KR20200078073A - Method for controlling drone and apparatus therefor - Google Patents

Abstract

The present invention relates to a drone control method and apparatus for accurately predicting and controlling a drone's posture through learning. According to the present invention, a method for controlling a drone flying at a long distance in a drone control system includes: a drone control device receiving and accumulating and storing flight data from a drone; Pre-processing the flight data by calculating a difference in attitude between the attitude information included in the stored flight data and the representative attitude information by the drone control device; The drone control apparatus repeatedly learning the posture prediction function by applying the posture difference and the flight data to a posture prediction function to converge the weights applied to the posture prediction function; And the drone control device transmitting the converged weights to the drone.

Description

Method for controlling drone and apparatus therefor

The present invention relates to a drone control technology, and more particularly, to a drone control method and apparatus for accurately predicting and controlling a drone's posture through learning.

Currently, various functions of drones are being developed. For example, various drones have been developed, such as shooting drones, agricultural drones, delivery drones, and air quality measurement drones, some of which are actually used in the field.

Furthermore, a technology for remotely controlling and controlling drones using a mobile communication network has emerged. Incidentally, in the past, the drone was controlled through short-range wireless communication such as Bluetooth communication, but a technique for controlling the drone via a mobile communication network has appeared to remotely control and control a drone located at a longer distance. The following patent document discloses a drone control system and method using an LTE network.

When the drone is controlled using the mobile communication network, the drone can be controlled to a greater distance, and the drone can also be controlled indoors.

When remote drones are controlled remotely, the control center transmits the drone flight route planning information to the drone, and the drone autonomously flies based on the flight route planning information. However, due to an external environment such as wind or an internal factor such as a sudden change of direction, a drone flying autonomously over a long distance may become unstable, and the drone may crash or collide with other obstacles.

Accordingly, various technologies are required to secure the drone's posture.

Korean Patent Publication No. 10-2018-0061514

The present invention has been proposed to solve this problem, and has an object to provide a drone control method and an apparatus therefor for accurately predicting a future drone posture and preventing a drone posture from becoming unstable.

Other objects and advantages of the present invention can be understood by the following description, and will be more clearly understood by examples of the present invention. In addition, it will be readily appreciated that the objects and advantages of the present invention can be realized by means of the appended claims and combinations thereof.

In a drone control system according to a first aspect of the present invention for achieving the above object, a method for controlling a drone flying at a long distance includes: a drone control apparatus receiving and accumulating and storing flight data from a drone; Pre-processing the flight data by calculating a difference in attitude between the attitude information included in the stored flight data and the representative attitude information by the drone control device; The drone control apparatus repeatedly learning the posture prediction function by applying the posture difference and the flight data to a posture prediction function to converge the weights applied to the posture prediction function; And the drone control device transmitting the converged weights to the drone.

The method may include applying, by the drone, the weights to a posture prediction function in which the weights are stored, and predicting a future drone posture based on the posture prediction function to which the weights are applied.

When the future drone posture identified based on the posture prediction function is predicted to be unstable, the drone may perform in-flight flight until the future drone posture is predicted to be stable.

The drone control device may select representative posture information for each speed. In this case, the step of pre-processing the flight data confirms the speed included in the flight data, and calculates a difference between the representative attitude information corresponding to the speed and the attitude information included in the flight data.

In the selecting step, the accumulated and stored flight data may be analyzed to select roll values, yaw values, and pitch values having the highest frequency at the corresponding speed as representative posture information of the corresponding speed.

According to a second aspect of the present invention for achieving the above object, the drone control apparatus for controlling a drone flying at a long distance includes a database for storing flight data of the drone; A data pre-processor for calculating posture differences between posture information and representative posture information included in the flight data and preprocessing the flight data; And a learning unit that repeatedly learns the posture prediction function by applying the posture difference and the flight data to the posture prediction function to converge the weights applied to the posture prediction function.

A drone capable of predicting a future position according to a third aspect of the present invention for achieving the above object includes: a storage unit for storing a posture prediction function; A wireless communication unit wirelessly communicating with a drone control device to receive weights from the drone control device; And applying the received weights to the attitude prediction function, collecting flight data, calculating a difference in attitude between the drone's attitude information and representative attitude information, and calculating the difference between the calculated attitude difference and the collected flight data. It includes a control unit for predicting the position of the future drone shown in the next cycle, input to the applied attitude prediction function.

The present invention has an advantage of learning a posture prediction function using a large amount of flight data and accurately predicting a drone's posture through an optimized posture prediction function.

In addition, the present invention derives a modeling function (that is, a posture prediction function) capable of accurately predicting the attitude of a drone by learning a posture prediction function by substituting a difference between representative posture information and actual posture information into the posture prediction function as an input value. There is an advantage to do.

In addition, the drone according to the present invention predicts the future drone posture through the optimized weighted posture prediction function, and when the predicted posture is determined to be unstable, the drone performs the in-flight flight to stabilize the future posture. It has the advantage of preventing falling.

The following drawings attached to this specification are intended to illustrate preferred embodiments of the present invention, and serve to further understand the technical idea of the present invention together with specific details for carrying out the invention, and thus the present invention is described in such drawings. It should not be interpreted as being limited to the matter.
1 is a view showing a drone control system according to an embodiment of the present invention.
2 is a view showing the configuration of a drone control device according to an embodiment of the present invention.
3 is a flowchart illustrating a method of collecting and storing flight data in a drone control device according to an embodiment of the present invention.
4 is a flowchart illustrating a method of learning a posture prediction function using a large amount of flight data in a drone control apparatus according to an embodiment of the present invention.
5 is a view showing a drone configuration according to an embodiment of the present invention.
6 is a flowchart illustrating a method of predicting a future posture based on a posture prediction function in a drone according to an embodiment of the present invention.

The above objects, features, and advantages will become more apparent through the following detailed description in connection with the accompanying drawings, and accordingly, those skilled in the art to which the present invention pertains can easily implement the technical spirit of the present invention. There will be. In addition, in the description of the present invention, when it is determined that the detailed description of the known technology related to the present invention may unnecessarily obscure the subject matter of the present invention, the detailed description will be omitted. Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a view showing a drone control system according to an embodiment of the present invention.

Referring to Figure 1, the drone control system according to an embodiment of the present invention includes a drone 100 and a drone control device 200, the drone 100 and the drone control device 200 is a mobile communication network 300 ) To communicate with each other. The mobile communication network 300 includes a Long Term Evolution (LTE) communication network and a fifth generation communication network, and can relay communication between the drone 100 and the drone control device 200.

Also, the mobile communication network 300 may include a base station, and the base station may provide GPS correction data to the drone 100 when establishing a radio link with the drone 100. The base station may check an error between its preset position and GPS coordinates currently received and measured from a plurality of satellites, and transmit the GPS error correction information capable of correcting the error to the drone 100.

The drone 100 includes a plurality of flight means such as a propeller, and by controlling the direction and rotational force of the flight means, it is possible to fly in the air in an intended direction or destination. The drone 100 is equipped with a GPS receiver, and it is possible to perform autonomous flight to the target location by comparing the current location and the target location through the GPS receiver. In addition, the drone 100 may be equipped with a camera, and may acquire a surrounding image through the camera and transmit it to the drone control device 200.

The drone 100 may be equipped with one or more sensors, such as a gyro sensor, wind direction and wind speed measurement sensor, and acceleration measurement sensor, and continuously collect flight data at predetermined cycle intervals using the sensor. The flight data includes current location information (ie, latitude, longitude and altitude) of the drone 100, attitude information (ie, roll, yaw and pitch), acceleration information, GPS information, battery information, distance remaining to the target point, weather Contains one or more of the information.

Here, the posture information includes measured values of roll, pitch, and yaw of the drone 100, and acceleration information includes X-axis acceleration, Y-axis acceleration, and Z-axis acceleration. In addition, the GPS information includes the number of satellites used to measure the speed and position of the drone 100 measured based on the GPS signal. In addition, the battery information includes the remaining battery power and the battery voltage, and the weather information includes the wind direction and wind speed. The drone 100 may transmit a plurality of flight data collected at the cycle intervals to the drone control device 200.

In addition, the drone 100 may receive flight route planning information from the drone control device 200, and the drone 100 may autonomously fly to a designated destination based on the flight route planning information. The flight route planning information includes continuous flight coordinates from the origin to the destination. As another embodiment, the drone 100 may receive only the destination coordinates from the drone control device 200, in this case, the drone 100 sets the current position as the starting position, and continuous coordinates from the starting position to the destination The flight route planning information recorded by itself can be generated, and the generated flight route planning information can be used to fly. The drone 100 may control the rotational speed and propulsion direction of a plurality of propellers and move to the route or destination direction.

On the other hand, the drone 100 stores a posture prediction function, receives a plurality of weights applied to the posture prediction function from the drone control apparatus 200 and applies it to the posture prediction function, and then recalls the current measured sensing value. By substituting the posture prediction function, it is possible to predict the future posture. In addition, the drone 100 may receive representative posture information for each speed used for pre-processing data from the drone control apparatus 200. When the predicted future posture is predicted to be an unstable posture, the drone 100 may perform in-situ flight at the current position without moving until it is predicted that the future posture is stabilized. The range of roll values representing the unstable posture, the range of pitch values and the range of yaw values are stored in the drone 100 in advance, and the drone 100 predicts the future posture of the drone 100 (ie, roll, pitch and yaw) Is compared with the range of the roll value indicating the unstable posture, the range of the pitch value and the range of the yaw value, and the drone 100 can determine whether the posture is unstable in the next cycle.

The drone control device 200 is a device capable of performing data communication, and a computing device such as a server may be employed. The drone control apparatus 200 may receive image data and status data from the drone 100, and may also control the drone 100 by transmitting a control command to the drone 100. The drone control device 200 may communicate with the drone 100 using a mobile communication network in which an LTE network including an EUTRAN (Evolved Universal Terrestrial Radio Access Network) is interlocked, and also a drone 100 using a fifth generation mobile communication network. ).

The drone control device 200 is equipped with a remote control interface for remotely controlling the drone 100. The drone control device 200 may acquire an image of the drone 100 in real time through the remote control interface, monitor the drone status, and also transmit a remote control command to the drone 100. The drone control apparatus 200 may transmit the destination coordinate information or flight route planning information to the drone 100, thereby moving the drone 100 to the designated place in the air.

In addition, the drone control apparatus 200 may collect and store flight data from the drone 100 and optimize each weight applied to the posture prediction function through machine learning as described below.

2 is a view showing the configuration of a drone control device according to an embodiment of the present invention.

2, the drone control device 200 includes a communication unit 210, a drone control unit 220, a data collection unit 230, a data pre-processing unit 240, a learning unit 250, and a database 260. Including, these components may be implemented in hardware or software, or through a combination of hardware and software.

In addition, the drone control apparatus 200 may include one or more processors and memory, and the functions of the drone control unit 220, the data collection unit 230, the data preprocessing unit 240, and the learning unit 250 are It may be mounted in the memory in the form of a program executed by the processor.

The communication unit 210 performs a function of performing communication with the drone 100 using a communication network such as a mobile communication network 300 or a wired communication network.

The database 260 is a storage device such as a storage device, a disk device, a tape device, and the like, and stores and stores a plurality of flight data. In addition, the database 260 may store status data received from each drone, and may store image data received from the drone 100.

The data collection unit 230 performs a function of collecting data from each drone 100. The data collection unit 230 receives image data from the drone 100 and stores it in the database 260. In particular, the data collection unit 230 collects flight data from a plurality of drones 100 and stores them in the database 260.

The data pre-processing unit 240 performs a function of pre-processing flight data stored in the database 260 when machine learning for optimizing the attitude prediction function is performed. Specifically, the data pre-processing unit 240 analyzes attitude information and speed information included in the flight data, and selects representative attitude information for each speed. The data pre-processing unit 240 analyzes attitude information and speed information included in each flight data, and selects roll values, yaw values, and pitch values having the most frequency at a specific speed as representative attitude information of the corresponding speed, thereby angular speed Star representative posture information can be selected. That is, the data pre-processing unit 240 analyzes attitude information and speed information included in a large amount of flight data, and selects roll values, yaw values, and pitch values representing the most frequent frequencies at a specific speed as representative attitude information at the speed. . As another embodiment, the data pre-processing unit 240 calculates the average of the roll value, the yaw value, and the pitch value at the corresponding speed, and represents the average value of the calculated roll value, the average value of the yaw value, and the average value of the pitch value at the speed. By selecting the posture information, representative posture information for each speed can be selected.

In addition, when the representative posture information for each speed is selected, the data preprocessing unit 240 checks the speed information and posture information included in the flight data, checks the representative posture information corresponding to the corresponding speed, and displays the representative posture information (that is, , Roll, yaw and pitch) and posture information included in flight data (ie, roll, yaw and pitch) to calculate the difference in posture, and preprocess each flight data. By the data pre-processing, the difference in attitude between the attitude information included in each flight data and the corresponding representative attitude information is calculated by the number of flight data.

The learning unit 250 trains the attitude prediction learning by repeatedly applying the flight data stored in the database 260 and the attitude difference calculated by the data pre-processing unit 240 to the attitude prediction function, and weights applied to the attitude prediction function It performs the function of optimizing. That is, the learning unit 250 applies posture information (ie, roll, pitch, and yaw) of the next cycle as an output layer, and applies flight data and preprocessing data as input values, thereby applying a plurality of weights applied to the posture prediction function Can be derived. At this time, the learning unit 250 uses a regression linear analysis model used in an artificial neural network, and inputs a result value (that is, a drone posture of the next cycle) and an input value into a posture prediction function to internally apply variables (input values). ) Can be optimized.

<Posture prediction function>

Here,'(location information)t' is the latitude, longitude, and altitude measured in the t period, and'(posture information)t' is the roll, pitch, and yaw measured in the t period. ,'(Acceleration information)t' is the X-axis acceleration, Y-axis acceleration, and Z-axis acceleration measured at t period. In addition,'(GPS information)t' is the number of satellites used to measure the drone speed in t period and the location in t period measured based on the GPS signal, and'(battery information)t' is the battery measured in t period Remaining amount and battery voltage, and'(posture difference)t' is the difference between the representative posture information calculated by pre-processing and the posture information measured at t period. Here, as representative attitude information, representative attitude information corresponding to the drone speed in the t period is used. In addition,'(remaining distance)t' is the distance remaining to the target point in the t period, and'(weather information)t' is the wind direction and wind speed measured in the t period. In addition,'(posture information)t+1' is the attitude information (ie, roll, pitch, and yaw) of the drone measured in the next period based on the t period.

As described above, as the input value, the difference between the sensed data measured in the t period and the pre-processed attitude difference based on the t period is substituted, and the drone attitude information of t+1 period is applied as the output value.

A regression model having a basis function (φ(x)) may be applied to the posture prediction function. The base function (φ(x)) may be a polynomial or a Gaussian function as follows.

Here, w _xo , w _yo , and w _{zo are} biases, and g _x , g _y , and g _z are forms in which the output values are divided into coordinate X, Y, and Y components. X _in is the input to the x component, Y _in is the input to the Y component, and Z _in is the input to the Z component. Incidentally, in X _in , the difference between the roll value of the drone measured _in the roll and t periods and the roll value included in the representative attitude information at the corresponding speed is used as an input value. In Y _in , the difference between the pitch value of the drone measured _in pitch and t period and the pitch value included in the representative attitude information at the corresponding speed is input. In addition, Z _in is the difference between the yogap included in the representative position information in the yaw (Yaw) and yogap of the drone and the measured speed from the period t is used as an input value. In addition, W is a weight, W _x is a weight applied to the x component, W _y is a weight input to the y component, and W _z is a weight input to the z component. The weight applied to the posture prediction function is optimized through repetitive learning of the learning unit 250, and the drone 100 acquires the weight and applies it to the posture prediction function that is being stored. When machine learning is repeatedly performed through the learning unit 250, the weight applied to the posture prediction function is not fixed, and converges to an optimal weight as the learning process is repeated.

The drone control unit 220 monitors the state of the drone through the communication unit 210 and performs a function of controlling the drone. Specifically, the drone control unit 220 may obtain and analyze the status information (eg, GPS coordinate information, various sensing information, etc.) of the drone 100 using the communication unit 210 to grasp the status of the drone. In addition, the drone control unit 220 receives the image from the drone 100, and stores it in the database 260. The drone control unit 220 communicates with the drone 100 through the communication unit 210 to continuously grasp the current location of the drone 100, but when a destination coordinate is input from an administrator, the risk stored in the database 260 In order to prevent the drone 100 from flying in the area, flight path planning information in which the current position of the drone 100 and sequential coordinates to the destination are recorded is generated, and the flight path planning information is transmitted to the drone 100 , The drone 100 is controlled to fly for a designated purpose.

3 is a flowchart illustrating a method of collecting and storing flight data in a drone control device according to an embodiment of the present invention.

Referring to FIG. 3, the drone control unit 220 checks the current position of the drone 100 using the communication unit 210 (S301). Subsequently, the drone control unit 220 receives the destination coordinates of the drone from the manager (S303), so that the drone 100 does not fly in the dangerous area stored in the database 260, and the current position of the drone 100 and the Flight route planning information in which sequential coordinates to a destination are recorded is generated (S305).

The drone control unit 220 transmits the flight route planning information to the drone 100 using the communication unit 210 to control the drone 100 to fly at the destination coordinates based on the flight route planning information. (S307).

When the flight route planning information is transmitted to the drone 100, the drone 100 collects flight data at predetermined intervals and transmits it to the drone control device 200, and the drone control device 200 transmits the flight data Is stored in the database 260 and stored (S309). The flight route planning information includes the current position information (ie, latitude, longitude and altitude) of the drone, attitude information (ie, roll, pitch and yaw) of the drone, acceleration information of the drone, GPS information, battery information, and remaining target points Distance and weather information is included.

4 is a flowchart illustrating a method of learning a posture prediction function using a large amount of flight data in a drone control apparatus according to an embodiment of the present invention.

Referring to FIG. 4, when a machine learning cycle arrives in a state where flight data is accumulated in a database over a predetermined number, the data pre-processor 240 extracts a plurality of stored flight data from the database 260 to accumulate. (S401). Subsequently, the data pre-processing unit 240 analyzes attitude information and speed information included in each of the extracted flight data, and selects roll values, yaw values, and pitch values having the most frequency at a specific speed as representative attitude information of the corresponding speed. By doing so, representative attitude information for each speed is selected (S403).

Next, the data pre-processing unit 240 checks the speed information and attitude information included in the flight data, checks the representative attitude information corresponding to the speed information, and the representative attitude information (that is, roll, yaw, and pitch) And posture differences between the posture information (that is, roll, yaw and pitch) included in the flight data, and preprocess each flight data (S405). Accordingly, the posture difference by the number of flight data is calculated by the data pre-processing unit 240.

When the data pre-processing is completed, the learning unit 250 learns the posture prediction learning by repeatedly applying the flight data stored in the database 260 and the posture difference calculated by the data pre-processing unit 240 to the posture prediction function. The learning proceeds (S407). The learning unit 250, as an input value, position information (ie, latitude, longitude, and altitude) of the drone of the t period, attitude information of the drone of the t period (ie, yaw, roll, and pitch), acceleration of the drone of the t period Information (i.e., x-axis acceleration, y-axis acceleration and z-axis acceleration), GPS information of t period (i.e., the number of satellites used to measure drone speed and position based on GPS signals), battery information of t period ( Battery remaining and battery voltage), pre-processed difference in attitude of t period, distance from t period to target point, and weather information of t period (i.e. wind direction and wind speed) are applied as input layer, and at t+1 period The drone posture information (ie, roll, pitch and yaw) is applied as the output layer.

If the repetitive learning is continuously performed, each weight of the posture prediction function is adjusted to converge to an optimal weight (S409).

The weights of the movement attitude prediction function adjusted by the learning unit 250 are provided to the drone 100, and the weights may be applied to the movement attitude prediction function held by the drone 100. In addition, representative attitude information for each speed selected by the dacher pre-processing unit 240 is provided to the drone 100.

5 is a view showing a drone configuration according to an embodiment of the present invention.

As shown in FIG. 5, the drone 100 includes a wireless communication unit 110, a driving unit 220, a GPS receiver 130, a storage unit 140, a sensing unit 150, and a control unit 160, These components can communicate with each other through a bus line.

The wireless communication unit 110 is a communication module capable of performing wireless communication with a base station of the mobile communication network 300, and communicates with the drone control device 200 via the mobile communication network 300. The wireless communication unit 210 transmits flight data to the drone control device 200 and also receives a plurality of weights applied to the attitude prediction function from the drone control device 200. In addition, the wireless communication unit 110 may receive representative attitude information for each speed from the drone control device 200.

The driving unit 220 performs a function of driving a plurality of propellers provided in the drone 100. The driving unit 220 drives the propeller of the drone 100 based on the motion control signal received from the control unit 160. The driving unit 220 may control the propeller rotation speed differently for each propeller.

The GPS receiver 130 performs a function of receiving a satellite signal detected in the surroundings.

The storage unit 140 is a storage means such as a memory, and stores various data necessary for the operation of the drone 100. In particular, the storage unit 140 stores a posture prediction function. In addition, the storage unit 140 stores representative attitude information for each speed, and stores a range of roll values, a range of pitch values, and a range of yaw values representing unstable attitudes.

The sensing unit 150 includes a gyro sensor 151, an acceleration sensor 152, and a wind speed sensor 153, and the sensing unit 150 uses a gyro sensor 151 and an acceleration sensor 152 to drone ( The yaw, pitch and roll of 100) are measured. In addition, the sensing unit 150 may measure the X-axis acceleration, the Y-axis acceleration, and the Z-axis acceleration of the drone 100 using the gyro sensor 151 and the acceleration sensor 152, respectively. In addition, the sensing unit 150 may measure the wind direction and wind speed around the drone 100 using the wind direction wind speed sensor 153.

The control unit 160, as a control means such as a microprocessor, controls the state of the drone 100, and also collects flight data at regular intervals, and then collects the flight data through the drone control device through the wireless communication unit 110 (200). At this time, the control unit 160 acquires the location information of the current drone through the GPS receiver 130 and checks the GPS information including the number of satellites and the speed of the drone. In addition, the control unit 160 checks the posture information including the roll, yaw, and pitch of the drone 100 through the sensing unit 150, and also includes acceleration of the X, Y, and Z axes of the drone 100. Check the acceleration information, and check the wind direction and wind speed around the drone. In addition, the controller 160 checks battery information including the voltage and the remaining amount of the battery, and checks the remaining distance from the current location to the destination.

The controller 160 includes drone position information, posture information, acceleration information, GPS information, battery information, remaining distance, and weather information in the flight data, and periodically transmits the flight data to the drone control device 200 .

The control unit 160 may store flight route planning information in the storage unit 140. At this time, the control unit 160 may receive the flight route planning information from the drone control device 200 through the wireless communication unit 110 and store the flight route planning information in the storage 140. As another embodiment, the control unit 160 may receive only the destination coordinates from the drone control apparatus 200, in this case, the control unit 160 sets the current location as the starting location, and continuous coordinates from the starting location to the destination The flight route planning information recorded by itself can be generated, and the generated flight route planning information can be used to fly.

The controller 160 controls the driving unit 220 so that the drone 100 reaches the destination by determining the current location through the GPS receiver 130 and comparing the current location with the location described in the flight route planning information. Can. In addition, the control unit 160 may receive a plurality of weights applied to the posture prediction function through the wireless communication unit 110, in which case the control unit 160 receives the posture prediction function stored in the storage 140. Apply one weight.

The controller 160 may predict a drone posture of the next cycle (ie, the future) using a weighted posture prediction function and selectively perform in-flight flight according to the predicted result.

6 is a flowchart illustrating a method for predicting a future posture based on a posture prediction function in a drone according to an embodiment of the present invention.

Referring to FIG. 6, the control unit 160 of the drone 100 controls the weights applied to the attitude prediction function, representative attitude information for each speed, and target point coordinates (or flight paths) through the wireless communication unit 110. Received from (200) and stored in the storage 140 (S601). Then, the controller 160 applies the received weights to the posture prediction function stored in the storage 140 (S603). If the weight is already applied to the attitude prediction function of the storage unit 140, the control unit 160 updates the weight of the attitude prediction function of the storage unit 140 with the weight received from the drone control apparatus 200.

Subsequently, the controller 160 generates a flight path based on the current location and the distance to the target point coordinates, and controls the driving unit 120 to perform autonomous flight so that the drone 100 moves to the target point (S605).

Next, when the autonomous flight starts, the control unit 160 collects data input to the attitude prediction function using the sensing unit 150 and the GPS receiver 130 (S607). Specifically, the control unit 160, based on the data obtained through the sensing unit 150 and the GPS receiver 130, the position information (ie, latitude, longitude and altitude) of the drone measured in the current cycle, the attitude information of the drone (Ie yaw, roll and pitch), drone acceleration information (i.e., x-axis acceleration, y-axis acceleration and z-axis acceleration), GPS information (i.e., the number of satellites used to measure drone speed and position based on GPS) ), battery information (battery level and battery voltage), and weather information (ie wind direction and wind speed) are collected. In addition, the control unit 160 checks the remaining distance to the target point, and also checks representative attitude information corresponding to the speed of the current drone 100 in the storage unit 140, and confirms the representative attitude information (that is, rolls). , Pitch and yaw) and the posture difference between the currently measured drone posture information (ie, roll, pitch and yaw) is calculated (S609).

In addition, the control unit 160 applies the calculated attitude difference, distance remaining to the target point, position information of the drone 100, attitude information of the drone, acceleration information, GPS information, battery information, and weather information to a weighted attitude prediction function. Substituting, the posture (ie, output value) of the drone of the next cycle (ie, the future) is predicted (S611).

The control unit 160 checks a posture range (ie, roll range, yaw range, and pitch range) indicating an unstable posture in the storage 140 and determines whether the predicted future drone posture is included in the posture range Then, it is determined whether the future drone posture becomes an unstable posture (S613).

When the posture of the drone 100 predicted as a result of the determination is an unstable posture, the controller 160 may temporarily stop moving to the target point and control the driving unit 220 to perform in-situ flight (S615). At this time, the control unit 160 maintains the current altitude as it is, or maintains a preset safety altitude, and performs in-flight flight in the air so that the latitude/longitude does not change. That is, if the attitude of the drone is predicted to be unstable in the next cycle, the control unit 160 performs the in-flight flight without moving to the target point until the drone attitude of the next cycle is predicted to be stable. Next, the controller 160 proceeds to step S607 again, collects data again, and substitutes the attitude prediction function, and when the drone attitude of the next cycle is predicted to be stabilized, step S605 to resume autonomous flight to the target point. To resume, the drone 100 is flying to a destination.

The process of steps S605 to S615 of FIG. 6 proceeds until the drone 100 reaches the destination.

While this specification includes many features, such features should not be construed as limiting the scope of the invention or the claims. Also, features described in individual embodiments herein may be implemented in combination in a single embodiment. Conversely, various features that are described in a single embodiment herein may be implemented in various embodiments individually or in combination as appropriate.

Although the operations in the drawings have been described in a specific order, it should not be understood that such operations are performed in a specific order as shown, or a series of sequences, or all described actions are performed to obtain a desired result. . In certain circumstances, multitasking and parallel processing may be advantageous. In addition, it should be understood that the division of various system components in the above-described embodiment does not require such division in all embodiments. The above-described program components and systems may be generally implemented as a package in a single software product or multiple software products.

The method of the present invention as described above may be implemented as a program and stored in a computer-readable form on a recording medium (CD-ROM, RAM, ROM, floppy disk, hard disk, magneto-optical disk, etc.). This process will not be described in detail any more as those skilled in the art to which the present invention pertains can easily carry out.

The present invention described above, as the person skilled in the art to which the present invention pertains, various substitutions, modifications and changes are possible within the scope of the technical spirit of the present invention, and the above-described embodiments and attached It is not limited by the drawings.

100: drone 200: drone control device
210: communication unit 220: drone control unit
230: data collection unit 240: data pre-processing unit
250: learning unit 260: database
110: wireless communication unit 120: drive unit
130: GPS receiver 140: storage unit
150: sensing unit 151: gyro sensor
152: acceleration sensor 153: wind speed sensor
160: control unit

Claims (15)

In the drone control system, as a method for controlling drones flying over long distances,
The drone control apparatus receives, accumulates and stores flight data from the drone;
Pre-processing the flight data by calculating a difference in attitude between the attitude information included in the stored flight data and the representative attitude information by the drone control device;
The drone control apparatus repeatedly learning the posture prediction function by applying the posture difference and the flight data to a posture prediction function to converge the weights applied to the posture prediction function; And
And transmitting, by the drone control device, the converged weights to a drone. According to claim 1,
After the step of transmitting,
The drone control method comprising the step of: applying the weights to a posture prediction function in which the weights are stored, and predicting a future drone posture based on the posture prediction function to which the weights are applied. According to claim 2,
The drone is characterized in that it comprises a; if the future drone posture confirmed based on the posture prediction function is predicted to be unstable, performing a flight in place until the future drone posture is predicted to be stable. Control method. According to claim 1,
The drone control device, selecting a representative posture information for each speed; further comprises,
The step of pre-processing the flight data checks the speed included in the flight data, and calculates a difference between the representative attitude information corresponding to the speed and the attitude information included in the flight data. According to claim 4,
The selecting step,
Drone control method, characterized in that by analyzing the stored and stored flight data, the roll value, yaw value and pitch value having the most frequency at the speed are selected as representative attitude information of the speed. According to claim 1,
The step of converging the weights,
The drone control apparatus applies the attitude difference and the flight data as input values to the attitude prediction function, checks attitude information from flight data collected after flight data applied as the input value, and retrieves this attitude information. Drone control method characterized in that it is applied as a result value to a posture prediction function The method of claim 6,
The flight data includes one or more of the position information of the drone, posture information, acceleration information, GPS information, battery information, a distance remaining to a target point, and weather information. In the drone control device for controlling drones flying at a long distance,
A database storing drone flight data;
A data pre-processor for calculating posture differences between posture information and representative posture information included in the flight data and preprocessing the flight data; And
And a learning unit that repeatedly learns the posture prediction function by applying the posture difference and the flight data to a posture prediction function to converge weights applied to the posture prediction function. The method of claim 8,
And a drone control unit that transmits the converged weights to the drone and controls the weights to be applied to a posture prediction function embedded in the drone itself. The method of claim 8,
The data pre-processing unit,
After selecting representative attitude information for each speed, check the speed included in the flight data, and calculate a difference between the representative attitude information corresponding to the speed and the attitude information included in the flight data. The method of claim 10,
The data pre-processing unit,
Drone control device characterized in that by analyzing the accumulated and stored flight data, the roll value, yaw value and pitch value having the most frequency at the speed are selected as representative attitude information of the speed. The method of claim 8,
The learning unit,
The posture difference and the flight data are applied to the posture prediction function as an input value, and posture information is checked from the flight data collected after the flight data applied as the input value, and this posture information is output to the posture prediction function. Drone control device, characterized in that applied to. For drones that can predict the future location,
A storage unit for storing a posture prediction function;
A wireless communication unit wirelessly communicating with a drone control device to receive weights from the drone control device; And
After applying the received weights to the attitude prediction function, collecting flight data, calculating a difference in attitude between the drone's attitude information and representative attitude information, and applying the calculated attitude difference and the collected flight data to the weights A drone comprising; a controller for inputting to a posture prediction function and predicting a position of a future drone represented in a next cycle. The method of claim 13,
The storage unit stores representative posture information for each speed,
The control unit, the drone characterized in that for calculating the difference between the attitude of the representative posture information and the attitude information of the drone corresponding to the speed of the drone. The method of claim 13,
The control unit,
A drone characterized in that if the future drone attitude confirmed based on the attitude prediction function is predicted to be unstable, the drone is controlled to perform in-flight flight until the future drone attitude is predicted to be stable.

Images