KR20210006138A - Indoor autonomousdriving system of transpotation robot - Google Patents

Abstract

According to one embodiment of the present invention, a method for assisting autonomous driving of a transportation robot by an autonomous driving control server comprises the steps of: obtaining an image of an indoor space photographed by a transportation robot from the transportation robot; and extracting a drivable area from the obtained image, and learning the drivable area for the indoor space.

Description

Indoor autonomous driving system of transport robot {INDOOR AUTONOMOUSDRIVING SYSTEM OF TRANSPOTATION ROBOT}

The present invention relates to an indoor driving system of a transport robot, and more particularly, learns a driving area through a machine learning algorithm of deep learning using an indoor space image captured by a transport robot, and within the driving area. The present invention relates to a method for accurately driving a transport robot according to a preset path by checking the existence of an obstacle, and an indoor driving system for performing the same.

With the recent rapid increase in logistics volume, the needs for logistics automation in the logistics service industry are increasing. Among these, as part of logistics automation, transport robots that perform logistics picking are being introduced on a pilot basis, and transport robots deliver various goods that are moved or delivered in large indoor spaces such as factories or logistics warehouses. can do.

On the other hand, in order to introduce a transport robot capable of working 24 hours while minimizing changes to existing processes, it can be said that accurately recognizing the environment of a large indoor space that changes in real time is the core of the autonomous driving operation of the transport robot.

Accordingly, various technologies have been developed for transport robots to respond to changes in the indoor environment. Representatively, Odometry, which measures the position change of the transport robot by attaching sensors such as encoders, infrared rays, and ultrasonic waves to the transport robot. Self-driving technology is mentioned.

However, if the transport robot runs autonomously without any information received from the outside, the transport robot may be pushed down depending on the material of the travel path, or due to various reasons such as malfunction of the sensor attached to the transport robot, failure, and the placement of new obstacles. Path errors inevitably occur.

Therefore, it is required to develop a new technology and system that can learn the driving range using the image of the indoor space acquired by the camera mounted on the transport robot, recognize obstacles, and travel according to the planned route set by the transport robot. The present invention relates to this.

Korean Registered Patent Publication No. 10-1944497 (2019.01.25.)

The technical problem to be solved by the present invention is to determine the existence of an obstacle by attaching a camera to a transport robot placed in an indoor environment where GPS is not available, acquiring an image of an indoor space, and learning it through a machine learning algorithm of deep learning. And to provide a method and system capable of inferring a drivable area.

Another technical problem to be solved by the present invention is a method and system capable of detecting whether a transport robot deviates from a preset route by using a learned driveable area, a camera attached to a transport robot, and a sensor, and correcting its position. Is to provide.

The technical problems of the present invention are not limited to the technical problems mentioned above, and other technical problems that are not mentioned will be clearly understood by those skilled in the art from the following description.

A method of assisting the autonomous driving of a transport robot by an autonomous driving control server according to an embodiment of the present invention, the method comprising: acquiring an image of an indoor space photographed by the transport robot from the transport robot, and driving from the acquired image The transport robot by extracting an area and learning a drivable region for the indoor space, by generating a travel path determined according to the driving status of a transport robot disposed in the indoor space based on the learned drivable region And transmitting.

According to an embodiment, when the transport robot is traveling, transmitting the travel path to the transport robot includes obtaining an image from the transport robot, and inputting the acquired image into the learned driveable area. The method may further include inferring a drivable area, and checking whether an obstacle exists in the inferred drivable area.

According to an embodiment, if it is confirmed that an obstacle exists in the checking step, a relative distance between the transport robot and the obstacle is calculated, and a driving path in the inferred driving area is corrected based on the relative distance. It may further include the step of.

According to an embodiment, the relative distance between the transport robot and the obstacle may be calculated based on the learned dimension of the drivable area or driving information of the transport robot.

According to an embodiment, when the obstacle is a different transport robot, determining a collision possibility based on driving information of each of the adjacent transport robots, and correcting a preset driving path for each of the adjacent transport robots. have.

According to an embodiment, if it is confirmed that there is no obstacle in the checking step, the current position coordinates of the transport robot are calculated based on the image and the learned driveable area, and the calculated position coordinates and driving It may further include transmitting a route determination signal to the transport robot.

According to an embodiment, after transmitting the travel route to the transport robot, receiving the relative position coordinates calculated by the transport robot together with image data captured by the transport robot from the transport robot, the reception Calculating absolute position coordinates of the transport robot based on one image data and the learned driveable area, determining whether a difference between the relative position coordinates and the absolute position coordinates is out of an error range, and out of the error range If it is determined that it is determined, the step of calculating a position correction value of the transport robot may be further included.

According to an embodiment, the transmitting of the travel route to the transport robot includes: receiving a route plan from the transport robot and inputting the received route plan into the learned driveable area, thereby defining a non-driving area. It may further include the step of confirming.

According to an embodiment, when it is determined that the non-driving area exists, generating a new route plan moving along a center line within the learned driving possible area may be further included.

According to an embodiment, when it is determined that the non-driving area exists, searching for a driving available area having a high similarity to the route plan among the learned driving available areas, and generating a new route plan, , The similarity may be calculated based on a moving direction of the transport robot or a moving distance according to the route plan.

According to the present invention, in an indoor environment where GPS is not available, by learning an available driving area through a machine learning algorithm using indoor spatial image information acquired through a camera attached to a transport robot, the relative distance between the transport robot and the obstacle, There is an effect of being able to accurately infer the relative position of a transport robot in an indoor space.

In addition, since a drivable area is inferred in an indoor space using a machine learning algorithm of deep learning, it is possible to accurately recognize a drivable area even if there is a change in the indoor space.

In addition, the position of the transport robot is estimated using auxiliary sensors installed in the transport robot (e.g., inertial sensor, encoder, lidar), and if it deviates from a preset path, the error is corrected based on the learned driving area. Thus, it is possible to calculate the optimal travel path at the current point in time, and there is an effect that the transport robot can travel efficiently.

The effects of the present invention are not limited to the above-mentioned effects, and other effects that are not mentioned will be clearly understood by those skilled in the art from the following description.

1 is a diagram showing a schematic configuration of an indoor autonomous driving system according to an embodiment of the present invention.
2 is a flowchart illustrating a flow of an indoor autonomous driving assistance method according to an embodiment of the present invention.
3 is a flowchart illustrating steps S210 and S220 shown in FIG. 2.
FIG. 4 is a diagram for explaining a method in which an autonomous driving control server learns an available driving area according to an embodiment of the present invention.
5 is a flowchart illustrating step S230 illustrated in FIG. 2.
6 is a flow chart illustrating step S2304 shown in FIG. 5.
7 is a diagram illustrating a method of correcting a driving route by an autonomous driving control server according to an embodiment of the present invention.
FIG. 8 is a diagram for explaining a method in which an autonomous driving control server corrects a driving route according to another embodiment of the present invention.
9 is a flowchart illustrating a method of determining a route plan by an autonomous driving control server according to an embodiment of the present invention.
10 is a flowchart illustrating a method of correcting a position of a transport robot by an autonomous driving control server according to an embodiment of the present invention.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. Advantages and features of the present invention, and a method of achieving them will become apparent with reference to the embodiments described below in detail together with the accompanying drawings. The same reference numerals refer to the same components throughout the specification.

Unless otherwise defined, all terms (including technical and scientific terms) used in the present specification may be used as meanings that can be commonly understood by those of ordinary skill in the art to which the present invention belongs. In addition, terms defined in a commonly used dictionary are not interpreted ideally or excessively unless explicitly defined specifically. The terms used in the present specification are for describing exemplary embodiments and are not intended to limit the present invention. In this specification, the singular form also includes the plural form unless specifically stated in the phrase.

As used herein, “comprises” and/or “comprising” refers to the recited component, step, operation, and/or element, of one or more other elements, steps, operations and/or elements. It does not exclude presence or addition.

1 is a diagram showing a schematic configuration of an indoor autonomous driving system 1000 according to an exemplary embodiment of the present invention, and is a diagram divided by functions performed by each component.

Referring to FIG. 1, it can be seen that the indoor autonomous driving system 1000 includes an autonomous driving control server 100 that learns an available driving area in an indoor space and a transport robot 200 that travels in an indoor space. The driving control server 100 includes a learning module 110, an inference module 120, and a communication module 130, and the transport robot 200 includes a camera module 210, a driving module 220, and a communication module 230. ) Can be included.

The learning module 110 of the autonomous driving control server 100 may learn a drivable area in which the transport robot 200 in the indoor space can drive based on an image photographed of the indoor space. Here, the image of the indoor space can be received from the transport robot 200, and after inferring the floor surface of the indoor space within the image, the width and length of the floor surface, and the grid are learned to drive the indoor space. A possible area model can be obtained.

The inference module 120 of the autonomous driving control server 100 includes the current position of the transport robot 200, the existence of an obstacle in the path the transport robot 200 travels, and the validity of the route plan set by the transport robot 200. , During autonomous driving of at least one or more transport robots 200 traveling in an indoor space, such as collisions between transport robots 200, various information necessary for autonomous driving may be inferred through the learned driveable area model.

For example, the inference module 120 receives path planning from the transport robot 200 and applies it to the learned drivable area model to determine if there is a blocked path, and another transport robot 200 It is possible to infer the validity of the route plan generated by the transport robot 200, such as whether there is a possibility of collision with or whether a more efficient route exists.

The communication module 130 of the autonomous driving control server 100 may receive various pieces of information for learning and inferring an available driving area of an indoor space from the transport robot 200 or a manager terminal (not shown). For example, the communication module 130 includes identification information of the transport robot 200, an image of the indoor space from the transport robot 200, the position coordinates of the transport robot 200 calculated by the transport robot 200, and the transport robot (200) driving information, etc. can be received, and identification information of facilities placed in the indoor space from the manager terminal, new equipment placed in the indoor space (e.g., racks for loading goods), and newly relocated information Map information of the indoor space, the number of transport robots 200 traveling in the indoor space, identification information, and the like may be received.

On the other hand, the transport robot 200 is an intelligent service robot that travels in space, and can operate autonomously by recognizing an external environment and determining a situation by itself.

The camera module 210 of the transport robot 200 is disposed at the front and rear of the transport robot 200 to continuously acquire an image of an indoor space along a path along which the transport robot 200 travels. To this end, the camera module 210 may be any one or more of a general analog camera, a network camera, a zoom camera, and a speed dome camera, and may be disposed in the front and rear as well as left and right/top and bottom to obtain an image of an indoor space.

The traveling module 220 of the transport robot 200 may include a hardware component for driving the transport robot 200 in an indoor space and a software component for generating a route plan for traveling in the indoor space. Here, the route plan generated by the driving module 220 may include global path planning and local path planning. At this time, the global route plan means an optimal route plan generated by the transport robot 200 from the starting point to the target point, and the regional route plan detects obstacles using sensors provided in the transport robot 200. , It means a path plan that is generated in real time to avoid obstacles.

To this end, the transport robot 200 may include sensors such as an encoder, an inertial sensor (IMU, Inertia Measurement Unit), an infrared sensor, an ultrasonic sensor, and a lidar, and transport through the above-described sensors. The x-coordinate, y-coordinate, and moving angle of the robot 200 may be calculated.

The communication module 230 of the transport robot 200 communicates with the autonomous driving control server 100 to provide various information on the driving of the transport robot such as an image of an indoor space, driving information, position coordinates, presence of obstacles, and route planning. I can receive it. To this end, the communication unit 110 may support cellular communication networks such as 2G/3G, Long Term Evolution (LTE), LTE-Advance (LTE-A), and 5G, and may support wireless networks such as Wi-Fi. .

On the other hand, since the position coordinates and route plans recognized by the transport robot 200 are relative position coordinates calculated through sensors, the information generated by the transport robot 200 is a change in the material of the driving path surface or frictional force in the indoor space. The accuracy is deteriorated in various unpredictable unexpected situations, such as the occurrence of, and a malfunction of a sensor provided in the transport robot 200. Accordingly, the transport robot 200 may receive information on a driving route based on the driving available area model learned by the autonomous driving control server 100 using the image of the indoor space.

Hereinafter, a method in which the autonomous driving control server 100 learns an available driving area and provides information on a driving route in order to provide information on an indoor space to the transport robot 200 will be described.

Hereinafter, a method of assisting indoor autonomous driving of a transport robot according to an embodiment of the present invention will be described with reference to FIGS. 2 to 10.

FIG. 2 is a flow chart showing a flow of an indoor autonomous driving assistance method according to an embodiment of the present invention, and FIG. 3 is a flowchart illustrating steps S210 and S220 shown in FIG. 2.

This is only a preferred embodiment for achieving the object of the present invention, and if necessary, some steps may be deleted or added, or one step may be included in another step and performed.

The autonomous driving control server 100 acquires an image of an indoor space captured by the transport robot 200 from the transport robot 200 (S210). To this end, the transport robot 200 traveling in the indoor space periodically acquires an image of the indoor space using a camera disposed in the front or rear, and this is an autonomous driving control server along with driving information of the transport robot 200 Can be sent to (100). Here, the driving information includes identification information of the transport robot 200 and driving information such as speed and location, and the image may include a time point at which the image is captured.

That is, the autonomous driving control server 100 may periodically receive an image of an indoor space from the transport robot 200 and collect the image of the indoor space to obtain an image of the indoor space.

After step S210, the autonomous driving control server 100 performs labeling for extracting the driving possible area from the previously acquired image, and learns the driving possible area for the indoor space (S220). Here, the image of the indoor space may be an image of the entire space or an image of a partial space.

FIG. 3 is a flow chart illustrating a method of learning an image of an indoor space of the autonomous driving control server 100, and step S220 will be described in more detail with reference to FIG. 3.

Referring to FIG. 3, the autonomous driving control server 100 acquires an image from the transport robot 200 (S2201) and preprocesses the acquired image (S2202). Here, image pre-processing is a process of purifying meaningful data from an image of an indoor space captured by the transport robot 200, excluding unnecessary information, and may be performed through various methods. For example, the image preprocessing may be performed using gray scale, binarization, enlargement/reduction, rotation/transformation, or the like.

That is, the autonomous driving control server 100 pre-processes the acquired image and converts the image to be used for learning.

After step S2202, the autonomous driving control server 100 checks whether or not the confirmed driving possible area is labeled (S2203). In other words, the autonomous driving control server 100 may determine the existence of the labeling data, and if the labeling data does not exist, it labels the confirmed driving available area (S2203-1, NO).

In this regard, FIG. 4 is a diagram for explaining a method in which the autonomous driving control server 100 according to an embodiment of the present invention learns a driving available area. Referring to FIG. 5, the autonomous driving control server 100 Receives the indoor space image as shown in (a) from the transport robot 200, and performs labeling of the drivable area (hatched).

Referring back to FIG. 3, if it is confirmed that the driving available area is labeled in step S2203, the autonomous driving control server 100 checks whether the currently collected data is sufficient (S2204, YES).

If it is determined that the data is insufficient in step S2204, the autonomous driving control server 100 can augment the data based on the previously learned content (S2204-1, NO), and, conversely, the data in step S2204. When it is confirmed that it is sufficient, the autonomous driving control server 100 learns a driving possible area for an indoor space by using a machine learning algorithm of deep learning using the acquired data (S2205, YES). That is, the autonomous driving control server 100 may check the drivable area based on the model learned in advance, and may determine whether an obstacle exists in an indoor space corresponding to the image.

Meanwhile, in order to learn the above-described items, the autonomous driving control server 100 may utilize various machine learning algorithms such as supervised learning.

After step S2205, the autonomous driving control server 100 verifies whether the model for checking the driving available area for the indoor space has been trained (S2206), and if the learning is completed, it may be stored (S2207, YES). More specifically, the autonomous driving control server 100 may determine whether training of the model has been completed based on the training data and the loss value of the verification data. That is, the autonomous driving control server 100 may check whether an obstacle exists in an image of an indoor space newly received from the transport robot 200 by using the trained model.

On the other hand, if it is determined that the learning has not been completed in step S2205, the autonomous driving control server 100 may continuously update the weight inside the model by using a back-propagation (S2206-1, NO. ). However, various algorithms that are not limited thereto may be used.

In this way, the autonomous driving control server 100 can learn the driving area, which is a key in the driving of the transport robot 200, using the image of the indoor space transmitted by the transport robot 200, and actually transport The robot 200 may provide information required in the process of driving the indoor space.

Returning to the description of FIG. 2 again.

Referring to FIG. 2, through step S220, the autonomous driving control server 100 may complete learning of a drivable area for an indoor space, and the autonomous driving control server 100 is an indoor space based on the learned drivable area. A travel route determined according to the driving status of the transport robot 200 disposed in is generated and transmitted to the transport robot 200 (S230). For example, the driving status of the transport robot 200 may include a situation in which the transport robot 200 is currently traveling in an indoor space, a situation in which the transport robot 200 wants to move according to a route plan set by itself, and the like. All situations related to the driving of the transport robot 200 may be included.

5 is a flow chart embodied in step S230 shown in FIG. 2, and FIG. 5 is a transport robot 200 is currently traveling in an indoor space, acquires an image of the indoor space, and transmits it to the autonomous driving control server 100 One case is described.

Referring to FIG. 5, the autonomous driving control server 100 acquires an image from the transport robot 200 (S2301). Through this, the autonomous driving control server 100 may acquire identification information of the transport robot 200, and at the same time acquire the identification information and load the previous travel path of the transport robot 200, Information for inferring the current location coordinates may be obtained.

After step S2301, the autonomous driving control server 100 may pre-process the image (S2302), and may extract data on an available driving area from the image transmitted by the transport robot 200.

After step S2302, the autonomous driving control server 100 inputs the preprocessed image data into the trained model (S2303), and infers a driving available area (S2304). That is, the inference of the driving possible area means that the autonomous driving control server 100 checks the existence of an obstacle in the image of the indoor space taken based on the current position of the transport robot 200, and determines the driving possible/impossible area. This can be done by using a drivable area divided by pixel units.

In addition, the non-driable area may be, for example, an area in which a warehouse shelf is placed, an area in which goods or another transport robot 200 is placed, and the autonomous driving server 100 can also classify reasons for which it is impossible. I can.

In this regard, FIGS. 6 and 7 are attached, and FIG. 6 is a flow chart embodied in step S2304 shown in FIG. 5, and FIG. 7 is a self-driving control server 100 according to an embodiment of the present invention to correct a driving route. It is a diagram for explaining the method of doing.

Referring to FIG. 6, the autonomous driving control server 100 may check whether an obstacle exists by inputting image data into the learned model (S2304-1).

That is, as shown in (a) of FIG. 7, the autonomous driving control server 100 can check the existence of an obstacle that did not exist in the learning process by inferring a driving area using the image transmitted by the transport robot 200. .

When it is determined that an obstacle exists in step S2304-1, the autonomous driving control server 100 calculates the relative distance between the transport robot 200 and the obstacle, and based on the relative distance, the driving in the inferred driving range The path can be corrected (S230-4-2, YES). Specifically, the autonomous driving control server 100 may calculate driving coordinates using the learned dimensions of the driving available area and driving information of the transport robot 200, wherein the dimensions of the driving available area are arranged in an indoor space to It may mean a distance (D1) between the racks in which are loaded, a distance (D2) of a driving area, a position coordinate based on a center line (L) of the driving area, and the like.

In this regard, referring to (b) of FIG. 7, when the location and driving information of the current transport robot 200 is inclined to the right with respect to the center line L of the driving available area, the autonomous driving control server 100 May correct the travel path so that the transport robot 200 can travel in the right direction toward the center line L.

That is, the autonomous driving control server 100 may correct the driving path so that the transport robot 200 to be moved can move along the solid line as shown in the dotted line.

In addition, for a more specific correction, the autonomous driving control server 100 calculates the position coordinates of the current transport robot 200 using the image transmitted by the transport robot 200, and based on the center line (L) of the driveable area. By checking the coordinates of the obstacle through the grid, it is possible to generate a driving route correction value including a wheel angle, speed/wheel rotation number of the transport robot 200.

On the other hand, if it is determined that the obstacle does not exist in step S2304-1, the autonomous driving control server 100 is based on the image transmitted by the transport robot 200 and the learned driveable area. Position coordinates may be calculated, and a signal for determining the calculated position coordinates and a driving route may be generated (S2304-3).

On the other hand, a plurality of transport robots 200 are arranged in the indoor space so that each of them travels according to the route plan, and accordingly, not only items that do not move, but also another transport robot 200 that is traveling is one transport robot ( 200) can be considered an obstacle.

FIG. 8 is a diagram for explaining a method in which the autonomous driving control server 100 corrects a driving route according to another embodiment of the present invention.

Referring to FIGS. 8A and 8B, the autonomous driving control server 100 may identify the transport robot 200 on the drivable area from the image data. That is, in the drawing shown in FIG. 8, A is the transport robot 200 for which the autonomous driving control server 100 currently determines whether or not the driving route is corrected, and B and C are in the image captured by the transport robot A 200. It may be an existing transport robot 200.

The autonomous driving control server 100 may identify the transport robot 200 present in the image captured by the transport robot A 200 and receive driving information from two adjacent transport robots 200. Through this, the autonomous driving control server 100 may determine the possibility of collision of the transport robot 200 in consideration of the path planning and the driving speed of the two adjacent transport robots 200 at present.

For example, the autonomous driving control server 300 may determine that when the B, C transport robot 200 moves along a solid line in (a) and (b), the adjacent transport robot 200 will not collide. , When the B, C transport robot 200 moves along the dotted line, it is determined that the adjacent transport robot 200 will collide, and a preset driving route may be corrected for each.

It will return to the description of FIG. 5 again.

After step S2304, the autonomous driving control server 100 may output the above-described inference result and transmit it to the transport robot 200 (S2305). Accordingly, the transport robot 200 may move according to the originally set route plan or move according to the travel path transmitted by the autonomous driving control server 100, thereby enabling safe driving without colliding with an obstacle. That is, the presence of the obstacle can be recognized faster than the detection of the obstacle through the sensors provided in the transport robot 200, so that the probability of collision can be minimized.

Meanwhile, the autonomous driving control server 100 may receive an image of the indoor space captured by the transport robot 200 from the transport robot 200 and a route plan established by the transport robot 200 itself. Can be evaluated. This is a description of FIG. 9 and will be described below.

9 is a flowchart illustrating a route plan determination method of the autonomous driving control server 100 according to an embodiment of the present invention.

Referring to FIG. 9, the autonomous driving control server 100 may receive and obtain a global route or route plan set by itself from the transport robot 200 (S2310), and whether there is an undriable area in the obtained route. Check (S2320). That is, the autonomous driving control server 100 may input the received route plan into the learned driving available area to check the area in which driving is impossible. Here, the global route is a travelable area for an indoor space created based on the information possessed by the transport robot 200, and the route plan is a driving set by the transport robot 200 to go from the current position to a preset arrival point. Means route planning.

If it is determined that there is an undriable area in step S2320, the autonomous driving control server 100 searches for the learned drivable area (S2320-1), and generates a new route plan or a local route (S2320-2). Here, the new route plan may be a route plan that moves the center line within the drivable area. In addition, even if it is not the center line, any path may be possible as long as it moves through the drivable area. For example, if it is determined that the current location of the transport robot 200 is not located on the center line within the drivable area, based on the center line within the drivable area, the autonomous driving control server 100 transports in a new route plan. Information for correcting a position and a direction may be included so that the robot 200 can move along the center line of the drivable area.

Meanwhile, the new route plan may be calculated based on a moving direction set by the transport robot 200, a flow, or a moving distance according to a route plan, in addition to following the center line within the driveable area. That is, the autonomous driving control server 100 searches for at least one drivable area based on the area in which driving is not possible, and among them, the area that does not change significantly from the original movement direction or the movement distance is similar to the area with high similarity. It is also possible to select and create a new route plan or local route through the area.

As described above, the autonomous driving control server 100 may terminate the inference by transmitting a newly set or determined route to the transport robot 200, and the transport robot 200 receives received information (eg, new driving). You can safely drive through the indoor space according to the route, route plan confirmation signal).

On the other hand, the autonomous driving control server 100 may also receive location information determined by the transport robot 200 itself from the transport robot 200 that continuously transmits an image of an indoor space. ) The location information determined by itself may cause errors depending on the friction of the indoor space, the weight of the goods loaded by the transport robot 200, and the material of the floor of the indoor space, so check this periodically and check the correction information. Must be created. This is a description of FIG. 11 and will be described below.

10 is a flowchart illustrating a method of correcting the position of the transport robot 200 by the autonomous driving control server 100 according to an embodiment of the present invention.

Referring to FIG. 10, the steps S310 to S330 shown are steps performed by the transport robot 200, and the transport robot 200 sets an initial position (S310), photographs an indoor space, and while driving. After recording the driving amount, that is, driving information through the sensor (S320), the position coordinates may be calculated by itself and transmitted to the autonomous driving control server 100. Here, the position coordinate calculated by the transport robot 200 may be understood as a relative position coordinate.

In this way, while the transport robot 200 calculates the position coordinates, the autonomous driving control server 100 acquires an image of the indoor space from the transport robot 200 (S410), and uses the image to determine the driving area. After inferring (S420), the absolute position coordinates of the transport robot 200 may be calculated based on the learned driveable area (S430).

After step S430, the autonomous driving control server 100 may determine whether a difference between the relative position coordinates and the absolute position coordinates is within an error range (S440). Here, the error range may be set based on a time or distance according to a route plan currently set in the transport robot 200.

If it is determined that the difference between the coordinates is out of the error range in step S430, the autonomous driving control server 100 calculates a position correction value of the transport robot 200 (S440-1, NO), and this is the transport robot 200 The location can be determined by sending it to (S450), and in the opposite case, a location determination signal can be transmitted to the transport robot 200 together with the absolute position coordinates calculated by the autonomous driving control server 100 (S450, YES). ).

So far, a method of providing various information related to driving of the transport robot 200 based on an image of an indoor space captured by the transport robot 200 by the autonomous driving control server 100 according to an embodiment of the present invention Explained. According to the present invention, the autonomous driving control server 100 goes beyond simply identifying the location of the transport robot 200 to infer a driving area of an indoor space and measure the relative distance, so that the transport robot 200 When a set route or location is deviated, information for correcting it may be provided, and thus the accuracy of driving and thus the efficiency of transportation may increase during the operation of the indoor autonomous driving system 1000.

Meanwhile, the present invention can also be implemented as a computer-readable code on a computer-readable recording medium. Computer-readable recording media include all storage media such as magnetic storage media and optical reading media. It is also possible to record the data format of the message used in the present invention on a recording medium.

Embodiments of the present invention have been described above with reference to the accompanying drawings, but those of ordinary skill in the art to which the present invention pertains can be implemented in other specific forms without changing the technical spirit or essential features. You can understand. Therefore, it should be understood that the embodiments described above are illustrative and non-limiting in all respects.

1000: indoor autonomous driving system
100: autonomous driving control server
110: learning module
120: reasoning module
130: communication module
200: transport robot
210: camera module
220: drive module
230: communication module

Claims (11)

As a method for the autonomous driving control server to assist the autonomous driving of a transport robot,
Obtaining an image of an indoor space photographed by the transport robot from the transport robot; And
Extracting a drivable area from the acquired image and learning a drivable area for the indoor space;
Indoor autonomous driving assistance method of a transport robot comprising a. The method of claim 1,
After the step of learning the drivable area,
Generating a determined travel path according to the driving status of the transport robot disposed in the indoor space based on the learned travelable area, and transmitting it to the transport robot;
Indoor autonomous driving assistance method of the transport robot further comprising a. The method of claim 2,
When the transport robot is running,
Transmitting the travel route to the transport robot,
Obtaining an image from the transport robot and inputting the acquired image to the learned drivable area to infer a drivable area; And
Checking whether an obstacle exists in the inferred drivable area;
Indoor autonomous driving assistance method of the transport robot further comprising a. The method of claim 3,
If it is confirmed that an obstacle exists in the step of checking,
Calculating a relative distance between the transport robot and the obstacle, and correcting a travel path in the inferred drivable area based on the relative distance;
Indoor autonomous driving assistance method of the transport robot further comprising a. The method of claim 4,
The relative distance between the transport robot and the obstacle,
Calculated based on the learned dimension of the driving available area or driving information of the transport robot,
How to assist transport robots in indoor autonomous driving. The method of claim 4,
If the obstacle is another transport robot,
Determining a possibility of collision based on driving information of each of the adjacent transport robots, and correcting a preset travel path for each of the adjacent transport robots;
Indoor autonomous driving assistance method of the transport robot further comprising a. The method of claim 3,
If it is confirmed that there is no obstacle in the step of checking,
Calculating the current position coordinates of the transport robot based on the image and the learned travelable area, and transmitting the calculated position coordinates and a driving route determination signal to the transport robot;
Indoor autonomous driving assistance method of the transport robot further comprising a. The method of claim 2,
After the step of transmitting the travel route to the transport robot,
Receiving a relative position coordinate calculated by the transport robot together with image data captured by the transport robot from the transport robot;
Calculating absolute position coordinates of the transport robot based on the received image data and the learned driveable area;
Determining whether a difference between the relative position coordinates and the absolute position coordinates is out of an error range; And
Calculating a position correction value of the transport robot when it is determined that it is out of the error range;
Indoor autonomous driving assistance method of the transport robot further comprising a. The method of claim 2,
Transmitting the travel route to the transport robot,
Receiving a route plan from the transport robot; And
Inputting the received route plan into the learned driveable area to confirm a non-driving area;
Indoor autonomous driving assistance method of the transport robot further comprising a. The method of claim 9,
Generating a new route plan for moving along a center line within the learned driveable area when it is determined that the non-driving area exists;
Implementation autonomous driving assistance method of the transport robot further comprising a. The method of claim 9,
When it is determined that the non-driving area exists, searching for a drivable area having a high similarity to the route plan among the learned drivable areas, and generating a new route plan; Including more,
The similarity is,
Calculated based on the moving direction of the transport robot or the moving distance according to the route plan,
How to assist transport robots in indoor autonomous driving.

Images