Layered grid map creation method, location estimation method based on layered grid map, layered grid map-based mobile robot unit, and mobile robot system comprising same

Abstract

The present invention provides a layered grid map creation method, a location estimation method using same, a layered grid map-based mobile robot unit for implementing same, and a mobile robot system comprising same, the layered grid map creation method comprising: a characterizing step (S110) for acquiring sensing information from a mobile environment, and confirming and creating a sensing information character for a character unit cell having a preset unit size through an occupancy state of a cell in a grid map corresponding to the mobile environment; a frame matching confirmation step (S120); a key frame matching confirmation step (S130); and a layered map creation step (S140) for creating a layered map by forming a multi-layered structure.

Description

Method for creating a layered grid map, method for estimating a location based on a layered grid map, mobile robot unit based on a layered grid map, and mobile robot system having the same

[0001] The present invention relates to a map creation method, a location estimation method using the same, and a robot and system for executing the same, and more specifically, to a method for enabling rapid map creation and rapid and accurate location estimation by minimizing data transmission load, and to a mobile robot unit and system.

[0002]

[0003] Robots are increasingly being used in diverse industries and in everyday life. Research, production, and deployment are underway for a wide range of robots, from household vacuum cleaners to manufacturing robots used in welding and press processes at automobile factories, to logistics systems like AGVs and AMRs.

[0004] In particular, guided or autonomous transport logistics robots, such as AGVs (Automatic Guided Vehicles) and AMRs (Autonomous Mobile Robots), are experiencing a dramatic expansion in their scope of use and application. In other words, advancements in drive technology have led to their expansion beyond clean facilities like semiconductor factories, encompassing not only the transportation service industry for services like shopping malls and freight transport, but also various logistics and delivery industries operating fulfillment systems, and even heavy industries requiring the transport of large loads, such as heavy industry.

[0005] Meanwhile, for mobile robots, mapping the driving environment and estimating location are crucial components for accurate and rapid driving. However, as the driving environment expands, the amount of map data increases exponentially. This results in a significant computational load during map creation and location estimation, resulting in computational delays that hinder immediate, fast, and accurate map creation and location estimation.

[0006]

[0007] Accordingly, the present invention aims to provide a map creation method having a structure that enables rapid and accurate map creation and minimizes data load, a location estimation based on the method, and a mobile robot unit that enables such map creation and / or location estimation, and a system including the same.

[0008] The present invention, in order to achieve the above technical task, may have the following configuration.

[0009] According to one aspect of the present invention, the present invention provides a method for creating a layered grid map, including a characterizing step (S110) of obtaining detection information from a driving environment and confirming and generating detection information characters for character unit cells of a preset unit size through an occupancy state of cells in a grid map corresponding to the driving environment, a frame matching confirmation step (S120) of matching or individually matching detection information characters in a plurality of frames in which detection information is obtained, a keyframe matching confirmation step (S130) of registering a frame that meets a keyframe preset condition among the frames as a keyframe, setting a character word for the detection information character in the corresponding keyframe, and executing matching or individual matching between the corresponding character words among the plurality of registered keyframes, and a layering map creation step (S140) of repeatedly dividing the grid map into a preset number of sections using the individual matching information between the corresponding character words among the plurality of keyframes to form a multi-layer structure, thereby creating a layered map.

[0010] In the above layering grid map creation method, the characterization step (S110) may include: an environment detection step (S111) in which environment detection is performed from a driving environment, a sensing information acquisition step (S113) in which point cloud detection information is acquired to confirm the occupancy status of cells in a grid map corresponding to the driving environment from the environment detection information, and a sensing character confirmation step (S115) in which a sensing information character for a character unit cell of the preset unit size is confirmed and generated from the point cloud detection information.

[0011] In the layering grid map creation method, the frame matching confirmation step (S120) may include: a frame sensing character matching step (S121) of obtaining sensing information from a driving environment and matching a plurality of frames including an occupancy state of a cell in a grid map corresponding to the driving environment, and the sensing information characters in the plurality of frames; a key frame confirmation step (S123) of checking whether a frame matched in the frame sensing character matching step (S121) satisfies a pre-key frame condition; and a key frame registration step (S125) of setting the corresponding frame as a key frame when it is determined in the key frame confirmation step (S123) that the frame satisfies a photo key frame condition.

[0012] In the above layering grid map creation method, the frame alignment confirmation step (S120) may include: a alignment error confirmation step (S127) in which an alignment error is confirmed when a frame and a sensing character are aligned in the frame sensing character alignment step (S121).

[0013] In the above layering grid map creation method, if a matching error is confirmed during the matching between the frame and the sensing character in the matching error confirmation step (S127), a position correction step (S160) may be included in which the position information acquired for the corresponding frame is corrected and confirmed based on the frame sensing character matching error.

[0014] In the above layering grid map creation method, the keyframe matching verification step (S130) may include a keyframe character word verification step (S131) ​​for verifying a character word formed by one or more sensing characters within the keyframes verified and registered in the keyframe registration step (S125).

[0015] In the above layering grid map creation method, the keyframe alignment confirmation step (S130) may include a keyframe character word alignment confirmation step (S133) for confirming the alignment of character words between a plurality of keyframes confirmed and registered in the keyframe registration step (S125).

[0016] In the above layering grid map creation method, the step of checking character word alignment between key frames (S133) may include: a key frame character word selection checking step (S1331) of selecting and checking character words between the plurality of key frames, a key frame character word similarity checking step (S1333) of performing a similarity comparison between the key frame character words selected and checked in the key frame character word selection checking step (S1331), and a key frame character word alignment confirmation step (S1335) of checking whether the key frame character words are aligned and checking the location of a newly registered key frame based on the similarity confirmation result in the key frame character word similarity comparison step (S1333).

[0017] In the above layering grid map creation method, the keyframe character word similarity confirmation step (S1333) may include: a keyframe character word similarity calculation step (S13331), a keyframe character word similarity comparison step (S13333), and a keyframe character word similarity status confirmation step (S13334).

[0018]

[0019] According to another aspect of the present invention, the present invention provides a layering grid map-based position estimation method, which comprises a position estimation characterization step (S210) of confirming and generating a sensing information character for a character unit cell of a preset unit size through an occupancy state of a cell in a grid map corresponding to a driving environment to obtain sensing information from a driving environment corresponding to a layering map formed into a multi-layer structure by repeatedly dividing by a preset number of sections, that is, a layering grid map, and estimating a position, and a position estimation matching confirmation step (S220) of confirming the sensing information characters in a frame in which the sensing information has been obtained, matching or individually matching them to confirm whether they are matched, and calculating and confirming the corresponding estimated position.

[0020] In the above layering grid map-based location estimation method, the location estimation characterization step (S210) may include: a location estimation environment detection step (S211) in which environment detection is performed from a driving environment, a sensing information acquisition step (S213) in which point cloud detection information is acquired to confirm the occupancy status of cells in a grid map corresponding to the driving environment from the environment detection information, and a sensing character confirmation step (S215) in which a sensing information character for a character unit cell of the preset unit size is confirmed and generated from the point cloud detection information.

[0021] In the layering grid map-based position estimation method, the position estimation matching confirmation step (S220) may include: a detection character word confirmation step (S221) of obtaining detection information from a driving environment and confirming a detection information character word within a frame that includes an occupancy state of a cell within a grid map corresponding to the driving environment, and a layering map matching position confirmation step (S223) of comparing and matching the detection information character word with the detection information character word within a key frame for each layering area within the layering map.

[0022] In the layering grid map-based location estimation method, the layering map matching location confirmation step (S223) may include: a layering map section confirmation step (S2231) of confirming a layering map section of a current location using the layering map, a layering map character word confirmation step (S2233) of confirming a character word within the layering map section, a character word matching adjustment step (S2235) of converting and adjusting the matching of the layering map section character word and the detection information character word, a similarity confirmation step (S2236) of confirming the character word similarity of the converted layering map section character word and the detection information character word, and a similarity comparison step (S2237) of comparing the character word similarity confirmed in the similarity confirmation step (S2236) with a preset character word similarity.

[0023] In the layering grid map-based location estimation method, the layering map matching location confirmation step (S223) may include: a location confirmation step (S2238) of estimating and confirming the current location using the matching adjustment transformation data confirmed in the character word matching adjustment step (S2235) when the character word similarity is determined to be greater than the preset character word similarity in the similarity comparison step (S2237).

[0024] According to another aspect of the present invention, a layering grid map-based mobile robot unit (10) is provided, including a detection unit (11) including a lidar detection unit (111) for acquiring detection information from a driving environment, a grid map generation control module (121) for generating detection information characters for character unit cells of a preset unit size through an occupancy state of cells in a grid map corresponding to the driving environment from the detection information, a characterization control module (123) for generating detection information characters for character unit cells of a preset unit size through the occupancy state of cells in the grid map, a layer leveling control module (124) for repeatedly dividing the grid map into a preset number of sections to form a multi-layer structure so as to enable creation of a layering map, a keyframe managing module (125) for registering and managing frames that meet keyframe preset conditions among a plurality of frames acquired by the detection information as keyframes, and a matching control module (126) for individually matching detection information characters in a plurality of frames acquired by the detection information and executing individual matching between corresponding character words among the registered plurality of keyframes.

[0025] In the layering grid map-based mobile robot unit (10) above, the matching control module (126) that matches or individually matches the detection information characters within a plurality of frames acquired by the detection information and performs the matching or individual matching between the corresponding character words between a plurality of registered key frames may include a position estimation control module (127) that estimates the current position in the driving environment by using the matching results of the detection information characters within a plurality of frames acquired by the detection information and the matching between the corresponding character words between a plurality of registered key frames.

[0026] According to another aspect of the present invention, a mobile layering grid map-based mobile robot server (20) is provided, including a grid map generation control module (221) for generating a grid map corresponding to a driving environment from sensing information about the driving environment acquired from a grid map-based mobile robot unit (10) having a sensing unit (11) including a lidar sensing unit (111), a characterization control module (223) for generating a sensing information character for a character unit cell of a preset unit size through the occupancy state of the cell in the grid map, a layer leveling control module (224) for repeatedly dividing the grid map into a preset number of sections to form a multi-layer structure so as to enable creation of a layering map, a keyframe managing module (225) for registering and managing frames that meet keyframe preset conditions among a plurality of frames acquired by the sensing information as keyframes, and a matching control module (226) for matching or individually matching the sensing information characters in the plurality of frames acquired by the sensing information and executing matching or individual matching between corresponding character words among the plurality of registered keyframes. A robot system (1) is provided.

[0027] In the mobile robot system (1) having the layering grid map-based mobile robot server (20), the grid map-based mobile robot server (20) may include: a position estimation control module (227) that estimates a current position in a driving environment by using the result of the alignment of the detection information characters within the plurality of frames acquired by the detection information and the alignment between the corresponding character words between the plurality of registered key frames in the alignment control module (126) that aligns or individually aligns the detection information characters within the plurality of frames acquired by the detection information and performs the alignment or individual alignment between the corresponding character words between the plurality of registered key frames.

[0028]

[0029] The effects of the mobile robot system and mobile robot system control method of the present invention configured as described above are as follows.

[0030] First, the mobile robot system and the mobile robot system control method of the present invention can increase the stability of the transport operation by forming a virtual node and a virtual link in a direction perpendicular to or intersecting a reference path and a reference path, thereby deriving a virtual path, and by returning to the reference path when escaping from the influence area of ​​an obstacle, enabling stable driving and completing a stable transport operation even when transporting an object with a large inertia due to its own weight through frequent path changes.

[0031] Second, the mobile robot system and the mobile robot system control method of the present invention form a virtual node and a virtual link in a direction perpendicular to or intersecting a reference path and a reference path, thereby deriving a virtual path, and when it escapes the influence area of ​​an obstacle, by returning to the reference path, and when it moves away from the influence area of ​​an obstacle, it enables stable driving and completes a stable transport task even when transporting an object with a large inertia due to its own weight through frequent path changes, thereby preventing the creation of an unreasonable alternating path in the opposite direction, thereby reducing the load applied to hardware, minimizing maintenance costs, and increasing durability.

[0032]

[0033] FIG. 1 is a schematic perspective view of a layering grid map-based mobile robot unit according to one embodiment of the present invention positioned in a driving environment.

[0034] FIG. 2 is a schematic diagram of a layering grid map-based mobile robot unit and a mobile robot system including the same according to one embodiment of the present invention.

[0035] FIG. 3 is a detailed configuration diagram of a layering grid map-based mobile robot unit according to one embodiment of the present invention.

[0036] FIG. 4 and FIG. 5 are specific configuration diagrams of a keyframe managing module and a matching control module of a layering grid map-based mobile robot unit according to one embodiment of the present invention.

[0037] FIG. 6 and FIG. 7 are a schematic configuration diagram of a mobile robot server and a schematic detailed configuration diagram of a server control unit according to one embodiment of the present invention.

[0038] FIG. 8 is a state diagram showing a continuous state of a plurality of frames and key frames detected when implementing a layering grid map creation method according to one embodiment of the present invention.

[0039] FIG. 9 is a state diagram of an example of a frame in which point cloud data detected during implementation of a layering grid map creation method according to one embodiment of the present invention is displayed.

[0040] FIG. 10 is an example state diagram showing the matching connection state of the detected information characters within the frames detected when implementing a layering grid map creation method according to one embodiment of the present invention.

[0041] FIG. 11 is an exemplary configuration diagram of a character detection information within a frame detected when implementing a layering grid map creation method according to an embodiment of the present invention.

[0042] FIG. 12 is a diagram illustrating the configuration of at least a portion of a key frame of a grid map formed from sensing information about a driving environment acquired from the location of a mobile robot unit based on a layering grid map when implementing a method for creating a layering grid map according to one embodiment of the present invention.

[0043] FIG. 13 is a diagram of a configuration of a key frame character word as a sensing information character string composed of a plurality of sensing information characters within a key frame among a grid map formed from sensing information about a driving environment acquired from the location of a mobile robot unit based on a layering grid map when implementing a layering grid map creation method according to one embodiment of the present invention.

[0044] FIG. 14 is a diagram illustrating a configuration of at least a portion of a key frame of a pre-stored grid map stored in a unit storage unit when implementing a method for creating a layered grid map according to an embodiment of the present invention.

[0045] FIG. 15 is a diagram of the configuration of a keyframe character word as a detection information character string composed of a plurality of detection information characters within the stored key frame of FIG. 14.

[0046] FIG. 16 is a diagram showing region A of a key frame in a grid map formed from sensing information about a driving environment acquired from the location of a grid map-based mobile robot unit of FIGS. 12 and 14 and region B of a key frame of a pre-stored grid map stored in a unit storage unit.

[0047] Figure 17 is a comparative diagram of the detection information character words for each of Figure 16.

[0048] Figure 18 is a diagram showing a configuration in which n layers of an n-th form are shown in a multi-layered division structure for a layered grid map corresponding to a driving environment.

[0049] Figure 19 is an example configuration diagram formed with a multi-layered division structure for a very small area of ​​the driving environment.

[0050] Figures 20 to 27 are flowcharts showing the control flow of the map creation and location estimation method of the present invention.

[0051] FIG. 28 is a diagram illustrating a state in which rotation and position transformation data generated in the process of matching the detected information character word within the keyframe of the present invention with the scanned confirmed detected information character word are inversely transformed to estimate and calculate the scanned position of the detected information character word within the keyframe.

[0052] Figure 29 is a schematic diagram of keyframes in a section within a pre-saved layering grid map.

[0053] Figure 30 is a diagram of the configuration of the detection information character word of the listing structure of the detection information character of Figure 29.

[0054] Figure 31 is a schematic diagram of keyframes in a section within a detected layering grid map.

[0055] Figure 32 is a diagram of the configuration of the detection information character word of the listing structure of the detection information character of Figure 31.

[0056] Figure 33 is a schematic comparative state diagram of Figures 30 and 32.

[0057] FIG. 34 is a schematic diagram of at least a portion of a keyframe of a section within a pre-stored grid map within a layering grid map that is acquired and verified in a driving environment and stored in a unit storage.

[0058] Figure 35 is a configuration diagram of the detection information character word of Figure 34.

[0059] Figure 36 is a state diagram of a key frame within a grid map section formed from sensing information about a driving environment acquired through a lidar sensing unit at the location of a grid map-based mobile robot unit of the present invention.

[0060] Figure 37 is a configuration diagram of the detection information character word of Figure 36.

[0061] Figure 38 is a state diagram of various selection structures of the character unit cell of Figure 36.

[0062] Figure 39 is a configuration state diagram of the detection information character word according to the selection of (a), (b), and (c) of the character unit cell of Figure 38.

[0063] Figure 40 is a diagram of a process for comparing the matching of detection information character words for Figure 39 (a).

[0064] Figure 41 is a diagram of the process for comparing the matching of the detection information character words for (b) of Figure 39.

[0065] Figure 42 is a diagram of the process for comparing the matching of the detection information character words for (c) of Figure 39.

[0066]

[0067] Hereinafter, specific details for implementing the configuration of the mobile robot unit, mobile robot system, and control method thereof of the present invention will be described based on examples with reference to the drawings. These examples are described in sufficient detail to enable those skilled in the art to practice the present invention. It should be understood that the various embodiments of the present invention, while different from each other, are not necessarily mutually exclusive. For example, specific shapes, structures, and characteristics described herein may be implemented in other embodiments without departing from the spirit and scope of the present invention. Furthermore, it should be understood that the positions or arrangements of individual components within each disclosed embodiment may be changed without departing from the spirit and scope of the present invention. Therefore, the following detailed description is not intended to be limiting, and the scope of the present invention is defined only by the appended claims, along with the full scope equivalents to which such claims are entitled, if properly described. Like reference numerals in the drawings designate the same or similar functions throughout.

[0068] Unless otherwise defined, all terms (including technical and scientific terms) used herein may be used in their common sense to those of ordinary skill in the art to which the present invention pertains. Furthermore, terms defined in commonly used dictionaries are not to be interpreted ideally or excessively unless explicitly and specifically defined otherwise.

[0069] In the embodiment described below, the server may be implemented as a standalone device or as part of a general-purpose processing device in the form of a collection of hardware including a central processing unit, a user I / F, an operating system (not shown), a memory storing various necessary data, an external communication port, a printer I / F, etc., and software that operates the same, which performs various functions required for the present invention described below by executing various software programs and / or command sets stored in a memory unit.

[0070]

[0071] A mobile robot system (1) according to one embodiment of the present invention includes one or more layered grid map-based mobile robot units (10) and a grid map-based mobile robot server (20). In some cases, an administrator terminal (30), such as a task manager's computer, may also be included in the overall system. Each component constituting the mobile robot system (1) can communicate with each other through a communication network and transmit and receive data.

[0072] Figure 1 schematically illustrates the positional state of a driving environment in which a layered grid map-based mobile robot unit (10) acquires detection information through detection by a lidar detection unit (111, described later).

[0073] Here, the communication, communication network or communication network may include, for example, a cellular communication protocol, for example, at least one of LTE, LTE-A, 5G, WCDMA, CDMA, UMTS, Wibro, and GSM. In addition, the communication network may be configured regardless of the communication type, such as wired or wireless, and may be implemented as various communication networks, such as a personal area network (PAN), a local area network (LAN), a metropolitan area network (MAN), and a wide area network (WAN). The communication, communication network or communication network may be the well-known World Wide Web (WWW), and include various communication methods, such as infrared (Infrared Data Association; IrDA) or Bluetooth or Bluetooth Low Energy and RF wireless communication.

[0074] First, one or more grid map-based mobile robot units (10) are provided. The grid map-based mobile robot units (10) drive along a reference path described below to execute a predetermined task, and enable stable task execution by driving along a detour path other than the reference path.

[0075] More specifically, the grid map-based mobile robot unit (10) of the present invention includes a unit detection unit (11), a unit control unit (12), a unit storage unit (13), a unit operation unit (14), a unit input unit (15), a unit communication unit (16), a unit output unit (17), and a unit driving unit (18).

[0076] The unit detection unit (11) is a detection means for acquiring the driving status and surrounding environment information of the grid map-based mobile robot unit (10). As illustrated in FIGS. 2 and 3, the unit detection unit (11) includes a laser sensor. The laser sensor is implemented as a lidar sensor (111) and is a component that performs a scan detection function. The laser sensor detects the presence of obstacles, such as other mobile robot units driving in the opposite direction requiring crossing or objects that may collide, i.e., objects that affect the driving operation, and extracts point cloud data of surrounding objects such as fixed pillars and walls, to enable map creation and location estimation of the driving environment. In addition, the unit detection unit (11) may further include an infrared detection unit (113) that acquires infrared information about the driving environment and a motor detection unit (115) that detects the driving operation of the unit driving unit described below. In addition, in some cases, an image sensor (not illustrated) may capture an image of the surroundings while the grid map-based mobile robot unit (10) is driving and convert and provide image information.

[0077]

[0078] The unit control unit (12) may apply a detection operation control signal to the unit detection unit (11) of the grid map-based mobile robot unit (10), receive the detected detection information to generate a grid map or enable location estimation, and transmit the information to the grid map-based mobile robot server (20) to generate a grid map.

[0079] The unit control unit (12) includes a grid map generation control module (121), a characterization control module (123), a layer leveling control module (124), a keyframe managing module (125), and a matching control module (126).

[0080] The grid map generation control module (121) generates a grid map through the occupancy status of cells in the grid map corresponding to the driving environment from the detection information.

[0081] The characterization control module (123) generates a detection information character for a character unit cell (Cu) of a preset unit size through the occupancy status of a cell in a grid map.

[0082] The layer leveling control module (124) enables the creation of a layering map by repeatedly dividing the grid map into a preset number of sections to form a multi-layer structure.

[0083] The keyframe management module (125) registers and manages frames that meet the keyframe preset conditions among the multiple frames acquired by detection information as keyframes. In more detail,

[0084] The keyframe managing module (125) includes a keyframe preliminary selection unit (1251) and a keyframe registration unit (1253). The keyframe preliminary selection unit (1251) selects a keyframe candidate frame from among a plurality of frames for which detection information has been acquired, checks whether the frame satisfies a preset condition, and if the frame satisfies the condition, the keyframe registration unit (1253) registers and manages the frame as a keyframe.

[0085] The alignment control module (126) individually aligns the detection information characters acquired by the detection information within a plurality of frames and individually performs alignment between the corresponding character words between the registered plurality of key frames. The alignment control module (126) may, in some cases, be equipped with a word character keyframe alignment unit (1261) and an inter-keyframe word character alignment unit (1263), as illustrated in FIG. 5. The character keyframe alignment unit (1261) may check whether the detection information character words within the keyframes are aligned, and the inter-keyframe word character alignment unit (1263) may check whether the detection information character words between keyframes are aligned. Alternatively, the alignment control module (126) may utilize the unit operation unit (14) or the like to check whether the detection information characters / character words within or between frames are aligned, or may check whether the detection information characters / character words within or between keyframes are aligned, and various other selections are possible.

[0086] In addition, the unit control unit (120) of the grid map-based mobile robot unit (10) further includes a position estimation control module (127), which individually or in multiple ways aligns the detection information characters within a plurality of frames acquired by the detection information and performs individual or multiple alignments between the corresponding character words between the plurality of registered key frames in the alignment control module (126), thereby estimating the current position in the driving environment by using the results of the alignment of the detection information characters within the plurality of frames acquired by the detection information and the alignment between the corresponding character words between the plurality of registered key frames.

[0087]

[0088] In this embodiment, the following description is made with a focus on the case where the grid map-based mobile robot unit (10) is mounted on the grid map-based mobile robot unit (10) and performs independent autonomous driving by executing the map creation method and position estimation method described below. However, it is clear from this technology that such functional elements may be mounted on the grid map-based mobile robot server (20) and transmitted to the grid map-based mobile robot unit (10) to perform an autonomous driving function under the control of the grid map-based mobile robot server (20).

[0089]

[0090] The unit storage unit (13) is operated according to the unit storage control signal of the unit control unit (12), and may perform the function of storing unit pre-data for driving operation or storing image information or detection data in the form of a point cloud detected by the unit detection unit (11).

[0091] The unit operation unit (14) may utilize unit dictionary data stored in the unit storage unit to execute a predetermined operation process according to the unit operation control signal of the unit control unit (12). In addition, the unit input unit (15) may enable direct data input into the grid map-based mobile robot unit (10) in addition to input through a terminal (not shown) of a user or management operator.

[0092] The unit communication unit (16) communicates with the grid map-based mobile robot server (20) through wired or wireless communication, and can receive a motion control signal transmitted from the grid map-based mobile robot server (20) or transmit detection data detected by the unit detection unit (11) of the grid map-based mobile robot unit (10).

[0093] The unit output unit (17) is attached to the grid map-based mobile robot unit (10) to output signals externally. The unit output unit (17) can be implemented as various output elements such as a speaker or display.

[0094] The unit drive unit (18) includes various power generation and power transmission means, such as wheels, drive motors, and reducers of the mobile robot unit. The grid map-based mobile robot unit (10) of the present embodiment is not limited to a specific drive type, and can be modified in various ways, such as a single drive type, a differential type, or a quad type.

[0095]

[0096] As illustrated in FIG. 6, a grid map-based mobile robot server (20) may create a map of a driving environment sensed by one or more grid map-based mobile robot units (10) and execute position estimation.

[0097] The server control unit (22) receives the detection information detected from the grid map-based mobile robot unit (10) to generate a grid map or enable location estimation, and can also transfer the information to the grid map-based mobile robot unit (10) to move and generate a grid map.

[0098] As illustrated in Fig. 7, the server control unit (22) includes a grid map generation control module (221), a characterization control module (223), a layer leveling control module (224), a keyframe managing module (225), and a matching control module (226).

[0099] The grid map generation control module (221) generates a grid map through the occupancy status of cells in the grid map corresponding to the driving environment from the detection information.

[0100] The characterization control module (223) generates a detection information character for a character unit cell of a preset unit size through the occupancy status of a cell in a grid map.

[0101] The layer leveling control module (224) enables the creation of a layering map by repeatedly dividing the grid map into a preset number of sections to form a multi-layer structure.

[0102] The keyframe managing module (225) registers and manages frames that meet keyframe preset conditions among multiple frames for which detection information has been acquired as keyframes.

[0103] The alignment control module (226) individually aligns the detection information characters within the plurality of frames in which the detection information has been acquired and individually performs alignment between the corresponding character words between the plurality of registered key frames.

[0104] In addition, the server control unit (220) of the grid map-based mobile robot server (20) further includes a position estimation control module (227), which individually or in multiple ways aligns the detection information characters within a plurality of frames acquired by the detection information and performs individual or multiple alignments between the corresponding character words between a plurality of registered key frames in the alignment control module (226), thereby estimating the current position in the driving environment by using the results of the alignment of the detection information characters within a plurality of frames acquired by the detection information and the alignment between the corresponding character words between a plurality of registered key frames.

[0105] The server storage unit (23) is operated according to the server storage control signal of the server control unit (22), and may store server dictionary data for driving operations or store image information or point cloud-type detection data detected by the server detection unit (21) from the grid map-based mobile robot unit. The server operation unit (24) may utilize server dictionary data, etc. stored in the server storage unit to execute a predetermined operation process according to the server operation control signal of the server control unit (22). The server communication unit (25) may execute a communication function with each grid map-based mobile robot unit (10) to transmit control signals and receive detection information.

[0106] A series of grid map-based mobile robot servers (20) can create a map of a driving environment sensed by one or more grid map-based mobile robot units (10) and execute position estimation. The server can also be configured to receive information sensed from mobile robot units, create a map using the overall sensed information, and execute position estimation of the corresponding mobile robot unit and transmit the information to the robot units in real time. The map creation method and the position estimation method are the same as when the layering grid map mobile robot unit executes them. The following description focuses on the map creation and position estimation method using the robot units.

[0107]

[0108] Hereinafter, with reference to the drawings of the present invention, a method for creating a layered grid map and a method for estimating a position based on a layered grid map using a grid map-based mobile robot (10) and / or a grid map-based mobile robot system according to the present invention will be described.

[0109] First, the method for creating a layering grid map using a grid map-based mobile robot (10) and / or a grid map-based mobile robot system (1) according to the present invention includes a characterization step (S110), a frame alignment confirmation step (S120), a key frame alignment confirmation step (S130), and a layering map creation step (S140).

[0110] First, a preparation step (S100) is provided at the tip of these steps, and the provision of a grid map-based mobile robot (10) and / or a grid map-based mobile robot system or the movement process of a grid map-based mobile robot (10) may be executed (see FIG. 20).

[0111] After that, the characterization step (S110) is executed, in which detection information is acquired from the driving environment and a detection information character for a character unit cell of a preset unit size is confirmed and generated through the occupancy status of cells in a grid map corresponding to the driving environment.

[0112] The characterization step (S110) more specifically includes an environment detection step (S111), a detection information acquisition step (S113), and a sensing character confirmation step (S115) (see Fig. 21).

[0113] First, in the environment detection step (S111), environmental detection of the driving environment is performed through the lidar detection unit (111) of the detection unit (11), and in the detection information acquisition step (S113), point cloud detection information is acquired from the environment detection information, and the occupancy / unoccupancy status of the cells in the grid map corresponding to the driving environment is confirmed (see Fig. 9). Here, the occupancy status of the cells in the grid map is expanded as the grid map-based mobile robot (10) moves and the detection area of ​​the detection unit (11) expands, and this series of processes can be controlled by the map generation control module (121).

[0114] The environmental information scanned through the lidar detection unit (111) of the unit detection unit (11) in the environmental detection step (S111) and / or the detection information acquisition step (S113) constitutes one frame, where the frame refers to the environmental information obtained by projection through the scanning operation, and the frame can be selected in various ways, such as being formed at time intervals according to the setting conditions and according to the implementation of a specific operation.

[0115] After that, the sensing character confirmation step (S115) is executed by the characterization control module (123) of the unit control unit (12), and in the sensing character confirmation step (S115), a sensing information character is confirmed and generated. Here, an occupied state / unoccupied state is formed due to the presence of a point cloud in a unit cell for a character unit cell of a preset unit size from the point cloud detection information, and the sensing information character refers to a character formed as an occupied state / unoccupied state in a character unit cell (Cu, see FIG. 10).

[0116] Figures 11(a) to (z) illustrate examples of detection information characters for a 5X5 sized character unit cell (Cu). Various detection information characters are generated and confirmed according to the occupied / unoccupied state in the cell within the character unit cell (Cu), which is the basic information unit of the present invention, and are 1 and 0, respectively, for a total of 2, according to the occupied / unoccupied state of each cell. 25 A detection information character consisting of information of a dog can be identified, and the detection information character is a total of 2 in this embodiment. 25 Various combinations of characters can be generated. That is, in the sensing character confirmation step (S115), the sensing information characters within the frame can be extracted and stored in the unit storage unit (13).

[0117] In some cases, there may not be a sensing character based on the occupied cells within a frame. However, if the character unit cells are all formed as non-occupied cells, these may be formed as empty sensing characters and formed as separate sensing characters. Various options are possible depending on the design specifications.

[0118]

[0119] After that, a frame alignment verification step (S120) is executed, and in the frame alignment verification step (S120), detection information characters within a plurality of frames from which detection information has been acquired can be individually and / or multiplely aligned.

[0120] The frame alignment verification step (S120) includes a frame sensing character alignment step (S121), a key frame verification step (S123), and a key frame registration step (S125) (see Fig. 22).

[0121] The frame sensing character matching step (S121) acquires sensing information from a driving environment and matches a plurality of frames containing the occupancy status of cells in a grid map corresponding to the driving environment with the sensing information characters within the plurality of frames. The frame sensing character matching step (S121) can be executed through a matching control module (126) of a unit control unit (12).

[0122] That is, the sensing character as described above is selected within a frame formed by scan data on the driving environment obtained through the lidar detection unit (111) according to time, location, or kinematic change, and the connection status of the sensing character is confirmed by integrating the alignment between the sensing characters within a plurality of frames.

[0123] In addition, the linkage status of the sensing characters is confirmed through the alignment between the sensing characters. For example, as illustrated in FIG. 10, the sensing information character (Sc1) in the identified character unit cell (Cu1) within the frame expressed as a grid map indicated by the drawing symbol frame 1 is displayed, and the sensing information characters (Sc1, Sc2) within the identified character unit cells (Cu1, Cu2) within the frame expressed as a grid map indicated by the drawing symbol frame 2 are displayed. Since the linkage status of the sensing information characters (Sc1) in frames 1 and 2 is confirmed through the alignment process, it may be possible to confirm the mutual positional relationship between frames 1 and 2, the positional information of the sensing information character (Sc1), and the relative positional information of the newly identified sensing information character (Sc2) in frame 2.

[0124]

[0125] In the frame sensing character matching step (S121), after the detection information characters in multiple frames are matched, the key frame verification step (S123) is executed. In the key frame verification step (S123), it is checked whether the frames matched in the frame sensing character matching step (S121) meet the pre-key frame conditions. In Fig. 8, the drawing symbols key frame 0, key frame 1, and key frame 2 are illustrated, and between the key frames, multiple frames (frame 01 to frame 0N, frame 0NP, frame 11 to frame 1N, frame 1NP, frame 21 to frame 2N, frame 2NP) are acquired, and among these frames, if a specific condition is met, such as a new detection information character (Sc) being generated, a preset kinematic motion being performed, or a preset time being reached, the corresponding frame (frame ONP, frame 1NP, frame 2NP) is set to key frame 0, key frame 1, and key frame 2.

[0126] In the keyframe registration step (S125), the registered keyframe is stored in the unit storage unit (13). At this time, one or more sensing information characters (Sc) are also provided in the keyframe, and these may include location information on a grid map within the driving environment through the frame sensing character matching process as described above. That is, after the keyframe registration step (S125) is executed, the position correction can be confirmed using the matching error confirmed in the matching process between the sensing information characters within the frame through the matching control module (126) of the unit control unit (12). The frame matching confirmation step (S120) includes a matching error confirmation step (S127), and in the matching error confirmation step (S127), if the matching control module (126) confirms that there is no matching error when matching the frame and the sensing character in the frame sensing character matching step (S121), the control flow ends step S120 and proceeds to step S130.

[0127] On the other hand, the alignment error confirmation step (S127) is executed after the keyframe registration step (S125), and if it is confirmed that an alignment error has occurred during alignment between the frame and the sensing character in the frame sensing character alignment step (S121), the control flow proceeds to step S160, and the position correction step (S160) is executed. At this time, the map generation control module (121) of the unit control unit (12) may, in some cases, correct the position information for map generation and update the generation of the grid map.

[0128] After the frame alignment verification step (S120) is completed, the control flow proceeds to the keyframe alignment verification step (S130). The keyframe alignment verification step (S130) is executed via the alignment control module (126). In the keyframe alignment verification step (S130), one or more detection information characters within the registered keyframe are set as character words, and individual and / or multiple alignments between the corresponding character words among the registered multiple keyframes can be executed.

[0129] As illustrated in FIG. 23, the keyframe alignment verification step (S130) includes a frame character word verification step (S131), in which the character word formed as a detection information character as one or more sensing characters in a character unit cell (Cu) formed as 5X5 within the keyframe verified and registered in the keyframe registration step (S125) is verified by the characterizing control module (123).

[0130] In Fig. 12, at least a part of a key frame of a grid map formed from sensing information about a driving environment acquired at a location of a grid map-based mobile robot unit (10) is displayed, and in Fig. 13, a key frame character word (Wc) as a sensing information character string composed of a plurality of sensing information characters (Sc1, Sc2, Sc3, Sc4, Sc5, Sc6, Sc7, Sc8) within a key frame is displayed.

[0131] Then, after the frame character word verification step (S131) ​​is completed, the character word alignment verification step (S133) between key frames is executed. That is, the key frame alignment verification step (S130) includes the character word alignment verification step (S133) between key frames, and in the character word alignment verification step (S133) between key frames, the character words between multiple key frames that were confirmed and registered in the key frame registration step (S125) are aligned through the alignment control module (126) of the unit control unit (12).

[0132] As illustrated in Fig. 23a, the step of checking the character word alignment between key frames (S133) includes a step of checking the key frame character word selection (S1331), a step of checking the key frame character word similarity (S1333), and a step of confirming the key frame character word alignment (S1335).

[0133] In the keyframe character word selection confirmation step (S1331), the character word between multiple keyframes is selected and confirmed.

[0134] In Fig. 14, at least a part of the key frames of the pre-stored grid map within the grid map stored in the unit storage (13), and in Fig. 15, a plurality of detection information characters (Sc1) within the stored key frames * ,Sc2 * ,Sc3 * ,Sc4 * ,Sc5 * ,Sc6 * ,Sc7 * ,Sc8 * ) as a detection information character string, the keyframe character word (Wc * ) is displayed.

[0135] In the keyframe character word similarity confirmation step (S1333), a similarity comparison is performed between the keyframe character words selected and confirmed in the keyframe character word selection confirmation step (S1331). Then, in the keyframe character word matching confirmation step (S1335), whether the keyframe character words are matched is confirmed based on the similarity confirmation result in the keyframe character word similarity comparison step (S1333), and the location of the newly registered keyframe is confirmed.

[0136] As illustrated in Fig. 23b, the keyframe character word similarity confirmation step (S1333) includes a keyframe character word similarity calculation step (S13331), a keyframe character word similarity comparison step (S13333), and a keyframe character word similarity status confirmation step (S13334).

[0137] In the keyframe character word similarity calculation step (S13331), the similarity between the keyframe character words selected and confirmed in the keyframe character word selection confirmation step (S1331) is calculated.

[0138] In the keyframe character word similarity comparison step (S13333), the similarity (Sw) between each keyframe character word calculated in the keyframe character word similarity calculation step (S13331) is compared with the preset character word similarity (Skw).

[0139] In the keyframe character word similarity status check step (S13334), the keyframe character word similarity status is checked based on the comparison result in the keyframe character word similarity comparison step (S13333).

[0140] If, in the keyframe character word similarity status check step (S13334), the matching control module (126) determines that the keyframe character word in the target keyframe and the keyframe character word in the corresponding keyframe stored in the unit storage unit (13) are similar, the control flow is transferred to step S13335 to determine whether the keyframe character word is matched and to check the location of the newly registered keyframe.

[0141] If, in the keyframe character word similarity status check step (S13334), the matching control module (126) determines that the keyframe character word in the target keyframe and the keyframe character word in the corresponding keyframe stored in the unit storage unit (13) are dissimilar, the status is checked as a dissimilar status (S13337), and the subsequent control flow is transferred to step S13331 so that the similarity comparison process may be repeated.

[0142]

[0143] In (a) and (b) of FIG. 16, at least a part of the key frame of the grid map formed from the sensing information about the driving environment acquired at the location of the grid map-based mobile robot unit (10) of FIG. 12 and FIG. 14 is displayed as region A, and at least a part of the key frame of the pre-stored grid map stored in the unit storage unit (13) is displayed as region B, and in FIG. 17, through the key frame character word selection confirmation step (S1331) and the key frame character word similarity confirmation step (S1333), a key frame character word (Wc) of the string 'ㄴㅏㅂlㅍㅡㄹㅏ' as an example of a sensing information character string composed of a plurality of sensing information characters (Sc1, Sc2, Sc3, Sc4, Sc5, Sc6, Sc7, Sc8) in the key frame of FIG. 13, and a plurality of sensing information characters (Sc1, Sc2, Sc3, Sc4, Sc5, Sc6, Sc7, Sc8) in the stored key frame of FIG. 15 * ,Sc2 * ,Sc3 * ,Sc4 * ,Sc5 * ,Sc6 * ,Sc7 * ,Sc8 * ) as a detection information character string, the keyframe character word (Wc *) are compared and displayed, and the similarity is calculated through individual matching comparison for each detection information character or detection information character word as a result of the comparison, and through comparison with the preset character word similarity (Skw), whether it is the same area or a different area is determined through the keyframe character word matching confirmation step (S1335) as a result of similarity, so that the grid map expansion update is possible through the matching process between key frames. In Fig. 16 (a), when comparing the detection information character word in the registered key frame stored as the key frame of the previous layer grid map with the detection information character word in the scan-confirmed frame (b), among the detection information character words formed by a total of 8 detection information characters of (a) and (b), only 3 detection information characters are the same and the remaining 5 are different, so that the similarity is 3 / 8, and whether or not it is matched can be confirmed through comparison with the preset character word similarity (Skw).

[0144] After a series of processes like this are executed, the layering map creation step (S140) is executed. That is, the layering map creation step (S140) is executed through the layer leveling control module (123) of the unit control unit (12), and by repeatedly dividing the grid map into a preset number of sections using individual or composite matching information between the corresponding character words between multiple key frames, a multi-layer structure is formed, and the layering map, i.e., the layering grid map, is expanded and updated to create an overall layering grid map.

[0145] The layer leveling control module (124) of the unit control unit (12) enables the creation of a layering map by repeatedly dividing the grid map into a preset number of sections to form a multi-layer structure. That is, each frame and key frame and the detection information character and character word formed therein are data structured into a multi-layer formed according to the layer leveling control module (124). The grid map has a multi-layered division structure with a preset number. When the grid map is placed as a first layer and divided into N×N planes, each section of N×N is placed as a second layer and divided into N×N again, and each section is divided again, thereby forming a repetitive re-repetition structure. For example, FIG. 18 illustrates a case where each layer forms a multi-layered division structure with a 3×3 structure, and the layers can form n layers in an n-th form.

[0146] In Fig. 19, an example of a multi-layered partition structure is shown for a very small area of ​​a small driving environment of 0.27 m x 0.27 m in an exemplary form in this embodiment, and is composed of three layers, and each layer forms a 3 x 3 layer-by-layer repeated partition structure, and the 3 x 3 partitioned section of the third layer at the bottom can be set as the character unit cell (Cu) described above, and 2 through occupancy / non-occupancy information in this character unit cell. 25 It can contain various detection information characters having data.

[0147] The keyframes mentioned above can be configured to be aligned with sections of a multi-layered segmentation structure layer, thereby minimizing the risk of performance degradation caused by signal delay due to the computational time and computational load of scanning the entire data of the grid map during the map creation process.

[0148]

[0149] Meanwhile, the present invention can execute a location estimation process using a layering grid map created according to the layering grid map creation method described above.

[0150] First, a layering grid map-based position estimation method according to another embodiment of the present invention includes a position estimation characterization step (S210) and a position estimation matching confirmation step (S220). Before these steps, a position estimation preparation step (S200) may be executed, and in the position estimation preparation step (S200), a layering grid map-based mobile robot unit (10) capable of moving within a driving environment in which a grid map layering map is formed in a multi-layer structure in the manner described above is provided and positioned, and the layering grid map-based mobile robot unit (10) includes the components described above, and a duplicate description thereof is omitted.

[0151] In the position estimation characterization step (S210), a layering map formed into a multi-layer structure by repeatedly dividing a preset number of sections is used to acquire sensing information from a corresponding driving environment, and a sensing information character for a character unit cell of a preset unit size is confirmed and generated through the occupancy status of cells in a grid map corresponding to the driving environment so as to estimate the location. That is, in the position estimation characterization step (S210), a sensing information character is formed through sensing information detected in the driving environment through a layering grid map-based mobile robot unit (10), and the position estimation characterization step (S210) includes a position estimation environment sensing step (S211), a sensing information acquisition step (S213), and a sensing character confirmation step (S215) (see FIG. 25).

[0152] In the position estimation environment detection step (S211), environment detection is performed from the driving environment through the lidar detection unit (111) of the unit detection unit (11). Information detected by the lidar detection unit (111) of the unit detection unit (11) can be transmitted to the unit storage unit (13), the unit operation unit (14), and the unit control unit (12) and subjected to predetermined signal processing.

[0153] After the location estimation environment detection step (S211) is executed, the detection information acquisition step (S213) is executed. In the detection information acquisition step (S213), point cloud detection information is acquired from the environment detection information to confirm the occupancy status of cells in the grid map corresponding to the driving environment.

[0154] In the detection information acquisition step (S213), point cloud detection information is acquired from the environment detection information, and the occupancy / unoccupancy status of the cells in the grid map corresponding to the driving environment is confirmed, which is the same as what was performed in the layering grid map creation process above (see Fig. 9). Here, the occupancy status of the cells in the grid map is expanded as the search area for position estimation expands according to the movement of the grid map-based mobile robot (10) and the expansion of the detection area of ​​the detection unit (11), and this series of processes can be controlled by the position estimation module (127) of the unit control unit (12).

[0155] After the detection information acquisition step (S213) is executed, the sensing character verification step (S215) is performed. In the sensing character verification step (S215), a detection information character for a character unit cell of a preset unit size is verified and generated from the point cloud detection information. The sensing character verification step (S215) is executed by the characterization control module (123) of the unit control unit (12). In the sensing character verification step (S215), a detection information character is verified and generated through the characterization control module (123).

[0156] Here, the occupied / unoccupied state is formed due to the presence of the point cloud in the unit cell for the character unit cell of the preset unit size from the point cloud detection information, and the configuration in which the detection information character refers to the character formed as the occupied / unoccupied state in the character unit cell (Cu, see FIG. 10) is the same as the detailed process of the layering grid map manufacturing method described above. For example, examples of detection information characters for character unit cells (Cu) of the size of 5X5 are shown in (a) to (z) of FIG. 11.

[0157] As described above, various detection information characters are generated and confirmed according to the occupied / unoccupied status in the character unit cell (Cu) as the basic information unit of the present invention, and a total of 2 are generated, 1 and 0, according to the occupied / unoccupied status of each cell. 25 A detection information character consisting of information of a dog can be identified, and the detection information character is a total of 2 in this embodiment. 25 Various combinations of characters can be generated. That is, in the sensing character confirmation step (S215), the sensing information characters within the frame can be extracted and stored in the unit storage unit (13).

[0158]

[0159] After that, the position estimation matching confirmation step (S220) is executed (see FIG. 26). The position estimation matching confirmation step (S220) is executed by the position estimation control module (127) of the unit control unit (12), and the characterization control module (123) and the matching control module (126) can be used. The position estimation control module (127) uses the detection information detected by the lidar detection unit (111) of the unit detection unit (11), and when the detection information character (see FIGS. 29 and 30) produced by the characterization control module (123) in the acquired frame is confirmed, the detection information character is individually or multiplely matched with the detection information character (see FIGS. 31 and 32) in the layering grid map stored in the unit storage unit (13) through the matching control module (116) to compare the similarity of the detection information character word composed of the corresponding detection information characters.

[0160] Based on the results of this similarity comparison, the estimated location of the point occupied by the current layering grid map-based mobile robot unit (10) can be accurately estimated.

[0161] More specifically, the position estimation matching confirmation step (S220) includes a detection character word confirmation step (S221) and a layering map matching position confirmation step (S223). The detection character word confirmation step (S221) is performed through the position estimation control module (127) and the characterizing control module (123), and the detection information acquired from the driving environment through the lidar detection unit (111) includes the occupancy status of cells in the grid map corresponding to the driving environment, and within the frame, a detection information character word formed as one or more sensing characters as the character unit cells (Cu) described above is confirmed, and the confirmation of the detection information characters or the detection information character words within this series of frames is substantially the same as in the map creation process.

[0162] In Fig. 31, a detection information character word (Wc) is illustrated as a detection information character string, such as in Fig. 32, which is composed of a plurality of detection information characters (Sc1, Sc2, Sc3, Sc4, Sc5, Sc6, Sc7, Sc8) that are confirmed for at least a part of a key frame within a grid map section formed from detection information about a driving environment acquired through a lidar detection unit (111) at the location of a grid map-based mobile robot unit (10).

[0163] After that, the layering map matching position confirmation step (S223) is executed (see Fig. 27). The layering map matching position confirmation step (S223) is performed through the position estimation control module (127) and the matching control module (126). Depending on the design specifications, the layer leveling control module (124), the keyframe managing module (125), and the matching control module (126) may additionally be utilized. That is, in the layering map matching position confirmation step (S223), the detection information character word confirmed in the detection character word confirmation step (S221) is compared and matched with the detection information character word in the keyframe registered for each layering area in the layering grid map of the multi-layered division structure. At this time, the detection information character word in the keyframe forms the data structure of the layering grid map of the multi-layered division structure, so that the comparison and confirmation work is performed quickly and accurately with the scanned confirmed detection information character word for position estimation, thereby reducing the computational load and enabling the execution of an accurate position estimation process.

[0164] In more detail, the layering map matching position confirmation step (S223) includes a layering map section confirmation step (S2231), a layering map character word confirmation step (S2233), a character word matching adjustment step (S2235), a similarity confirmation step (S2236), and a similarity comparison step (S2237).

[0165] The layering map section verification step (S2231) can be executed using the layer grid map information stored in the unit storage (13) through the position estimation control module (127), and in some cases, the layer leveling control module (124) can be used to use layers for key frames. That is, in the layering map section verification step (S2231), the section of the layering grid map of the current position to be estimated is verified using the layering grid map, and the key frame of the corresponding section is verified.

[0166] After that, the layering map character word verification step (S2233) is executed to verify the detection information character word within the corresponding layering grid map section within the corresponding layering grid map, and the detection information character word within a series of key frames is stored as layering grid map data of a multi-layered division structure in the unit storage unit (13).

[0167] FIG. 29 illustrates at least a portion of a key frame of a section within a pre-stored grid map within a layering grid map that is acquired in a driving environment and stored in a unit storage unit (13), wherein a plurality of detection information characters (Sc1) within the key frame stored in the unit storage unit (13) are illustrated. * ,Sc2 * ,Sc3 * ,Sc4 * ,Sc5 * ,Sc6 * ,Sc7 * ,Sc8 * ) is composed of a keyframe character word (Wc) that is arranged in multiple rows as in Fig. 30. * ) is shown.

[0168] After that, the character word alignment adjustment step (S2235) is executed, in which the detection information character word within the keyframe of the layering grid map section and the scanned detection information character word are rotated and position-transformed to individually align and adjust the character word as a series of detection information characters. At this time, the rotation and position-transformation data of the detection information character confirmed during the alignment adjustment process can be used as inverse operation data values ​​for confirming accurate location information for the current location search point.

[0169] After that, the similarity confirmation step (S2236) is executed, which is substantially the same as or similar to the similarity comparison process in the previous layering grid map creation process. That is, the similarity confirmation step (S2236) is executed through the matching control module (126) according to the control signal of the position estimation control module (127), and the similarity is calculated and confirmed by the identity ratio calculated using the number of corresponding and compared detection information characters and the total number of characters in the entire detection information character word, and ultimately, the character word similarity between the layering grid map section character word stored in the converted unit storage unit (13) and the scanned and detected detection information character word is calculated and confirmed.

[0170] After that, a similarity comparison step (S2237) is executed, and the detection information character word similarity confirmed in the similarity confirmation step (S2236) is compared with the preset character word similarity (Ss).

[0171] In Fig. 33, keyframe character words (Wc) on a layered grid map configured in multiple lists as in Fig. 30 are *) and a comparison is performed on the detected detection information character word (Wc) as in FIG. 32, and this comparison is performed through the layering map matching position confirmation step (S223) described above. In some cases, the comparison may be performed by comparing each corresponding character individually, or in some cases, the entire comparison may be performed by comparing each word to calculate the matching rate and compare the similarity, etc. Various options are possible depending on the design specifications.

[0172]

[0173] If the detection information character word similarity is less than the preset character word similarity (Ss) as a result of the comparison in the similarity comparison step (S2237), the control flow switches to the layering map section verification step (S2231) to repeat the layering grid map section verification process. That is, the layering grid map section verification process is repeated, and a comparison target other than the section key frame or the detection information character word within the key frame determined to be dissimilar as a result of the comparison in the similarity comparison step (S2237) is selected, and a position where the similarity is satisfied can be confirmed through a predetermined iterative search process, and this search process can be executed more quickly and accurately due to the data structure formed in the layering grid map method.

[0174]

[0175] On the other hand, if the detection information character word similarity is greater than or equal to the preset character word similarity (Ss) as a result of the comparison in the similarity comparison step (S2237), the control flow switches to the location confirmation step (S2238). The location confirmation step (S2238) of the layering map matching location confirmation step (S223) is executed through the location estimation control module (127). In the location confirmation step (S2238), if the character word similarity is determined to be greater than the preset character word similarity in the similarity comparison step (S2237), the current location is estimated and confirmed using the matching adjustment transformation data confirmed in the character word matching adjustment step (S2235). That is, when the detection information character word and the scanned detection information character word within the keyframe of the layering grid map section are individually aligned or adjusted as a character word by rotating and positioning the detection information character unit in a series of detection information characters, the rotation and positioning data of the detection information character confirmed in the alignment adjustment process can be used as the inverse operation data value to determine the exact location information for the current location search point.

[0176] In Fig. 28, when the detected information character word in the keyframe confirmed in the labeling grid map section stored in the unit storage unit (13) and the scanned confirmed detected information character word are matched, the rotation and position transformation data ([R1|t1], [R2|t2], [R3|t3], [R4|t4], [R5|t5], [R6|t6], [R7|t7], [R8|t8].) generated in the process of matching the detected information character word in the keyframe and the scanned confirmed detected information character word are inversely transformed to estimate and produce the scanned position of the detected information character word in the corresponding keyframe.

[0177] That is, a grid map corresponding to the surrounding environment to be estimated in this way is formed in a multi-layered structure, and n layers in an n-th form can be formed, and each layer is divided into, for example, N×N, so that it is divided into N×N in the first layer on a plane, and each N×N section of the first layer is placed as a second layer, and this is divided into N×N again, and each section is re-divided again, thereby forming a repetitive re-repeated layer to form a multi-layered divided structure (see Figs. 18 and 19), and the detection information character word formed by the detection information character in the scan-detected detection information is compared with the detection information character word in the corresponding layering grid map to confirm the similarity, thereby executing a practical primary location estimation in the corresponding driving space, and also, when the similarity between the detection information character word formed by the detection information character in the scan-detected detection information and the detection information character word in the corresponding layering grid map satisfies the preset similarity condition, the detection information character word matching is calculated when the character words are compared and matched. By inversely calculating using displacement and rotation values, a secondary position estimation process can be performed to more accurately derive position information of a point where the corresponding scan-detected detection information character word is obtained.

[0178] In this way, the layering map matching position confirmation step (S223) for a series of position estimation confirmations including the layering map section confirmation step (S2231), the layering map character word confirmation step (S2233), the character word alignment adjustment step (S2235), the similarity confirmation step (S2236), and the similarity comparison step (S2237) is the same as the primary similarity comparison confirmation and the secondary position estimation process using displacement and rotation transformation data, as shown in FIGS. 29 to 33 and FIG. 28, which are cases for the general environment described above.

[0179] Meanwhile, the present invention is not limited to the cases of FIGS. 29 to 33, which are exemplified as cases of general driving environments with many features. That is, FIGS. 34 to 42 illustrate another implementation example of the present invention, which can ensure robustness of estimation by enabling accurate position estimation even in environments with fewer features.

[0180] As in the previous cases, FIG. 34 illustrates at least a part of a key frame of a section within a pre-stored grid map within a layering grid map that is acquired and confirmed in a driving environment and stored in a unit storage unit (13), and a plurality of detection information characters (Sc1) within the key frame stored in the unit storage unit (13). * ,Sc2 * ,Sc3 * ,Sc4 * ,Sc5 * ,Sc6 * ,Sc7 * ,Sc8 * ) is composed of a keyframe character word (Wc) that is arranged in multiple rows as in Fig. 35. * ) is shown.

[0181] Unlike the cases of FIGS. 29 and 30 shown previously, the detection information characters and character words within the pre-stored grid zone of FIGS. 34 and 35 are shown as being arranged in a clockwise direction from the upper right, and this viewpoint and direction can be selected in various ways depending on the design specifications.

[0182] In Fig. 36, a plurality of detection information characters (Sc1, Sc2, Sc3, Sc4, Sc5) are identified for at least a part of the key frames within the grid map section formed from detection information about the driving environment acquired through the lidar detection unit (111) at the location of the grid map-based mobile robot unit (10), and a detection information character word (Wc) is illustrated as a detection information character string as in Fig. 37. In this case, unlike the previous case, if a separate detection information character does not exist for an area identified as an empty space, a separate detection information character may not be assigned (see Fig. 37), and in some cases, an empty detection information character (eSc6, eSc7, eSc8) that assigns an empty state may be assigned to form a character string (Fig. 36), and various other options are possible.

[0183] Similarity is confirmed through comparison of the stored character word with the detected character word, and whether or not it matches is determined through the similarity comparison, and location estimation can be performed by inverse calculation using displacement and rotation data during the matching process, as described above.

[0184] Meanwhile, in this case, in the process of confirming and generating the detection information character through the occupancy status of the cells in the driving environment response grid map for the detected detection information involved by the characterizing control module (123), even if the character unit cell (Cu) changes, the position estimation method of the present invention can secure position estimation robustness because the result value according to the change is substantially the same. That is, in FIG. 38, for the detection information detected through the lidar detection unit (111) of the unit detection unit (11), the character unit cell (Cu) is distinguished into three cases (a), (b), and (c), and the detection information character words (Wc) of these three cases are displayed in FIG. 39, and the three cases (a), (b), and (c) have different detection information character word values ​​because the positions of the segmented character unit cells (Cu) are different.

[0185] The detection information character words of the three cases (a), (b), and (c) derived from the scan-detected detection information of Fig. 39 are the detection information characters (Sc1) within the keyframe of the layering grid map data pre-stored in the unit storage unit (13) of Fig. 35. * ,Sc2 * ,Sc3 * ,Sc4 * ,SC5 * ) listed as detection information character word (Wc * ) is compared to calculate the similarity, and the comparison process of three cases (a), (b), and (c) is illustrated in Figs. 40 to 42, and the detection information character word (Wc * ) and the detection information character word (Wc) for each of the three cases are compared, at which time the character word alignment adjustment step (S2235) is executed to form a displacement or rotation transformed detection information character word (Wcm) of the detection information character word (Wc) for each of the three cases, ultimately forming the detection information character word (Wc *) and the converted detection information character word (Wcm) are compared, and in the case of (b) of FIG. 41, displacement transformation of the third and fourth detection information characters (Sc3, Sc4) occurs, and in the case of (c) of FIG. 42, displacement transformation of the first, second, third, and fifth detection information characters (Sc1, Sc2, Sc3, Sc5) occurs, and displacement and rotation transformation data ([R|T]) are generated during the matching process.

[0186] In the case of (a) and (b) of FIG. 40 and FIG. 41, the similarity due to alignment is the same, and in the case of (c) of FIG. 42, the similarity has a slightly different configuration. Although there are some differences depending on the selection of the entire character unit cell (Cu), since alignment is achieved through conversion of the detection information character during the alignment process, the robustness of the position estimation result to the difference in the selection of the character unit cell (Cu) can be secured to a certain extent.

[0187] Meanwhile, in the previous embodiments, a structure was adopted to calculate the similarity of the entire sensed information character word through the process of matching the sensed information characters of the comparison target and whether they are the same or different, but the similarity calculation method of the present invention is not limited to this.

[0188] In some cases, the ratio of the number of overlapping occupied cells of the detected information character (Sc) confirmed by scan detection to the number of occupied cells of the detected information character (Sc*) in the layering grid map data within the character unit cell (Cu) may be calculated as a similarity weight, and may be formed as the sum of the similarity weights of the detected information character (Sc) confirmed by scan detection to the number of characters of the entire compared target detected information character word.

[0189] For example, in the case of Fig. 40, if according to the above criteria, the similarity (S) is calculated to have a similarity of 0.375 through 3 identical detection information characters out of a total of 8, but if the similarity weight is utilized, in the case of Sc1, 2 out of 4 cells overlap and have a similarity of 2 / 4, in the case of Sc2, 5 out of 5 cells overlap and have a similarity of 1, in the case of Sc3, 5 out of 5 cells overlap and have a similarity of 1, in the case of Sc4, 5 out of 5 cells overlap and have a similarity of 1, and in the case of Sc5, 2 out of 5 cells overlap and have a similarity reflected with a weight of 2 / 5, so that the overall similarity can be configured to have a similarity of (2 / 4+1+1+1+2 / 5) / 8=0.43.

[0190] By enabling adjustments in the similarity judgment depending on the presence or absence of such weights and the values ​​of the weights, it is possible to enable variable selection according to the driving environment in terms of accuracy and speed.

[0191]

[0192] While the present invention has been illustrated and described with reference to preferred embodiments intended to illustrate the principles of the invention, it is not intended to be limited to the exact configuration and operation described herein. Rather, those skilled in the art will readily appreciate that numerous modifications and variations are possible without departing from the spirit and scope of the appended claims.

[0193]

[0194] The present invention is used to control a mobile robot unit that enables rapid and accurate operation through rapid map creation, and this technology can be applied to various industrial fields and daily life fields, such as delivery within the city center and indoor driving, in addition to industrial sites.

Claims

1. A characterization step (S110) for obtaining detection information from the driving environment and generating a detection information character for a character unit cell of a preset unit size through the occupancy status of cells in a grid map corresponding to the driving environment, A frame alignment verification step (S120) that aligns the detection information characters within multiple frames from which the detection information has been acquired, and A keyframe matching confirmation step (S130) in which a frame that meets the keyframe preset conditions among the above frames is registered as a keyframe, a detection information character within the keyframe is set as a character word, and matching between the corresponding character words among the registered multiple keyframes is performed; A method for creating a layering grid map, comprising a layering map creation step (S140) of creating a layering map by repeatedly dividing the grid map into a preset number of sections using the matching information between the corresponding character words between the plurality of key frames to form a multi-layer structure.

2. In paragraph 1, The above characterization step (S110) is: An environment detection step (S111) in which environment detection is performed from the driving environment, A detection information acquisition step (S113) in which point cloud detection information is acquired to confirm the occupancy status of cells in a grid map corresponding to the driving environment from the above environmental detection information, A layering grid map creation method characterized in that it includes a sensing character verification step (S115) in which a sensing information character for a character unit cell of the preset unit size is verified and generated from the point cloud detection information.

3. In paragraph 2, The above frame alignment verification step (S120) is: A frame sensing character matching step (S121) for matching the sensing information characters within the plurality of frames, which includes the occupancy status of cells in a grid map corresponding to the driving environment by acquiring sensing information from the driving environment, and A key frame verification step (S123) for checking whether the frame aligned in the above frame sensing character alignment step (S121) meets the pre-key frame conditions, A layering grid map creation method characterized in that it includes a key frame registration step (S125) of setting the corresponding frame as a key frame if it is determined that the photo key frame condition is met in the above key frame confirmation step (S123).

4. In paragraph 3, The above frame alignment verification step (S120) is: In the above frame sensing character alignment step (S121) A method for creating a layering grid map, characterized in that it includes a matching error checking step (S127) in which a matching error is checked when matching between a frame and a sensing character.

5. In paragraph 4, A layering grid map creation method characterized in that it includes a position correction step (S160) in which, if a matching error is confirmed during the matching between the frame and the sensing character in the above matching error confirmation step (S127), the position information acquired for the corresponding frame is corrected and confirmed based on the frame sensing character matching error.

6. In paragraph 4, A layering grid map creation method, characterized in that the keyframe alignment verification step (S130) includes a keyframe character word verification step (S131) ​​of verifying a character word formed by one or more sensing characters within the keyframes verified and registered in the keyframe registration step (S125).

7. In paragraph 6, A layering grid map creation method characterized in that the above keyframe alignment confirmation step (S130) includes a keyframe character word alignment confirmation step (S133) for confirming alignment of character words between a plurality of keyframes confirmed and registered in the keyframe registration step (S125).

8. In paragraph 7, The character word alignment check step (S133) between the above key frames is: A keyframe character word selection confirmation step (S1331) for selecting and confirming a character word among the above multiple keyframes, A keyframe character word similarity confirmation step (S1333) for performing a similarity comparison between the keyframe character words selected and confirmed in the above keyframe character word selection confirmation step (S1331), A layering grid map creation method characterized by including a keyframe character word matching confirmation step (S1335) for confirming whether keyframe character words are matched based on the similarity confirmation result in the keyframe character word similarity comparison step (S1333) and confirming the location of a newly registered keyframe.

9. In paragraph 8, The above keyframe character word similarity check step (S1333) is: In the keyframe character word similarity calculation step (S13331), A keyframe character word similarity calculation step (S13331) in which the similarity between the keyframe character words selected and confirmed in the above keyframe character word selection confirmation step (S1331) is calculated, and A keyframe character word similarity comparison step (S13333) in which the similarity (Sw) between each keyframe character word calculated in the keyframe character word similarity calculation step (S13331) is compared with a preset character word similarity (Skw), A layering grid map creation method characterized in that it includes a keyframe character word similarity status confirmation step (S13334) in which the keyframe character word similarity status is confirmed according to the comparison result in the keyframe character word similarity comparison step (S13333).

10. A position estimation characterization step (S210) for obtaining detection information from a driving environment corresponding to a layered grid map formed into a multi-layer structure by repeatedly dividing into a preset number of sections, and generating a detection information character for a character unit cell of a preset unit size through the occupancy status of cells in the grid map corresponding to the driving environment to estimate a position, and A layering grid map-based location estimation method characterized by including a location estimation alignment confirmation step (S220) of checking and aligning the detection information characters within the acquired frame to confirm whether they are aligned and calculating and confirming the corresponding estimated location.

11. In paragraph 10, The above location estimation characterization step (S210) is: A location estimation environment detection step (S211) in which environment detection is performed from the driving environment, and A detection information acquisition step (S213) in which point cloud detection information is acquired to confirm the occupancy status of cells in a grid map corresponding to the driving environment from the above environmental detection information, A layering grid map-based location estimation method characterized in that it includes a sensing character verification step (S215) in which a sensing information character for a character unit cell of the preset unit size is verified and generated from the point cloud detection information.

12. In paragraph 11, The above location estimation alignment verification step (S220) is: A detection character word verification step (S221) for obtaining detection information from the driving environment and verifying the detection information character word within the frame that includes the occupancy status of the cell in the grid map corresponding to the driving environment, A layering grid map-based location estimation method, characterized in that it includes a layering map matching position confirmation step (S223) of comparing and matching the above-mentioned detection information character word and the detection information character word in the key frame of each layering area in the layering map.

13. In paragraph 12, The above layering map alignment position confirmation step (S223) is: A layering map section confirmation step (S2231) for confirming the layering map section of the current location using the above layering map, A layering map character word verification step (S2233) for verifying the character word in the above layering map section, A character word alignment adjustment step (S2235) for adjusting alignment by converting the layering map section character word and the detection information character word, A similarity check step (S2236) for checking the character word similarity between the converted layering map section character word and the detection information character word, A layering grid map-based location estimation method, characterized in that it includes a similarity comparison step (S2237) for comparing the character word similarity confirmed in the above similarity confirmation step (S2236) with a preset character word similarity.

14. In paragraph 13, The above layering map alignment position confirmation step (S223) is: A layering grid map-based location estimation method characterized by including a location confirmation step (S2238) for estimating and confirming a current location using the alignment adjustment transformation data confirmed in the character word alignment adjustment step (S2235) when the character word similarity is determined to be greater than the preset character word similarity in the similarity comparison step (S2237).

15. A detection unit (11) including a lidar detection unit (111) that obtains detection information from the driving environment, A grid map generation control module (121) through the occupancy status of cells in the grid map corresponding to the driving environment from the above detection information, A characterization control module (123) that generates and confirms detection information characters for character unit cells of preset unit sizes through the occupancy status of cells in the above grid map, A layer leveling control module (124) that enables the creation of a layering map by repeatedly dividing the above grid map into a preset number of sections to form a multi-layer structure, A keyframe managing module (125) that registers and manages frames that meet keyframe preset conditions among the multiple frames obtained by the above detection information as keyframes, A layering grid map-based mobile robot unit (10) including a matching control module (126) that matches the detection information characters within a plurality of frames acquired from the above detection information and performs matching between the corresponding character words between a plurality of registered key frames.

16. In paragraph 15, A layering grid map-based mobile robot unit (10), characterized in that it includes a position estimation control module (127) that estimates a current position in a driving environment by using the results of the alignment of the detection information characters within the plurality of frames acquired by the detection information and the alignment between the corresponding character words between the plurality of registered key frames in the alignment control module (126) that aligns the detection information characters within the plurality of frames acquired by the detection information and executes the alignment between the corresponding character words between the plurality of registered key frames.

17. A grid map generation control module (221) that obtains sensing information about the driving environment from a grid map-based mobile robot unit (10) equipped with a sensing unit (11) including a lidar sensing unit (111) and uses the occupancy status of cells in the grid map corresponding to the driving environment, and A characterization control module (223) that generates and confirms detection information characters for character unit cells of preset unit sizes through the occupancy status of cells in the above grid map, A layer leveling control module (224) that enables the creation of a layering map by repeatedly dividing the above grid map into a preset number of sections to form a multi-layer structure, A keyframe managing module (225) that registers and manages frames that meet keyframe preset conditions among the multiple frames acquired by the above detection information as keyframes, A mobile robot system (1) comprising a grid map-based mobile robot server (20) including a matching control module (226) that matches the characters of the detection information acquired in the plurality of frames and performs matching between the corresponding character words between the plurality of registered key frames.

18. In paragraph 17, The above grid map-based mobile robot server (20): A mobile robot system (1) characterized in that it includes a position estimation control module (227) that estimates a current position in a driving environment by using the results of the alignment of the detection information characters within the plurality of frames acquired by the detection information and the alignment between the corresponding character words between the plurality of registered key frames in the alignment control module (126) that aligns the detection information characters within the plurality of frames acquired by the detection information and performs the alignment between the corresponding character words between the plurality of registered key frames.

Images