sampling board and the first RL sensor, and the tangent of the vertical resolution of the first RL sensor, And the height of the sampling board is set so that the height of the sampling board is greater than the height of the sampling board. The method according to claim 1,
Wherein the occupied grid forming portion comprises:
Wherein the plurality of cells are configured on the sampling board based on the target information, and the occupancy grid is formed by calculating a height and a width of each of the plurality of cells. 6. The method of claim 5,
Wherein the occupied grid forming portion comprises:
And calculating a height and a width of the cell for forming the occupation grid based on the distance between the first Rather sensor and the sampling board and the vertical resolution of the first Rather sensor. Crystal device. The method according to claim 1,
Wherein the optimization processing unit includes:
A chromosome is constructed based on the position and attitude of the remaining RL sensor and a population including a plurality of chromosomes is defined and the occupancy And calculating an occupancy evaluation function value for the occupancy of the grid. 8. The method of claim 7,
Wherein the optimization processing unit includes:
Wherein a chromosome having a maximum occupancy rate is detected based on the occupancy evaluation function value and an arrangement position of the plurality of the RDA sensors is estimated using a detected chromosome, And a positioning device. 8. The method of claim 7,
Wherein the optimization processing unit includes:
And assigning different weights to the sampling boards to calculate the occupancy evaluation function value. 10. The method of claim 9,
Wherein the optimization processing unit includes:
Wherein the weighting unit assigns a weight higher than a predetermined reference value to the sampling board located in a region of interest or a direction of interest of the user. The method according to claim 1,
The placement position determination unit determines,
Further determining a placement position of the predetermined peripheral region based on the estimated placement position and determining a placement position determined to have the highest occupancy as the final placement position, . A method for determining a placement position of a plurality of Lidar sensors by a placement position determination apparatus,
A sampling board setting step of setting target information on a sampling board disposed in the vicinity of a first row sensor of the plurality of row sensors;
Forming an occupancy grid on the sampling board based on the target information;
An optimization process for estimating an optimized placement position by adjusting placement positions of the first Rather sensors except for the first Rather sensor among the plurality of Rather sensors, ; And
A placement position determination process of determining placement positions of the plurality of row sensors based on the estimated placement positions
Wherein the positioning of the lidar sensor comprises: 13. The method of claim 12,
The sampling board setting process includes:
Calculating a height and a width of the sampling board using at least one condition information among a distance between the sampling board and the first Rather sensor, a number of channels of the first Rather sensor, and a vertical axis resolution of the first Rather sensor And setting the target information based on the position information. 13. The method of claim 12,
The occupancy grid forming process includes:
Wherein a plurality of cells are configured on the sampling board based on the target information, and a height and a width of each cell are calculated to form the occupancy grid. 15. The method of claim 14,
The occupancy grid forming process includes:
And calculating a height and a width of the cell for forming the occupation grid based on the distance between the first Rather sensor and the sampling board and the vertical resolution of the first Rather sensor. Determination method. 13. The method of claim 12,
The optimization process includes:
A chromosome is constructed based on the position and attitude of the remaining RL sensor and a population including a plurality of chromosomes is defined and the occupancy And calculating an occupancy evaluation function value for the occupancy of the grid. 17. The method of claim 16,
The optimization process includes:
Wherein a chromosome having a maximum occupancy rate is detected based on the occupancy evaluation function value and an arrangement position of the plurality of the RDA sensors is estimated using a detected chromosome, Wherein said positioning means comprises: 13. The method of claim 12,
The positioning process includes:
Further comprising the step of: determining a position of the predetermined peripheral region on the basis of the estimated position, and determining a placement position of the occupant to be the highest position, Way.

Images