KR20160071236A - Mini Integrated-control device - Google Patents

Abstract

According to a preferable embodiment of the present invention, a small integrated control device comprises: a main processor for processing bulk sensor data by using a multi-core CPU; an auxiliary processor for processing the bulk sensor data in parallel by using the same clock as the main processor; and a graphic processing unit for using the same clock as the main processor and processing in parallel a calculation which occurs when the bulk sensor data is processed.

Description

[0002] Mini Integrated-control device [0003]

The present invention relates to a small-sized integrated controller that receives a large-capacity sensor data and performs parallel processing to obtain quick results.

Figure 1 shows an example of a conventional controller. In the conventional controller system, as shown in FIG. 1, a large number of CPU cores are required in order to process large capacity sensor data. In order to secure large capacity sensor data, a plurality of PCs 110, 111, 112, 113, 114 and 115 are connected to a gigabit switch (120).

In addition to a large number of PCs, external sensors and additional parts for controlling the system are connected to individual products, so that the controller system has considerable size and volume.

"Towards Fully Autonomous Driving: Systems and Algorithms", 2011 IEEE Intelligent Vehicles Symposium (IV) Baden-Baden, Germany, June 5-9, 2011

When the large capacity sensor data 100 shown in FIG. 1 is input to the PC1 110, the PC1 110 is connected to another PC such as PC2, PC3, PC4, PC5 and PC6 111, 112, 113, 114 and 115 And transmitted the large amount of sensor data received and shared. However, in the case of transmitting and sharing data, the Gigabit Ethernet switch is theoretically 1 Gb / s, but actually has a transmission speed of 50 MB / s, so that it is not possible to efficiently share a large amount of data among PCs.

In order to solve such a problem, the PC 1 110 selects and shares only a few important frames among the received large-capacity sensor data, so that the PCs 111, 112, 113, 114 and 115 other than the PC 1 110 There is a problem in that an accurate calculation result value can not be obtained because the calculation is performed based on the sensor data whose resolution is degraded.

In one preferred embodiment of the present invention, the small integrated controller includes a main processor for processing large-capacity sensor data using a multi-core CPU, a secondary processor for processing the large-capacity sensor data in parallel using the same clock as the main processor, And a graphic processing unit which uses the same clock as the main processor and processes the operations occurring in the large capacity sensor data processing in parallel.

Preferably, the graphics processing unit receives the three-dimensional distance data among the large-capacity sensor data, converts the three-dimensional distance data into voxel data, and stores the three-dimensional point cloud data included in each voxel An arithmetic unit for calculating an average and a covariance; Calculating the surface direction angle and height of each voxel based on the eigen value and the eigenvector value calculated on the basis of the average and covariance values, and calculating a traversability, which is a probability value for whether to pass through each voxel A coefficient operation unit; And a frequency detector for digitizing the number of times the respective voxels have been scanned (Occupancy).

Preferably, the graphics processor transmits the pass coefficient for each voxel and the number of times of scanning each voxel to the main processor, and the main processor calculates a pass coefficient for each voxel and a value of the number of times each voxel is scanned, To generate a map.

Preferably, the moving object is returned by using the traveling route image photographed by the image photographing sensor mounted on the moving object and the map generated by the main processor.

As a preferred embodiment of the present invention, the main processor includes a layered stack including a system layer, an interface layer, a core layer, and an application layer. .

As another preferred embodiment of the present invention, a method for processing large-capacity sensor data in a small integrated control apparatus includes processing large-capacity sensor data using at least one or more multicore CPUs in a main processor; Processing the sensor data related to the environment recognition among the mass sensor data using the same clock as the main processor in the coprocessor and outputting the large capacity sensor data using the same clock as the main processor using the multi- Wherein the main processor, the coprocessor, and the graphics processor process the large-capacity sensor data in parallel.

As another preferred embodiment of the present invention, a method for processing large-capacity sensor data in a small integrated control apparatus includes processing large-capacity sensor data using at least one or more multicore CPUs in a main processor; Processing the laser-based sensor data among the large-capacity sensor data in a coprocessor using the same clock as that of the main processor, and in a graphics processing unit using the same clock as the main processor, Wherein the main processor, the coprocessor, and the graphics processor process the large-capacity sensor data in parallel, and the graphic processor processes the three-dimensional distance data among the large-capacity sensor data into a rectangle , Calculates a probability value that can pass through each voxel based on the average and covariance values of three-dimensional point cloud data in each voxel, The accuracy of the probability values that can pass through each voxel Characterized in that for calculating the weights.

As a preferred embodiment of the present invention, the compact integrated control device has an effect that it can be utilized in artificial intelligence, military equipment, factory automation, mobile server equipment, and autonomous mobile robot.

1 shows an example of a general controller.
2 shows an example of a small integrated device 200, which is a preferred embodiment of the present invention.
3 to 4 show an internal configuration diagram of a small integrated control apparatus 300 as a preferred embodiment of the present invention.
FIGS. 5 to 7 illustrate an embodiment in which an operation process is performed in a graphics processing unit according to a preferred embodiment of the present invention.
FIG. 8 illustrates a layered stack 800 supported by a main processor 310 according to a preferred embodiment of the present invention.
Fig. 9 shows an embodiment of a moving object equipped with a small integrated control device or using a small integrated control device, according to a preferred embodiment of the present invention.
Fig. 10 shows a flowchart for processing large capacity sensor data in parallel in a graphics processing unit of a small integrated control apparatus, in a preferred embodiment of the present invention.
11 shows a flow chart for performing arithmetic processing in a graphics processing unit of a small integrated control apparatus in a preferred embodiment of the present invention.

The terminology used herein is for the purpose of illustrating embodiments and is not intended to be limiting of the present invention. In the present specification, the singular form includes plural forms unless otherwise specified in the specification. It is noted that the terms "comprises" and / or "comprising" used in the specification are intended to be inclusive in a manner similar to the components, steps, operations, and / Or additions.

Unless defined otherwise, all terms (including technical and scientific terms) used herein may be used in a sense commonly understood by one of ordinary skill in the art to which this invention belongs. Also, commonly used predefined terms are not ideally or excessively interpreted unless explicitly defined otherwise.

Embodiments of the present invention are not limited to server computer systems, desktop computer systems, laptops, handheld devices, smartphones, tablets, other thin notebooks, systems on a chip (SOC) Devices, and other applications such as embedded applications. Handheld devices include cellular phones, Internet protocol devices, digital cameras, PDAs (personal digital assistants), and handheld PCs. In addition, the devices, methods, and systems described herein are not limited to physical computing devices, and may also be utilized for software optimization for energy savings and efficiency.

2 shows an example of a small integrated device 200, which is a preferred embodiment of the present invention.

The small integrated device 200 includes a main processor 210 for processing large-capacity sensor data using at least one multicore CPU, a secondary processor 220 for processing large-capacity sensor data using the same clock as the main processor 210, And a graphics processing unit 230 that uses the same clock as the main processor and performs arithmetic processing of the large-capacity sensor data. As a preferred embodiment of the present invention, the main processor 210, the coprocessor 220, and the graphic processor 230 are capable of processing large-capacity sensor data in parallel.

The small integrated device 200 can be implemented to expand the main processor 210, the coprocessor 220 and the graphic processing unit 230 in series or in parallel using the input / output interface 240.

In one preferred embodiment of the present invention, the main processor 210 may be any type of data processor, including a general purpose or special purpose central processing unit (CPU), an application specific integrated circuit (ASIC) or a digital signal processor .

For example, the main processor 210 may be a general purpose processor such as a CoreTM i3, i5, i7, 2 Duo and Quad, XeonTM, or ItaniumTM processor. The main processor 210 may be a processor, such as a network or communications processor, a compression engine, a graphics processor, a co-processor, a built-in processor, etc., designed for a special purpose. The main processor 210 may be implemented with one or more chips included in one or more packages. The main processor 210 may utilize any of a number of process technologies, such as, for example, BiCMOS, CMOS, or NMOS.

In a preferred embodiment of the present invention, the main processor 210 may include at least one multicore CPU 212, 214. At least one of the multicore CPUs 212 and 214 can communicate using a QPI (Quick Path Interconnect) protocol. The main processor 210 may interconnect the at least one multicore CPU 210, 212 using a packet-based point-to-point interconnect bus.

In one preferred embodiment of the present invention, the graphics processing unit 230 includes logic for executing graphics commands, such as 3D or 2D graphics commands. Graphics processing unit 230 may execute industry standard graphics commands such as those specified by Open GL and / or Direct X application programming interfaces (APIs) (e.g., OpenGL 4.1 and DirectX 11).

The input / output interface 240 can support network communication such as a local area network, a wide area network, or the Internet, and communication with an internal unit or an external device. The input / output interface 240 may further include an adapter, a hub, and the like to provide access and network communication with an internal unit or an external device. The input / output interface 240 may be implemented on the same chip or the same board as the main processor 210 or in a separate chip and / or package connected to the main processor 210.

3 to 4 show an internal configuration diagram of a small integrated control apparatus 300 as a preferred embodiment of the present invention.

The small integrated control device 300 includes a main processor 310, a coprocessor 320, and a graphics processing unit 330.

Preferably, the small integrated control device 300 may further include an input / output interface 340 or an Ethernet switch 350.

The small integrated control device 300 may further include a main processor 310, a coprocessor 320 and a power supply unit (not shown) for supplying power to the graphic processing unit 330. [

The micro integrated controller 300 may further include an MCU (Micro Control Unit) (not shown) for controlling the main processor 310, the coprocessor 320, and the graphics processor 330. In this case, the MCU can perform communication such as Ethernet or RS232 with the main processor 310, and is connected to the power unit through CAN communication, so that the MCU can be implemented to control the follow-up measures of the peripheral devices for the power state and the fault situation have.

In one preferred embodiment of the present invention, the small integrated control device 300 can process large-capacity sensor data in parallel using the main processor 310, the coprocessor 320, and the graphics processor 330 using the same system clock .

As one preferred embodiment of the present invention, it is assumed that the small integrated control device 300 is implemented as an autonomous mobile robot, a mobile robot, a moving object, or the like.

In this case, the main processor 310 selects the traveling path by itself from the autonomous mobile robot, the mobile robot, and the moving object, and calculates the position data of the autonomous mobile robot, the mobile robot, And an automobile map or a map indicating a movement route for autonomous travel based on data of an obstacle in a position close to an autonomous mobile robot, a mobile robot, or a mobile object.

In this case, the coprocessor 320 processes an operation related to environment recognition among the large-capacity sensor data received by the main processor 310, and the graphic processor 330 performs parallel processing to process image-related operations among the large- Can be performed.

Operations associated with the perception of the environment performed by the coprocessor 320 include operations to process laser-based sensor data. The laser-based sensor data includes sensor data detected by a laser scanner or the like.

An operation related to the image performed by the graphic processing unit 330 includes a camera image, an arithmetic process, and the like. Also, the computation processing derived from the laser-based sensor data can be processed in the graphic processing unit 330.

In a preferred embodiment of the present invention, the main processor 310 may include LADAR sensor information indicating position data, distance data, data on obstacles in close proximity, etc. in an autonomous mobile robot, a mobile robot, Information, CAMERA image information, and the like. In this case, the main processor 310 may be configured to receive the large-capacity sensor data using the Ethernet switch 350 or the like.

The main processor 310 transmits the data to the graphic processor 330 to process the data of the distance data and the obstacles in the proximate positions of the large capacity sensor data. And generate an on-road map or map based thereon. One embodiment of processing an operation in the graphics processing unit 330 is illustrated in Figures 5-6.

FIG. 8 illustrates a layered stack 800 supported by a main processor 310 according to a preferred embodiment of the present invention.

The layered stack 800 includes a system layer 810, an interface layer 820, a core layer 830, and an application layer 840.

The Interface Layer 820 supports MCUs, sensors, and communication interfaces.

The core layer 830 is used for locating the dynamic environment in real time, locating the visual distance, locating such as laser scan matching, dynamic obstacle detection and tracking, environment recognition such as laser based environment recognition, global route planning based on RRT sampling, planning, basic waypoint following and fast way following, such as costom planning and dynamic obstacle avoidance, and other math and utility libraries.

The coprocessor 320 supports environmental recognition such as dynamic obstacle detection and tracking and a laser-based environment recognition among the core layer 830, for example. It can also be implemented to handle parts that can optimize parallel processing. The planning SW can also be divided into individual grids and processed by the auxiliary processor 320 if parallelization is possible.

FIGS. 5 to 6 illustrate an embodiment for performing arithmetic processing in the graphics processing unit according to a preferred embodiment of the present invention.

The graphics processor 500 uses the same clock as the main processor and processes the large-capacity sensor data in parallel by using the multicore.

The graphic processing unit 500 includes an operation unit 532, a pass coefficient operation unit 534, and a frequency detection unit 536.

The calculation unit 532 can receive the three-dimensional distance data, the camera image, the traveling data, and the like among the large-capacity sensor data from the main processor 510. In one preferred embodiment of the present invention, the three-dimensional distance data is formed based on at least one of three-dimensional image data of the obstacle, distance information, and movement information.

The operation unit 532 converts the three-dimensional distance data into voxel data as in the embodiment of FIG. Then, the average and covariance of the three-dimensional point cloud data contained in each voxel are calculated.

Referring to FIG. 6, in a preferred embodiment of the present invention, the operation unit 532 divides the three-dimensional distance data into voxels of a rectangular size having a predetermined size (S610, S620, S630). In this case, since each of the voxels S610, S620, and S630 is independent of each other, the calculation for each voxel can be processed quickly by a multi-core graphics processing unit. In one preferred embodiment of the present invention, the height of the voxel may be set to a maximum height that can be detected by a three-dimensional distance sensor such as a velodyne sensor.

In a preferred embodiment of the present invention, the operation unit 532 calculates a mean and a covariance of three-dimensional point cloud data included in each of the voxels S610, S620, and S630 .

The pass coefficient calculator 534 calculates the eigenvalue and the eigenvector using the mean and covariance of each voxel calculated by the calculator 532. [ In one preferred embodiment of the present invention, the height (FIG. 6, S611) of the three-dimensional point cloud data in each voxel is obtained using the Eigen value, and the height The surface direction angle (FIG. 6, S613) of the point cloud data is obtained.

The pass coefficient calculator 534 calculates the height and the surface direction angle of the three-dimensional point cloud data in each voxel using the eigen value and the eigenvector value, Calculate the probability value. The surface orientation angle can be calculated via surface normal. In the present invention, this probability value is referred to as a traversability. The pass coefficient can be calculated as shown in Equation (1).

In Equation (1), N (x) is a function for calculating surface normal, and x represents an index of each voxel.

The frequency detector 536 digitizes the number of times each voxel is scanned (Occupancy). It should be noted that it is not limited to being able to perform the number of times each voxel is scanned only in the graphic processor 530, but can also be performed in the main processor 510.

As a preferred embodiment of the present invention, a laser sensor mounted on a mobile robot, a moving object, an autonomous robot or the like is higher in the number of times that an object at a close distance is scanned than the number of times that an object at a distance is scanned. In one preferred embodiment of the present invention, the frequency detector 536 is located nearer to the preset number of occasions (Occupancy) of each voxel than the predetermined number of times the voxel has been scanned using a laser scanner, a laser sensor, And so on. In addition, it is possible to reduce the error in sensor data and errors in position recognition in mobile robots, mobile objects, autonomous robots, and the like.

In one preferred embodiment of the present invention, the graphics processor 530 transmits Occupancy information of each voxel and the pass coefficient of each pixel to the main processor 510.

When the main processor 510 receives Occupancy information of each voxel and information on the pass coefficient of each pixel, the main processor 510 accumulates Occupancy information of each voxel and pass coefficient of each pixel, And generates a three-dimensional running map or a map through the running map generating unit 511. [ The main processor 510 can obtain the movement information and the traveling map according to time by performing histogram matching or the like on the three-dimensional distance data continuously obtained in front of the traveling route. The generated three-dimensional map can be stored in an internal chip, an internal board or a remote device.

The main processor 510 generates the driving map as shown in Equation (2).

In this case, Weight _t = Weight _{t -1} + Occupancy _t

In a preferred embodiment of the present invention, the moving object, the mobile robot, the autonomous robot, and the like can automatically return the traveling route by referring to the traveling map generated by the main processor 510. [

In a preferred embodiment of the present invention, the driving map generation is not performed by the main processor 510 alone, but is performed in parallel in parallel with the operation processing in the graphic processor 530 and the environment recognition in the coprocessor have.

FIG. 7 illustrates an example of recognizing three-dimensional distance data and calculating a pass coefficient in a graphic processor according to a preferred embodiment of the present invention.

The graphic processing unit recognizes the three-dimensional distance data (710), converts the recognized three-dimensional distance data into voxel data (720), and then converts the three-dimensional distance data into the average and covariance values The eigen value and the eigen vector value are calculated to detect the pass coefficient. The larger the value of the pass coefficient is, the darker it is, indicating that an obstacle exists. The shaded portion of the portion where the numerical value of the pass coefficient is large is displayed (711), and the lower the numerical value of the pass coefficient is, the shaded is displayed. For example, if the coefficient of passage is 0.1, the shadows are displayed softly and can represent the leaf part of a tree. When the coefficient of passage is 0.7, the shade is displayed in bold and can represent the column of a tree.

7C shows an example of setting a path by calculating a pass coefficient value in a moving object, a mobile robot, and an autonomous robot 700 using a small integrated control device according to a preferred embodiment of the present invention.

The higher the value of the pass coefficient, the lower the probability of passage. The higher the value of the pass coefficient, the higher the probability that the shade will pass, and the shade is lightly displayed.

Fig. 9 shows an embodiment of a moving object equipped with a small integrated control device or using a small integrated control device, according to a preferred embodiment of the present invention.

The moving body 900 may be implemented to include a 3D sensor unit, a sensor unit, a GPS transceiver unit, a control unit, and an output unit (internal configuration not shown). The sensor unit may include a steering sensing unit, a speed sensor, an acceleration sensor, a position sensor, and the like.

The 3D sensor unit is a camera system that captures omnidirectional, rearward, and / or side orientations at once using a rotating mirror, a condenser lens, and an image pickup device. It is applied to security facilities, surveillance cameras, robot vision, and the like. The shape of the rotating body reflector may be hyperbolic, spherical, conical, or hybrid. Also, a CCD (Charge Coupled Device) or a CMOS (Complementary Metal Oxide Semiconductor) is used as the imaging element. An image (i.e., an omnidirectional image) projected on the imaging surface of the imaging element is a reflection image that is reflected on the rotating mirror and is a distorted image that is not suitable for human observation. Therefore, in order to observe the image accurately, the 3D sensor unit converts the coordinates of the output of the image pickup device through a microprocessor or the like to produce a new panoramic image.

The 3D sensor unit includes a stereo camera, a depth camera, a moving stereo camera, and a light detector (RDA) to capture three-dimensional distance data by three-dimensionally capturing all directions. LIDAR) equipment.

A stereo camera is a video device composed of a plurality of cameras. The omnidirectional image obtained through the 3D sensor unit 110 provides two-dimensional information about the periphery of the 3D sensor unit. If a plurality of images taken from different directions through a plurality of cameras are used, three-dimensional information about the periphery of the 3D sensor unit can be obtained. Such a stereo camera is also used for location recognition and map generation of a moving object or a mobile robot.

The depth camera is a camera that shoots or measures obstacles and extracts images and distance data. That is, the depth camera creates image or image data by capturing an obstacle such as a general camera, and measures the distance from the camera at an actual position corresponding to a pixel of each image to produce distance data.

A mobile stereo camera refers to a camera in which the position of a stereo camera is actively changed according to the distance of an obstacle, thereby fixing the main observer to an observation obstacle. Stereo cameras generally position two cameras in parallel and acquire images, and calculate the distances to the obstacles according to the stereo parallax of the acquired images.

Such a stereo camera is a passive camera in which the optical axis is always arranged in parallel and fixed. On the other hand, the mobile stereo camera actively changes the geometrical position of the optical axis to fix the main view. The control of the angle of view of the stereo camera according to the distance of the obstacle is called main vision control.

Main Visual Control Stereo camera maintains stereo parallax for moving obstacles at all times to provide more natural stereoscopic images to stereoscopic viewers and provides useful information for distances to obstacles and stereo image processing.

The LIDAR equipment is provided to detect the presence and distance of the obstacles located on the front surface of the mobile unit 900. Raida is a type of active remote sensing that uses the same principle as radar to acquire the desired information without direct contact with objects. The Lada equipment uses a laser to target the target to acquire the desired distance information by sensing the parallax and energy changes of the electromagnetic waves reflected from the target.

The Rada equipment is divided into three types: DIAL (DIfferent Absorption LIDAR), Doppler LIDAR, and Range finder LIDAR, depending on the object or object to be measured. DIAL is used to measure the concentration of water vapor, ozone, and pollutants in the atmosphere using two lasers with different degrees of absorption for the object to be measured. Doppler LIDAR uses the Doppler principle to move objects It is used for speed measurement. However, in general, the term "RDA" refers to a range finder (LIDAR), which includes a Global Positioning System (GPS), an Inertial Navigation System (INS), a Laser Scanner ) System to obtain 3D terrain information by combining the distance information with the object.

The Lada equipment detects three-dimensional distance data by sensing the presence of an obstacle located on the moving path front of the moving object 900, the distance to the obstacle and the movement of the obstacle, and transmits the obtained data to the control unit, Thereby allowing the moving body 900 to move.

The output unit includes a display unit and displays a driving route determined through the driving map generated by the control unit or the driving map generating unit of the main processor (FIGS. 5 and 511) using a UI (User Interface) or a GUI (Graphic User Interface) Can be implemented.

Fig. 10 shows a flowchart for processing large capacity sensor data in parallel in a graphics processing unit of a small integrated control apparatus, in a preferred embodiment of the present invention.

 A method for processing large-capacity sensor data in a small integrated controller includes processing large-capacity sensor data using at least one multi-core CPU in a main processor (S1010) Processing sensor data related to environment recognition among the sensor data (S1020) and

(S1030) of performing operation processing among the large-capacity sensor data using the same clock as that of the main processor by using a multi-core in a graphics processing unit.

The graphic processing unit also performs operation processing on the large-capacity sensor data received from the main processor in parallel. Referring to FIG. 11, the graphic processing unit of the small integrated control apparatus receives three-dimensional distance data among the large capacity sensor data from the main processor (S1110). Thereafter, the three-dimensional distance data is converted into voxel data (S1120), and the average and covariance of the three-dimensional point cloud data included in each voxel are calculated (S1130).

Then, the graphic processing unit calculates an eigen value and an eigenvector value based on the averaged and covariance values (S1140), calculates the surface direction angle and height of each voxel based on the eigen value and the eigenvector value, .

Thereafter, the number of times the voxel is scanned is digitized, and a weight for the accuracy is given to the calculated pass coefficient (S1160).

The method of the present invention can also be embodied as computer readable code on a computer readable recording medium. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored. Examples of the computer-readable recording medium include a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like, and may be implemented in the form of a carrier wave (for example, transmission via the Internet) . The computer-readable recording medium may also be distributed over a networked computer system so that computer readable code can be stored and executed in a distributed manner.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the invention. Accordingly, the scope of protection of the present invention should be construed in accordance with the following claims, and all technical ideas within the scope of equivalents and equivalents thereof should be construed as being covered by the scope of the present invention.

Claims (40)

A main processor for processing mass sensor data using at least one multicore CPU;
An auxiliary processor for processing laser-based sensor data among the large-capacity sensor data using the same clock as the main processor;
And a graphic processor for processing image-based sensor data among the large-capacity sensor data in parallel using a multi-core, using the same clock as the main processor,
Wherein the main processor, the coprocessor, and the graphics processor process the large-capacity sensor data in parallel. The apparatus of claim 1, wherein the graphics processing unit
Dimensional distance data among the large-capacity sensor data, converts the three-dimensional distance data into voxel data, and calculates an average and a covariance of three-dimensional point cloud data included in each voxel ;
Calculating the surface direction angle and height of each voxel based on the eigen value and the eigenvector value calculated on the basis of the average and covariance values, and calculating a traversability, which is a probability value for whether to pass through each voxel A coefficient operation unit;
And a frequency detector for digitizing the number of times the individual voxels are scanned. 3. The method of claim 2,
Wherein the graphics processing unit transmits the pass coefficient for each voxel and the number of times of scanning each voxel to the main processor,
Wherein the main processor generates a map by cumulatively using a pass coefficient for each voxel and a number of times of scanning each voxel. The method of claim 3,
Wherein the moving object returns the traveled route by using the traveling route image photographed by the image photographing sensor mounted on the moving object and the map generated by the main processor. The method of claim 1, wherein the mass sensor data
And data received by at least one or more sensors mounted on the moving object. The method according to claim 1,
Wherein the graphic processing unit further processes sensor data related to image processing among the laser-based sensor data. A main processor for processing mass sensor data using at least one multicore CPU;
An auxiliary processor for processing laser-based sensor data among the large-capacity sensor data using the same clock as the main processor;
And a graphics processor for processing the large-capacity sensor data in parallel using the same clock as the main processor, wherein the main processor, the coprocessor, and the graphics processor are connected in parallel to the large- And processing the sensor data. 8. The apparatus of claim 7, wherein the graphics processing unit
The distance data and the height of each voxel are calculated on the basis of the average and covariance values of the three-dimensional point cloud data included in each voxel, And calculates a pass coefficient that is a probability value as to whether or not the signal can pass through the voxel. 8. The method of claim 7, wherein the main processor
And generates a map by cumulatively using the number of occasions of scanning each voxel and the pass coefficient calculated for each voxel. 9. The method of claim 8,
And an input / output interface for implementing the main processor, the coprocessor, and the graphics processing unit to expand in series or in parallel. A main processor for processing mass sensor data using at least one multicore CPU;
A coprocessor that uses the same clock as the main processor and processes the large capacity sensor data in parallel with the main processor;
And a graphics processor for performing a calculation process using the same clock as the main processor, wherein the main processor, the coprocessor, and the graphics processor process the large-capacity sensor data in parallel, Device. 12. The method of claim 11,
Wherein the coprocessor processes laser-based sensor data among the large-capacity sensor data, and the laser-based sensor data includes sensor data obtained through a laser scanner. 12. The method of claim 11,
Wherein the graphic processing unit processes image-based sensor data among the large-capacity sensor data in real time, and the image-based sensor data includes a camera image. 13. The system of claim 12, wherein the main processor
And controls the graphic processing unit to process sensor data operations related to image processing among the laser-based sensor data. 12. The method of claim 11,
And an input / output interface for implementing the main processor, the coprocessor, and the graphics processing unit to expand in series or in parallel. 12. The method of claim 11,
The main processor receives mass sensor data generated when the mobile robot is driven, and the mass sensor data includes sensor data detected by a sensor used by the mobile robot, real-time distance information obtained by moving the mobile robot, And image information on the basis of the image information. The apparatus of claim 11, wherein the graphics processing unit
Dimensional distance data and traveling path information detected during traveling of the mobile robot among the large-capacity sensor data, converts the three-dimensional distance data into voxel data, and outputs the three-dimensional point cloud included in each voxel and an operation unit for calculating an average value and a covariance of cloud data and extracting an eigen value and an eigenvector value,
The graphics processing unit calculates a pass coefficient indicating the possibility of passing through each voxel based on the eigen value and the eigenvector value, grasps the number of times the each voxel has been scanned, And transmits the scan count value and the scan count value to the main processor. 18. The system of claim 17, wherein the main processor
And generates a map indicating a travel route of the mobile robot by cumulatively using a pass coefficient and a scan count value for each voxel. 12. The method of claim 11,
And the main processor receives the large capacity sensor data via Ethernet. 12. The method of claim 11,
Wherein the main processor interconnects the at least one multicore CPU using a packet-based point-to-point interconnect bus. 12. The method of claim 11,
Wherein the auxiliary processor and the graphics processor synchronize the processed sensor data using the same clock as the main processor. 12. The method of claim 11,
And the main processor receives the large capacity sensor data through the Ethernet switch. 12. The method of claim 11,
Further comprising a power supply unit for supplying power to the main processor, the auxiliary processor, and the graphic processing unit. 12. The method of claim 11,
Further comprising an MCU (Micro Control Unit) for controlling the main processor, the coprocessor, and the graphics processing unit. 12. The system of claim 11, wherein the main processor
Wherein the control unit supports a system layer, an interface layer, a core layer, and an application layer. 12. The system of claim 11, wherein the main processor
A layered stack including a system layer, an interface layer, a core layer, and an application layer. 18. The method of claim 17,
Wherein the height of the voxel is set to a maximum height that can be detected by the three-dimensional distance detection sensor. 18. The method of claim 17,
Wherein the voxel is in the form of a rectangle. A main processor for processing mass sensor data using at least one multicore CPU;
An auxiliary processor for processing laser-based sensor data among the large-capacity sensor data using the same clock as the main processor;
And a graphic processor for processing image-based sensor data among the large-capacity sensor data in parallel using a multi-core, using the same clock as the main processor,
Wherein the main processor, the coprocessor, and the graphics processor process the large-capacity sensor data in parallel, and the graphics processor divides the three-dimensional distance data among the large-capacity sensor data into a rectangular voxel unit, Calculating a probability value that can pass through each voxel based on the average and covariance values of cloud data and calculating a weight value of the accuracy of the probability value that can pass through each voxel using the number of scans of each voxel And a control unit for controlling the operation of the small integrated controller. A method for processing large capacity sensor data in a small integrated control device,
Processing large-capacity sensor data using at least one or more multicore CPUs in a main processor;
Processing sensor data related to environment recognition among the mass sensor data using the same clock as the main processor in a coprocessor;
And performing a calculation process among the large-capacity sensor data using the same clock as the main processor by using a multi-core in the graphics processor
Wherein the main processor, the coprocessor, and the graphics processing unit process the large capacity sensor data in parallel. The apparatus of claim 30, wherein the graphics processing unit
The calculation unit receives three-dimensional distance data among the large-capacity sensor data, converts the three-dimensional distance data into voxel data, and calculates average and covariance of three-dimensional point cloud data included in each voxel Calculating;
A traversability calculation unit which calculates a surface direction angle and a height of each voxel based on an eigen value and an eigenvector calculated on the basis of the averaged and covariance values and calculates a traversability, ;
And a step of quantifying the number of times the voxel has been scanned by the frequency detector and giving a weight to the accuracy of the pass coefficient by using the number of times of scanning each of the digitized voxels. . 32. The method of claim 31,
Wherein the graphics processing unit transmits the pass coefficient for each voxel and the number of times of scanning each voxel to the main processor,
Wherein the main processor generates a map by cumulatively using a pass coefficient for each voxel and a number of times of scanning each voxel. 32. The method of claim 31,
Wherein the height of the rectangle is set to a maximum height that can be detected by the three-dimensional distance detection sensor. 31. The method of claim 30,
The main processor receives mass sensor data generated when the mobile robot is driven, and the mass sensor data includes sensor data detected by a sensor used by the mobile robot, real-time distance information obtained by moving the mobile robot, And image information. ≪ Desc / Clms Page number 20 > A method of processing large capacity sensor data in a small integrated control device, the method comprising:
Processing the mass sensor data using at least one or more multicore CPUs in the main processor
Processing laser-based sensor data among the large-capacity sensor data in a coprocessor using the same clock as the main processor;
And processing the image-based sensor data among the large-capacity sensor data in parallel using a multi-core in a graphics processing unit using the same clock as the main processor, wherein the main processor, the coprocessor, Large-capacity sensor data is processed in parallel,
The graphic processing unit divides the three-dimensional distance data among the large-capacity sensor data into a rectangular voxel unit, calculates a probability value that can pass through each voxel based on an average and a covariance value of three-dimensional point cloud data in each voxel And calculates a weight on the accuracy of a probability value that can pass through each of the voxels using the number of times of scanning each voxel. The system of claim 35, wherein the main processor
A system layer, an interface layer, a core layer, and an application layer. The method of claim 35, further comprising:
And expanding the main processor, the coprocessor, and the graphics processing unit in serial or parallel. 36. The method of claim 35,
Wherein the main processor interconnects the at least one multicore CPU using a packet-based point-to-point interconnect bus. 36. The method of claim 35,
Wherein the main processor receives the large capacity sensor data using an Ethernet switch. 37. The method of claim 35, wherein the mass sensor data
And data received from at least one or more sensors mounted on the moving object.

Images