KR20220143621A - Unmanned aerial vehicle and method for moving obstacle collision avoidance - Google Patents

Abstract

The present invention relates to an unmanned aerial vehicle capable of moving obstacle automatic avoidance and an avoidance method thereof. The unmanned aerial vehicle according to one embodiment of the present invention may comprise: an obstacle detection unit which detects a first obstacle by a predetermined method to generate a first sphere including the detected first obstacle, and generates first obstacle information including at least one of the diameter and the radius of the first sphere and the central point of the first sphere; and a collision avoidance processing unit which uses the first obstacle information sequentially inputted to expect a first moving route of the first obstacle, extracts the n number of candidate aiming points corresponding to the first moving route, detects the mth candidate aiming point which has a higher order of priority among the n number of candidate aiming points, and determines collision possibility with the first obstacle while moving to the mth candidate aiming point. Therefore, the moving obstacle may be detected and collision thereagainst may be avoided.

Description

Dynamic obstacle collision avoidance drone and its avoidance method

The present invention relates to an unmanned aerial vehicle capable of automatic collision avoidance of dynamic obstacles and a method for avoiding the same.

In general, an aircraft refers to a machine that can acquire bearing capacity in the atmosphere by the reaction of air against the ground. Among aircraft, an unmanned aerial vehicle (UAV) is an aircraft that does not have a pilot on board, and has recently attracted a lot of attention not only in the military field but also in the civilian field. Representative UAVs include quadrotor and hexacopter, which are types of rotary wing UAVs capable of vertical take-off and landing.

UAVs can efficiently and effectively perform various missions in hazardous environments such as battlefields and disaster areas, and are widely used for missions such as surveillance, reconnaissance, exploration, and transportation. In performing these tasks, it is required to be close to various obstacles including natural and man-made structures, and in order to successfully perform a given task, it is necessary to avoid colliding with these obstacles, For this purpose, it is required to detect and avoid obstacles from the unmanned aerial vehicle.

There are sensors developed and used for UAVs for this purpose. Sensors can be classified based on whether the energy they sense is artificial or natural. Among them, the active sensor operates by generating energy into the environment, and then measures the reflection of the beam. For example, radar, laser radar, lidar, etc. there may be

Korean Patent Laid-Open Patent No. 10-2013-0037697 (System and Method for Collision Prevention of Unmanned Aircraft) states that automatically moving to the next flight stop based on a two-dimensional electronic map, such as mountains, buildings, etc. In this case, there is disclosed a technology for notifying the pilot of the risk of collision during automatic flight in advance to correct the altitude or change the waypoint. However, in this case, an obstacle is detected using an ultrasonic sensor without using image information, and the pilot is warned in advance to modify the altitude or flight path of the unmanned aerial vehicle moving to the next destination. .

Recently, active sensors such as lidar have been widely installed on unmanned aerial vehicles due to their low cost, and thus a method for detecting dynamic obstacles and avoiding collisions using them is required.

The above-mentioned background art is technical information possessed by the inventor for the derivation of the present invention or acquired in the process of derivation of the present invention, and cannot necessarily be said to be a known technique disclosed to the general public prior to the filing of the present invention.

Korean Patent Publication No. 10-2015-0136209

SUMMARY OF THE INVENTION The present invention is to solve the above problems, and an object of the present invention is to provide an obstacle avoiding unmanned aerial vehicle capable of detecting and avoiding collisions using an active sensor and a method of avoiding the same.

According to an embodiment of the present invention, a first obstacle is detected according to a preset method, a first sphere including the detected first obstacle is generated, and at least a diameter or a radius of the first sphere is detected. an obstacle detector for generating first obstacle information including one and a center point of the first sphere; and predicting a first movement path of the first obstacle using sequentially input first obstacle information, extracting n candidate aiming points corresponding to the first moving path, and prioritizing the n candidate aiming points A collision avoidance processing unit that detects an m-th candidate aiming point having a high An unmanned aerial vehicle, which is a natural number less than or equal to n, is disclosed.

According to an embodiment, the obstacle detecting unit generates the first obstacle information in consideration of a detection time and an obstacle moving distance before a preset condition is satisfied after the first obstacle is detected, and the collision avoidance processing unit comprises: The first movement path of the first obstacle may be tracked using the first obstacle information input before the condition is satisfied, and the prediction may be performed using the tracked first movement path after the condition is satisfied.

According to an embodiment, the obstacle detection unit detects the first obstacle by clustering information input from the provided active sensor according to a preset method after the condition is satisfied, and the collision avoidance processing unit includes the condition After this is satisfied, the prediction may be performed using a preset Kalman filter.

According to an embodiment, the condition corresponds to a preset initial time, but the initial time may be calculated from a moment when the first obstacle is first detected.

According to an embodiment, the collision avoidance processing unit may generate a pipe corresponding to the prediction and extract the n candidate aiming points included in the pipe.

According to an embodiment, the collision avoidance processing unit may detect, as the mth candidate aiming point, the closest distance to a preset target point among the n candidate aiming points.

According to an embodiment, the collision avoidance processing unit may determine that there is a possibility of collision with the first obstacle if there is a case where the distance to the first obstacle is less than or equal to the radius while moving to the mth candidate aiming point.

According to an embodiment, the collision avoidance processing unit detects, as the p-th candidate aiming point, the closest distance to a preset target point among the n candidate aiming points except for the m-th candidate aiming point, and as the p-th candidate aiming point. A possibility of colliding with the first obstacle is determined while moving, wherein p may be a natural number different from m among natural numbers less than or equal to n.

According to an embodiment, the obstacle detecting unit detects a second obstacle according to the preset method, generates a second sphere including the detected second obstacle, and a diameter or radius of the second sphere generates second obstacle information including at least one of at least one and the center point of the first sphere, and the collision avoidance processing unit predicts a second movement path of the second obstacle by using the sequentially input second obstacle information extracting q candidate aiming points corresponding to the first and second moving paths, detecting the a-th candidate aiming point having a higher priority among the q candidate aiming points, and moving to the a-th candidate aiming point While determining the possibility of collision with the first obstacle and the second obstacle, q may be a natural number greater than n, and a may be a natural number less than or equal to q.

According to an embodiment, the obstacle detector generates the first obstacle information in consideration of the detection time and the obstacle movement distance before a preset condition is satisfied after the first obstacle is detected, and the second obstacle is detected. before the condition is satisfied, the second obstacle information is generated in consideration of the detection time and the obstacle movement distance, and the collision avoidance processing unit is configured to generate the second obstacle information using the first obstacle information input before the condition is satisfied. Tracks a first movement path of a first obstacle, predicts the first movement path using the tracked first movement path after the condition is satisfied, and uses the second obstacle information input before the condition is satisfied 2 Track the second movement path of the obstacle, and predict the second movement path using the tracked second movement path after the condition is satisfied, wherein the condition corresponds to a preset initial time, and the initial time is the The first obstacle may be counted from the moment it is first detected.

According to an embodiment, the obstacle detecting unit detects at least one of the first obstacle and the second obstacle by clustering information input from the provided active sensor according to a preset method after the condition is satisfied, After the condition is satisfied, the collision avoidance processing unit may predict at least one of the first movement path and the second movement path using a preset Kalman filter.

According to an embodiment, the collision avoidance processing unit generates a first pipe corresponding to the prediction of the first movement path, generates a second pipe corresponding to the prediction of the second movement path, and the The q candidate aiming points included in the first pipe and the second pipe may be extracted.

According to an embodiment, the collision avoidance processing unit may detect, as the a-th candidate aiming point, the closest distance to a preset target point among the q candidate aiming points.

According to an embodiment, the collision avoidance processing unit may include, while moving to the a-th candidate aiming point, when the distance to the first obstacle is less than or equal to the radius of the first sphere or the distance to the second obstacle is the second sphere's If the radius or less exists, it may be determined that there is a possibility of collision with the first obstacle or the second obstacle.

According to an embodiment, the collision avoidance processing unit is configured to detect, as the bth candidate aiming point, the closest distance to a preset target point among the q candidate aiming points except for the a-th candidate aiming point, and the b-th candidate aiming point. A possibility of collision with the first obstacle and the second obstacle is determined while moving to , wherein b may be a natural number different from a among natural numbers less than or equal to q.

The present invention can detect a dynamic obstacle using an active sensor and automatically avoid collision with the dynamic obstacle.

1 is a block diagram of an obstacle avoiding unmanned aerial vehicle according to an embodiment of the present invention.
2 to 4 are diagrams for explaining an operation of detecting and avoiding obstacles of an unmanned aerial vehicle according to an embodiment of the present invention.
5 to 10 are diagrams for comparing and explaining an obstacle avoidance simulation result of an unmanned aerial vehicle according to an embodiment of the present invention.
11 is a flowchart of a method for detecting and avoiding an obstacle according to another embodiment of the present invention.

Specific structural or functional descriptions of the embodiments according to the concept of the present invention disclosed in this specification are only exemplified for the purpose of explaining the embodiments according to the concept of the present invention, and the embodiments according to the concept of the present invention are It may be implemented in various forms and is not limited to the embodiments described herein.

Since the embodiments according to the concept of the present invention may have various changes and may have various forms, the embodiments will be illustrated in the drawings and described in detail herein. However, this is not intended to limit the embodiments according to the concept of the present invention to specific disclosed forms, and includes changes, equivalents or substitutes included in the spirit and scope of the present invention.

Terms such as first or second may be used to describe various elements, but the elements should not be limited by the terms. The above terms are used only for the purpose of distinguishing one component from other components, for example, without departing from the scope of rights according to the concept of the present invention, a first component may be named as a second component, Similarly, the second component may also be referred to as the first component.

When an element is referred to as being “connected” or “connected” to another element, it is understood that it may be directly connected or connected to the other element, but other elements may exist in between. it should be On the other hand, when it is mentioned that a certain element is "directly connected" or "directly connected" to another element, it should be understood that there is no other element in the middle. Expressions describing the relationship between components, for example, “between” and “between” or “neighboring to” and “directly adjacent to”, should be interpreted similarly.

The terms used herein are used only to describe specific embodiments, and are not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly dictates otherwise. In the present specification, terms such as “comprise” or “have” are intended to designate the presence of the specified feature, number, step, operation, component, part, or combination thereof, but one or more other features, number, or step , it should be understood that it does not preclude the possibility of the existence or addition of , operation, components, parts or combinations thereof.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Terms such as those defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present specification. does not

Hereinafter, an obstacle avoiding unmanned aerial vehicle and a method for avoiding the same according to an embodiment of the present invention will be described with reference to the accompanying drawings.

1 is a block diagram of an obstacle avoiding unmanned aerial vehicle according to an embodiment of the present invention, and FIGS. 2 to 4 are diagrams for explaining an obstacle detection and avoiding operation of the unmanned aerial vehicle according to an embodiment of the present invention.

Referring to FIG. 1 , an unmanned aerial vehicle 100 according to an embodiment of the present invention may include an active sensor unit 110 , an obstacle detection unit 120 , a collision avoidance processing unit 130 , and an unmanned aerial vehicle starting unit 140 . have.

The unmanned aerial vehicle starter 140 may operate the rotor according to the operation mode of the unmanned aerial vehicle 100 to change the position and posture of the unmanned aerial vehicle 100 to start the unmanned aerial vehicle 100 . The operation mode according to an embodiment of the present invention may include a target point starting mode and a collision avoidance mode. The target point maneuvering mode may be an operation mode in which the vehicle automatically navigates in an optimal path toward a preset target point. The collision avoidance mode may be an operation mode when an obstacle is detected in the navigation mode and a collision is expected. Hereinafter, a case in which an obstacle is detected and the unmanned aerial vehicle 100 is operated in a collision avoidance mode will be described.

The active sensor unit 110 may generate information sensing a surrounding object through a provided sensor (eg, an active sensor such as lidar). Alternatively, the active sensor unit 110 may include a vision sensor, and may generate information sensing a surrounding object through a vision sensor or the like.

The obstacle detector 120 may detect an obstacle based on the information generated by the active sensor unit 110 . For example, the obstacle detection unit 120 may use the 3D information sensed by the active sensor unit 110 to detect a surrounding object, a topographical feature, and the like, and model it.

Here, the active sensor unit 110 and the obstacle detection unit 120 may be set to detect an obstacle when a preset event occurs. For example, the active sensor unit 110 may generate information sensing a surrounding object at a preset time period, and the obstacle detection unit 120 detects obstacles whenever information is input from the active sensor unit 110 . existence can be detected.

When an obstacle is detected, the obstacle detector 120 may generate a sphere including the obstacle. For example, the obstacle detector 120 may generate a sphere that is a virtual space including all detected obstacles, but may generate the sphere by applying a preset margin. That is, the center point of the generated sphere is the same as the center point of the detected obstacle, and the volume of the sphere may be 'by a preset amount' greater than the volume of the obstacle.

Referring to FIG. 2 , the obstacle detecting unit 120 of the unmanned aerial vehicle 100 may detect obstacles 210 - 1 and 210 - 2 by analyzing information sensed through provided sensors, and individual obstacles 210 - 1 and 220-2) while including all of the spheres 220-1 and 220-2 to which a preset margin is applied may be generated.

The obstacle detector 120 may generate obstacle information including a center point and a diameter (or radius) of the generated sphere. For example, when the first obstacle 210-1 is detected for the first time, the obstacle detector 120 may generate a first sphere 220-1 corresponding to the first obstacle 210-1 according to a preset method. can Also, the obstacle detector 120 may generate first obstacle information including a center point and a diameter (or radius) of the first sphere 220 - 1 . Also, when the second obstacle 210 - 2 is first detected, the obstacle detector 120 may generate a second sphere 220 - 2 corresponding to the second obstacle 210 - 2 according to a preset method. . Also, the obstacle detector 120 may generate second obstacle information including the center point and the diameter (or radius) of the second sphere 220 - 2 .

Meanwhile, the obstacle detector 120 may continuously detect an obstacle at a preset time period. That is, when a preset time elapses after first detecting the first obstacle 210-1, the obstacle detector 120 may detect the first obstacle 210-1 again. This may also be applied to the second obstacle 210 - 2 . Accordingly, the obstacle detection unit 120 is the first sphere 220-1 and/or the first obstacle information at the first time, the first sphere 220-1 and/or the first obstacle information at the second time. , the first sphere 220-1 and/or the first obstacle information at the x-th time may be respectively generated and output. Similarly, the obstacle detection unit 120 includes the second sphere 220-2 and/or the second obstacle information at the first time, the second sphere 220-2 and/or the second obstacle information at the second time, The second sphere 220-1 and/or second obstacle information at the x-th time may be generated and output, respectively. Here, x may be any natural number. Also, the x-th time may mean after a preset time has elapsed from the x-1 time.

An operation in which the active sensor unit 110 generates sensing information and the obstacle detector 120 detects an obstacle through the sensing information is disclosed by a number of prior arts such as Korean Patent Application Laid-Open No. 10-2015-0136209 Therefore, a more detailed description thereof will be omitted.

The collision avoidance processing unit 130 may anticipate the possibility of collision between the unmanned aerial vehicle 100 and the obstacles 210 - 1 and 210 - 2 , and when it is determined that there is a possibility of collision as a result of the prediction, it may create a new path point, a selective aiming point. In this case, the obstacles 210-1 and 210-2 may be dynamic obstacles. That is, the collision avoidance processing unit 130 predicts the movement path of the obstacles 210 - 1 and 210 - 2 , and the path and the obstacles 210 - 1 and 210 - to the preset movement target point 230 of the unmanned aerial vehicle 100 . The movement paths of 2) overlap, and whether there is a possibility of a collision may be determined in consideration of the speed of the unmanned aerial vehicle 100 and the speed of the obstacles 210 - 1 and 210 - 2 . Hereinafter, the collision avoidance operation of the obstacles 210 - 1 and 210 - 2 of the collision avoidance processing unit 130 will be described.

Referring to FIG. 3 , a 1-1 sphere 310-1 (shown by a dotted line) is generated in response to the first 'obstacle 210-1, and a 1-2 sphere 310-2 (in a solid line) is generated. A case in which the positions of the 1-3 spheres 310-3 (shown by a solid line), and the 1-4th spheres 310-4 (shown by a solid line) (shown by a solid line) is predicted is exemplified. Also, referring to FIG. 3 , a 2-1 sphere 320-1 (shown by a dotted line) is generated in response to the second obstacle 210-2, and a 2-2 sphere 320-2 (solid line) is generated. A case in which the positions of the 2-3 spheres 320-3 (shown by solid lines) are predicted is exemplified. That is, the collision avoidance processing unit 130 can predict that the first obstacle 210-1 is moving from the center point of the 1-1 sphere 310-1 to the center point of the 1-4 sphere 310-4. have. Similarly, the collision avoidance processing unit 130 can predict that the second obstacle 210-2 is moving from the center point of the 2-1 sphere 320-1 to the center point of the 2-3 sphere 320-3. have.

Each of the spheres shown in FIG. 3 is illustrated as a circle, but the circle is merely a circle corresponding to the moving direction of the obstacle among numerous circles constituting the surface of the sphere.

Here, the collision avoidance processing unit 130 analyzes one or more of the plurality of spheres corresponding to the first obstacle 210-1 (particularly by analyzing the center point of the spheres) to determine the state variable of the first obstacle 210-1. (position, velocity, acceleration, etc.) can be predicted. In this case, the collision avoidance processing unit 130 may use a Kalman filter. Similarly, the collision avoidance processing unit 130 analyzes one or more of the plurality of spheres corresponding to the second obstacle 210-2 (particularly by analyzing the center points of the spheres) to determine the state variable ( position, velocity, acceleration, etc.) can be predicted. Even in this case, the collision avoidance processing unit 130 may use a Kalman filter.

However, when the Kalman filter is used, since it takes time for the state variable estimates for the detected obstacles to converge to the normal state, the collision avoidance processing unit 130 performs the first obstacle 210-1 or the second obstacle 210- 2) is detected for the first time, and after a preset time has elapsed, the state variable of the first obstacle 210 - 1 or the second obstacle 210 - 2 may be predicted using the Kalman filter.

In addition, the collision avoidance processing unit 130 "tracks" the position of each obstacle 210-1, 210-2 before predicting the state variable using a Kalman filter, etc., and uses the tracked position to predict the state variable later. can For example, a case before the first obstacle 210-1 is recognized for the first time and a preset condition is satisfied is limited. At this time, the collision avoidance processing unit 130 may "track" the movement of the center point of the first obstacle 120-1 using the 1-1 obstacle information to the 1-y obstacle information (where y is a natural number). have. In addition, the collision avoidance processing unit 103 predicts the state variable of the Kalman filter later on the movement path (center point movement) of the first obstacle 120-1 tracked at the initial stage (that is, after the initial recognition and before the preset condition is satisfied) . is available for The method of tracking the first obstacle 210 - 1 of the collision avoidance processing unit 130 may be similarly applied to the second obstacle 210 - 2 .

On the other hand, the obstacle detection unit 120 may use a preset clustering technique when classifying obstacles using the information input from the active sensor unit 110. The clustering technique is not performed until a predetermined time elapses after the obstacle is detected for the first time. may not be stable. This is because, when an obstacle is detected for the first time, only a part of the shape of the obstacle may be detected, and over time, the entire shape may be detected. Accordingly, the obstacle detecting unit 120 detects the first obstacle 210-1 after the first obstacle 210-1 is first detected and before a preset condition is satisfied, the first obstacle 210-1 is detected. And/or the first obstacle information may be generated in consideration of moving distances of other obstacles.

For example, it is assumed that the active sensor unit 110 includes a vision sensor, and information output from the vision sensor corresponds to a point cloud (point cloud). In this case, the obstacle detector 120 may detect the obstacle by classifying the point cloud through a preset clustering technique (clustering, for example, a center-based clustering technique, a density-based clustering technique, etc.) . Therefore, the obstacle detection unit 120 considers the movement distance/speed and detection time of the other obstacles (hereinafter, abbreviated as 'obstacles') that have already been detected among the point clouds input by the active sensor unit 110, Point clouds that do not belong to the obstacle may be clustered and detected as the first obstacle 210-1.

In addition, since the shape of the first obstacle 210-1 may continue to be deformed for a certain period of time after the first obstacle 210-1 is detected for the first time (the shape of the first recognized obstacle 210-1) is not all), the obstacle detection unit 120 considers the moving distance, speed, and detection time of obstacles (including the newly detected first obstacle 210-1) until a preset condition is satisfied. It will be possible to cluster point clouds. For example, if points are newly detected while the obstacle is moving at speed A (however, A is a real number), it can be determined whether the new points can belong to the obstacle, and thereafter, the movement speed of all obstacles , it is possible to determine which obstacle the newly detected point belongs to in consideration of the detection time.

The obstacle detection method of the obstacle detector 120 may be similarly applied to the second obstacle 210 - 2 .

Here, the preset condition of the obstacle detection unit 120 and/or the collision avoidance processing unit 130 may correspond to the initial time. The initial time may be a time calculated from the moment the obstacle is first detected. For example, when the initial time is set to 1 second, it may be determined that the initial time is satisfied when 1 second has elapsed after the first obstacle 210-1 is first detected.

In addition, when a preset condition is satisfied (eg, when the initial time has elapsed), the collision avoidance processing unit 130 uses the sequentially input first obstacle information and second obstacle information to generate the first obstacle 210-1. ) and/or the movement path of the second obstacle 210 - 2 may be predicted. In this case, the Kalman filter may be used as described above.

In the example of FIG. 3 , the first obstacle recognized before the 1-1 sphere 310-1 and the points 310-11 connected thereto of the first recognized obstacle 210-1 are satisfied before a preset condition is satisfied. may correspond to information. In addition, the 1-2 spheres 310-2 to 1-4 spheres 310-4 refer to the movement paths of the first obstacle 210-1 predicted by the collision avoidance processing unit 130 . Similarly, the first 2-1 sphere 320-1 of the first recognized second obstacle 210-2 and the points 320-11 following it correspond to the second obstacle information recognized before the preset condition is satisfied, The 2-2 spheres 320 - 2 to 2 - 3 spheres 320 - 3 refer to the movement paths of the second obstacle 210 - 2 predicted by the collision avoidance processing unit 130 .

Also, the collision avoidance processing unit 130 may generate a first virtual pipe corresponding to the prediction of the movement path of the first obstacle 210-1. Similarly, the collision avoidance processing unit 130 may generate a virtual second pipe corresponding to the prediction of the movement path of the second obstacle 210 - 2 .

Referring to FIG. 4 , the diameter of the first pipe 410 is the same as the diameter of the first sphere, and the centerline of the first pipe 410 is the same as the centerline of the first obstacle 210-1. Also, a case where the diameter of the second pipe 420 is the same as the diameter of the second sphere and the centerline of the second pipe 420 is the same as the centerline of the second obstacle 210-2 is exemplified. Here, the center line of the first obstacle 210 - 1 may mean a virtual line connecting the predicted center points of the first obstacle 210 - 1 , and the center line of the second obstacle 210 - 2 is the predicted center line of the first obstacle 210 - 1 . It may mean a virtual line connecting the center points of the second obstacle 210 - 2 .

Thereafter, the collision avoidance processing unit 130 may extract points included in the first pipe 410 and/or points included in the second pipe 420 as candidate aiming points. When the drone 100 moves to a point included in the first pipe 410 or the second pipe 420 , there is a risk of colliding with the first obstacle 210-1 and/or the second obstacle 210-2. Because.

Also, the collision avoidance processing unit 130 may detect a candidate aiming point having the highest priority among the candidate aiming points as the mth candidate aiming point (where m is a natural number equal to or less than the total number of candidate aiming points). For example, the priority may correspond to the order closest to the target point 230 of the UAV 100 (ie, the point to which the UAV 100 was facing).

In addition, the collision avoidance processing unit 130 determines that the distance to the first obstacle 210-1 is less than or equal to the first radius 430 of the first sphere while the drone 100 goes to the m-th candidate aiming point at a preset speed. It may be determined whether there is a case or the distance to the second obstacle 210 - 2 is less than or equal to the second radius 440 of the second sphere. The collision avoidance processing unit 130 must determine the existence of the above-mentioned case over all the time that the unmanned aerial vehicle 100 goes to the mth candidate aiming point. This is because the unmanned aerial vehicle 100 may collide with any one of the obstacles before reaching the mth candidate aiming point.

As a result of the determination, the collision avoidance processing unit 130 sets the distance between the first obstacle 210-1 and the first radius 430 or less or the second obstacle 220-2 while the UAV 100 moves to the m-th candidate aiming point. ) and the second radius 440 or less, it is possible to read the p th candidate aiming point, which is the next priority. Here, p may be a natural number other than m among natural numbers less than or equal to the total number of candidate aiming points. The p-th candidate aiming point may be a candidate aiming point having a higher priority than the m-th candidate aiming point among all the candidate aiming points. For example, the p-th candidate aiming point may be a candidate aiming point having the closest distance to the target point 230 after the m-th candidate aiming point among all the candidate aiming points.

The collision avoidance processing unit 130 becomes the distance between the first obstacle 210-1 and the first radius 430 or less or the second obstacle 220-2 and the second obstacle while the UAV 100 moves to the p-th candidate aiming point. It may be determined whether there is a case where the distance is less than or equal to 2 radius 440 . As a result of the determination, if there is no such case, the collision avoidance processing unit 130 may designate the p-th candidate aiming point as the selective aiming point.

Thereafter, the unmanned aerial vehicle starter 140 may operate the rotor so that the unmanned aerial vehicle 100 faces a point corresponding to the selected aiming point. Accordingly, the unmanned aerial vehicle 100 may automatically and efficiently avoid the first obstacle 210-1 and the second obstacle 210-2.

In the above description, it is assumed that there are two obstacles, but the above case can be easily applied even when there are three or more obstacles. Therefore, it is obvious that the number of obstacles cannot limit the scope of the present invention.

5 to 10 are diagrams for comparing and explaining an obstacle avoidance simulation result of an unmanned aerial vehicle according to an embodiment of the present invention.

5 illustrates a movement path of the unmanned aerial vehicle 100 (UAV) and three dynamic obstacles according to an embodiment of the present invention. 6 to 8 show the Euler angle, angular velocity, speed, speed, cluster management, and minimum distance to obstacles of the unmanned aerial vehicle 100 according to an embodiment of the present invention.

When the flight starts, the unmanned aerial vehicle 100 may fly toward a target point and detect 0 obstacles (clusters). Thereafter, the unmanned aerial vehicle 100 may detect the first obstacle, and when a preset initial time has elapsed after detection, the Kalman filter may be applied to the first obstacle to generate a pipe. The pipe shown in FIG. 9 may correspond to the first obstacle.

Thereafter, the unmanned aerial vehicle 100 may detect the second obstacle, but cannot apply the Kalman filter until the initial time elapses after detecting the preset second obstacle. As shown in FIG. 9 , it can be seen that the pipe for the second obstacle is not yet created.

If the initial time elapses after detection of the second obstacle, a pipe to the second obstacle may also be created. Similarly, when the initial time elapses after detecting the third obstacle, a pipe for the third obstacle may also be created. As shown in FIG. 10 , it can be confirmed that pipes for the second obstacle and the third obstacle are also generated.

Thereafter, the unmanned aerial vehicle 100 may generate a selective aiming point after predicting the possibility of collision with all detected obstacles, and may efficiently avoid the obstacles by operating the rotor to face the selected aiming point.

11 is a flowchart of a method for detecting and avoiding an obstacle according to another embodiment of the present invention.

Each of the steps to be described below may be steps performed in each component of the UAV 100 described with reference to FIG. 1 , but for convenience of understanding and description, they will be collectively described as being performed in the UAV 100 .

In step S1110, the unmanned aerial vehicle 100 may detect one or more dynamic obstacles. The unmanned aerial vehicle 100 may include an active sensor such as a vision sensor and lidar, and may detect a dynamic obstacle through this.

In step S1120, the unmanned aerial vehicle 100 may shape the obstacle detected through the sensor into a sphere according to a preset method.

In step S1130, the unmanned aerial vehicle 100 may shape the sphere of the obstacle in consideration of the detection time, the speed, the distance, and the like of the obstacle for the initial preset time.

In step S1140 , when the initial time has elapsed, the unmanned aerial vehicle 100 may predict the movement path of the obstacle according to a preset method. For example, the unmanned aerial vehicle 100 may predict the movement path of the obstacle through the Kalman filter using the center point of the tracked obstacle.

In step S1150, the unmanned aerial vehicle 100 may extract n candidate aiming points included in the predicted movement path. When there are a plurality of obstacles, the drone 100 may shape a plurality of pipes corresponding to the prediction of the movement path of the obstacle, and extract candidate aiming points from n points included in the plurality of pipes (however, n is a natural number greater than or equal to 2)

In step S1160 , the unmanned aerial vehicle 100 may detect an mth candidate aiming point having the highest preset priority (eg, a rank having a close distance to the target point) among the n candidate aiming points.

In step S1170 , the unmanned aerial vehicle 100 may determine a distance to an individual dynamic obstacle during the entire time it moves from the current position to the mth candidate aiming point.

In step S1180 , as a result of the determination, if there is a case where the distance to the individual obstacle is less than or equal to the radius of the obstacle, the drone 100 may determine that there is a risk of collision with the obstacle when moving to the mth candidate aiming point. Accordingly, the unmanned aerial vehicle 100 may detect the p-th candidate aiming point corresponding to the preset next priority and re-perform steps S1160 to S1180.

In step S1170 , the unmanned aerial vehicle 100 may determine the mth candidate aiming point as the selective aiming point if there is no case in which the distance to the individual obstacle is less than or equal to the radius of the obstacle during the entire time of moving to the mth candidate aiming point. Thereafter, the unmanned aerial vehicle 100 may efficiently avoid obstacles by operating the rotor to face the selected aiming point.

It is to be understood that the above-described embodiments are illustrative in all respects and not restrictive, and the scope of the present invention will be indicated by the following claims rather than the foregoing detailed description. And it should be construed that all changes and modifications derived from the meaning and scope of the claims as well as equivalent concepts are included in the scope of the present invention.

100: drone
110: active sensor unit
120: obstacle detection unit
130: collision avoidance processing unit
140: unmanned aerial vehicle maneuvering unit

Claims (11)

A first obstacle is detected according to a preset method, a first sphere including the detected first obstacle is generated, and at least one of a diameter or a radius of the first sphere and a center point of the first sphere are determined. an obstacle detector for generating first obstacle information including; and
Using the sequentially input first obstacle information, the first movement path of the first obstacle is predicted, n candidate aiming points corresponding to the first moving path are extracted, and the priority among the n candidate aiming points is determined. a collision avoidance processing unit detecting a high m-th candidate aiming point and determining a possibility of collision with the first obstacle while moving to the m-th candidate aiming point;
including,
Wherein n is a natural number greater than or equal to 2, and m is a natural number less than or equal to n,
The obstacle detection unit,
After the first obstacle is detected and before a preset condition is satisfied, the first obstacle information is generated in consideration of a detection time and an obstacle movement distance,
After the condition is satisfied, the first obstacle is detected by clustering the information input from the provided active sensor according to a preset method,
The collision avoidance processing unit,
tracking the first movement path of the first obstacle using the first obstacle information input before the condition is satisfied, and performing the prediction using the tracked first movement path after the condition is satisfied;
After the condition is satisfied, the UAV performs the prediction using a preset Kalman filter.
According to claim 1,
The condition corresponds to a preset initial time,
The initial time is calculated from the moment the first obstacle is detected for the first time, the unmanned aerial vehicle.
A first obstacle is detected according to a preset method, a first sphere including the detected first obstacle is generated, and at least one of a diameter or a radius of the first sphere and a center point of the first sphere are determined. an obstacle detector for generating first obstacle information including; and
Using the sequentially input first obstacle information, the first movement path of the first obstacle is predicted, n candidate aiming points corresponding to the first moving path are extracted, and the priority among the n candidate aiming points is determined. a collision avoidance processing unit detecting a high m-th candidate aiming point and determining a possibility of collision with the first obstacle while moving to the m-th candidate aiming point;
including,
Wherein n is a natural number greater than or equal to 2, and m is a natural number less than or equal to n,
The collision avoidance processing unit,
A pipe corresponding to the prediction is generated, the n candidate aiming points included in the pipe are extracted, and the closest distance to a preset target point among the n candidate aiming points is detected as the mth candidate aiming point. to do, drone.
4. The method of claim 3,
The collision avoidance processing unit,
and determining that there is a possibility of collision with the first obstacle if the distance to the first obstacle is equal to or less than the radius while moving to the mth candidate aiming point.
5. The method of claim 4,
The collision avoidance processing unit,
Among the n candidate aiming points excluding the mth candidate aiming point, the closest distance to the preset target point is detected as the pth candidate aiming point, and the possibility of collision with the first obstacle while moving to the pth candidate aiming point to judge,
The p is a natural number different from the m among the natural numbers less than or equal to n, the unmanned aerial vehicle.
A first obstacle is detected according to a preset method, a first sphere including the detected first obstacle is generated, and at least one of a diameter or a radius of the first sphere and a center point of the first sphere are determined. an obstacle detector for generating first obstacle information including; and
Using the sequentially input first obstacle information, the first movement path of the first obstacle is predicted, n candidate aiming points corresponding to the first moving path are extracted, and the priority among the n candidate aiming points is determined. a collision avoidance processing unit detecting a high m-th candidate aiming point and determining a possibility of collision with the first obstacle while moving to the m-th candidate aiming point;
including,
Wherein n is a natural number greater than or equal to 2, and m is a natural number less than or equal to n,
The collision avoidance processing unit,
generating a pipe corresponding to the prediction, and extracting the n candidate aiming points included in the pipe;
The obstacle detection unit,
Detect a second obstacle according to the preset method, generate a second sphere including the detected second obstacle, and at least one of a diameter or a radius of the second sphere and a center point of the first sphere Generates second obstacle information comprising a,
The collision avoidance processing unit,
Using the sequentially input second obstacle information, predicting the second movement path of the second obstacle, extracting q candidate aiming points corresponding to the first movement path and the second movement path, and the q detecting the a-th candidate aiming point having a higher priority among the candidate aiming points, and determining the possibility of collision with the first obstacle and the second obstacle while moving to the a-th candidate aiming point;
The obstacle detection unit,
After the first obstacle is detected, the first obstacle information is generated in consideration of a detection time and an obstacle movement distance before a preset condition is satisfied, and after the second obstacle is detected, the detection time before the condition is satisfied , generating the second obstacle information in consideration of the obstacle movement distance,
The collision avoidance processing unit,
tracking the first movement path of the first obstacle using the first obstacle information input before the condition is satisfied, and predicting the first movement path using the tracked first movement path after the condition is satisfied;
A second movement path of the second obstacle is tracked using the second obstacle information input before the condition is satisfied, and the second movement path is predicted using the tracked second movement path after the condition is satisfied,
The condition corresponds to a preset initial time, and the initial time is calculated from the moment the first obstacle is first detected.
7. The method of claim 6,
The obstacle detection unit,
After the condition is satisfied, at least one of the first obstacle and the second obstacle is detected by clustering information input from the provided active sensor according to a preset method,
The collision avoidance processing unit,
After the condition is satisfied, predicting at least one of the first movement path and the second movement path using a preset Kalman filter.
7. The method of claim 6,
The collision avoidance processing unit,
A first pipe corresponding to the prediction of the first movement path is generated, a second pipe corresponding to the prediction of the second movement path is generated, and the first pipe and the pipe included in the second pipe are generated. extracting the q candidate aiming points.
9. The method of claim 8,
The collision avoidance processing unit,
and detecting, as the a-th candidate aiming point, the closest distance to a preset target point among the q candidate aiming points.
10. The method of claim 9,
The collision avoidance processing unit,
If the distance to the first obstacle is less than or equal to the radius of the first sphere or the distance to the second obstacle is less than or equal to the radius of the second sphere while moving to the a-th candidate aiming point, the first obstacle or determining that there is a possibility of collision with the second obstacle, the unmanned aerial vehicle.
11. The method of claim 10,
The collision avoidance processing unit,
Among the q candidate aiming points excluding the a-th candidate aiming point, the closest distance to a preset target point is detected as the b-th candidate aiming point, and the first obstacle and the second aiming point are detected while moving to the b-th candidate aiming point. Determining the possibility of collision with obstacles,
wherein b is a natural number different from a among natural numbers less than or equal to q.

Images