Unmanned aerial vehicle obstacle avoidance system based on obstacle behavior prediction

Abstract

The present invention relates to the technical field of unmanned aerial vehicle obstacle avoidance, and in particular, to an unmanned aerial vehicle obstacle avoidance system based on obstacle behavior prediction. It includes a route planning module, an obstacle type distinction module, an obstacle state capture module, and an avoidance path planning module. The present invention obtains the changing state of various obstacles in real time through the obstacle state capture module, predicts the behavior logic of the obstacle, obtains the current operating speed and operating state of the unmanned aerial vehicle, predicts the overlapping trajectory of the obstacle's operating route and the unmanned aerial vehicle's operating route, establishes an unmanned aerial vehicle path avoidance model, establishes an unmanned aerial vehicle obstacle avoidance operating path, the unmanned aerial vehicle performs an operating state offset according to the operating path, avoids obstacles, distinguishes different types of obstacles in advance, predicts different types of obstacle operating rules, formulates an adaptive avoidance path for the unmanned aerial vehicle, ensures the avoidance success rate, and can improve the avoidance efficiency and reduce the response time of avoiding obstacles.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle obstacle avoidance, and in particular to a unmanned aerial vehicle obstacle avoidance system based on obstacle behavior prediction. Background Art

[0002] With the continuous advancement of drone technology, especially breakthroughs in navigation, obstacle avoidance, and stable control, the practicality and safety of drone delivery have been significantly improved. Drone delivery is regarded as an important development direction for the logistics industry due to its high efficiency and low cost.

[0003] When drones are delivering goods in cities, they need to plan their routes in advance. This not only ensures that the route is as short as possible and saves delivery time, but also determines the obstacles in the route, including obstacles with fixed positions and obstacles with uncertain positions, such as buildings, telephone poles, and overpasses. The avoidance routes of drones and the corresponding locations should be planned according to the positions and heights of the obstacles, and the flight status should be adjusted in advance. However, obstacles with uncertain positions often occur during flight, such as flying birds and moving cars. These obstacles have one thing in common, that is, the moving position and direction are uncertain, which makes it impossible to plan the route in advance. To avoid obstacles, traditional UAV avoidance systems mainly use relative avoidance methods to deal with this type of obstacle, such as falling objects, moving from top to bottom. At this time, you only need to adjust the UAV's running speed or adjust the UAV to move upward before the two paths overlap, and avoid obstacles by relative motion. This processing method is not suitable for all types of obstacles, including obstacles with regular motion, such as obstacles with reciprocating motion. That is to say, after the UAV avoids the current obstacle for the first time, it is easy to contact the UAV in relative motion during its reciprocating motion. If its motion behavior cannot be predicted before avoiding the obstacle, the avoidance route will fail.

[0004] In order to deal with the above problems, there is an urgent need for a drone obstacle avoidance system based on obstacle behavior prediction. Summary of the invention

[0005] The purpose of the present invention is to provide a drone obstacle avoidance system based on obstacle behavior prediction to solve the problems raised in the above background technology.

[0006] To achieve the above purpose, a UAV obstacle avoidance system based on obstacle behavior prediction is provided, including a route planning module, an obstacle type distinction module, an obstacle state capture module and an avoidance path planning module;

[0007] The route planning module obtains the location information of the initial location and the destination, and globally plans the flight route of the drone and the height above the ground during the flight;

[0008] The obstacle type distinguishing module includes a city 3D map simulation unit and a drone real-time monitoring unit. The city 3D map simulation unit establishes a city 3D simulation map, and combines the flight route of the drone and its altitude above the ground during flight to obtain objects in the flight route area that form obstacles to the drone and whose mobile area is fixed, and marks them as map obstacles. The drone real-time monitoring unit uses real-time monitoring equipment to capture objects that are different from map obstacles during the drone's flight path, and marks them as unconventional obstacles.

[0009] The obstacle state capture module obtains the changing state of various obstacles in real time, predicts the behavior logic of obstacles, obtains the current running speed and running state of the UAV, and predicts the overlapping trajectory of the obstacle's running route and the UAV's running route;

[0010] The avoidance path planning module establishes an obstacle avoidance operation path for the UAV, and the UAV performs an operation state deviation according to the operation path to avoid obstacles.

[0011] As a further improvement of the technical solution, the map obstacles include fixed obstacles and obstacles with regularly changing positions, and the unconventional obstacles include non-directional obstacles and abnormal obstacles;

[0012] Wherein, the fixed obstacle is an obstacle whose position and height do not change;

[0013] The regularly changing position obstacle is an obstacle that reciprocates in a specific area;

[0014] The non-directional obstacle is an obstacle to non-directional movement in a specific area;

[0015] The abnormal obstacles are regular obstacles that are not within the flight trajectory of the drone.

[0016] As a further improvement of the technical solution, the method for the route planning module to globally plan the flight route of the drone includes the following steps:

[0017] S1. Obtain the location information of the initial location and the destination, and locate the area points that the drone passes through from the initial location to the destination;

[0018] S2. Establish a shortest distance regional point set P, and use the set P to store the regional points with the shortest distance from the initial point;

[0019] S3. Establish a set Q of unknown area points, and store the remaining unknown area points through the set Q;

[0020] S4. Create an array , determine whether each regional point is in the unknown regional point set Q or in the shortest distance regional point set P;

[0021] When the array When the output is 1, it means that the region point is within the shortest distance region point set P;

[0022] When the array When the output is 0, it means that the region point is in the unknown region point set Q;

[0023] S5. Assume that the shortest distance from the initial point to itself is 0, and the distance of the regional points that the initial point can reach is expressed as , where A represents the initial location, N represents the code of the determined regional point, Z represents the distance between the two locations, and the distance of the regional point that the initial point cannot reach is represented as ∞;

[0024] S6. Select the regional point U closest to the initial point in the unknown regional point set Q. The distance value of the regional point U is Minimum, and added to the shortest distance area point set P;

[0025] S7, taking the regional point U as the starting point, obtaining the regional point V adjacent to the regional point U, and calculating the distance between the adjacent regional point V and the regional point U ,when Less than When Instead of the shortest distance area point set P ;

[0026] S8. Repeat step S7 until the unknown area points in the unknown area point set Q complete the distance calculation, and the final distance value in the shortest distance area point set P is the shortest distance from the initial point to the destination.

[0027] As a further improvement of the present technical solution, the obstacle state capture module obtains in real time the changing states of various obstacles including obstacle size state changes, obstacle speed state changes, obstacle path state changes and obstacle shape state changes.

[0028] As a further improvement of the technical solution, the obstacle state capture module predicts the behavior logic of the obstacle including the following steps:

[0029] S20, capturing obstacles appearing in the flight trajectory through the configured camera, and classifying obstacle types according to map obstacle and unconventional obstacle distinction rules;

[0030] S21. For obstacles whose position and height do not change, a fixed obstacle avoidance algorithm is used to calculate the avoidance route of the UAV;

[0031] S22. For obstacles with regular position changes, a regular obstacle avoidance algorithm is used to calculate the avoidance route of the drone;

[0032] S23, using a configured camera to capture the final speed and direction of an obstacle in non-directional motion in a specific area that leaves the monitoring screen, predicting the obstacle behavior when the drone flies to the position closest to the point where the obstacle leaves the non-directional motion, predicting the obstacle collision range according to the shape and size of the obstacle in non-directional motion, and the drone determines the obstacle avoidance route according to the predicted obstacle collision range;

[0033] S24, using the configured camera to capture conventional obstacles that are not within the flight trajectory of the drone, and hovering to the current flight position, specifying a hovering monitoring unit time, and combining the captured image data to determine the type of conventional obstacles that are not within the flight trajectory of the drone;

[0034] When the obstacle is one with a regularly changing position, the regular obstacle avoidance algorithm in step S22 is used to calculate the avoidance route of the drone;

[0035] If the obstacle is non-directionally moving, repeat step S23 to determine the obstacle avoidance route.

[0036] As a further improvement of the technical solution, the regular roadblock avoidance algorithm in S22 includes the following steps:

[0037] S22.1, let the UAV's trajectory be a radius of The circular trajectory of , the current operating flight speed is ;

[0038] S22.2. The drone is transformed into a particle, and a rectangular coordinate system of relative position is established. The obstacle of the monitored position change rule is Z, and Z is the running radius. + The position of obstacle Z relative to the drone W is , the speed of the drone W relative to the obstacle Z is , is the running speed of the monitored obstacle Z;

[0039] S22.3. Establishing unit time , according to the relative position rectangular coordinate system of S22.2, it is converted into a relative speed rectangular coordinate system, and the relative obstacle speed set is obtained according to the established relative speed rectangular coordinate system ;

[0040] When the relative speed between the drone W and the obstacle Z In the relative obstacle speed set Within, then in time Inside, the drone W and the obstacle Z will definitely collide;

[0041] S22.4. Set the running speed of obstacle Z The value range is the set , get The Minkowski sum of , translate along the moving direction Unit distance, get The brush path, along the relative speed rectangular coordinate system, is the set of areas painted. ;

[0042] When the drone is running at a speed , the drone will not collide with obstacle Z.

[0043] As a further improvement of the technical solution, the regular roadblock avoidance algorithm in S22 further includes the following steps:

[0044] S22.5. Obtain the expected speed of the drone W and the desired speed of obstacle Z ,satisfy ;

[0045] S22.6. Calculation arrive The nearest vector to the edge ,vector The direction of and the normal to the tangent plane Consistency;

[0046] S22.7, in the distance Find the Minkowski sum along the tangent line of the circle, where is the weight value, the specific value is 0.5, and the expected speed of drone W is obtained Allowed range, marked as , when the desired speed of the drone W In the allowed range .

[0047] As a further improvement of the technical solution, the obstacle avoidance algorithm in S21 includes the following steps:

[0048] S21.1. Establish fixed position obstacles The fixed position rectangular coordinate system between the drone and the fixed position obstacle is marked as line segment O;

[0049] S21.2, let the UAV's trajectory be a radius of The circular trajectory of , the current operating flight speed is ;

[0050] S21.3. Establishing unit time , convert the fixed position rectangular coordinate system of S21.1 into a fixed speed rectangular coordinate system, and obtain the fixed obstacle speed set according to the established speed rectangular coordinate system ;

[0051] S21.4. Obtain the fixed speed rectangular coordinate system origin to the fixed obstacle speed set The shortest distance, marked as the desired speed of the drone , at a fixed obstacle speed set Make a tangent at the nearest point to obtain the expected speed set of the drone .

[0052] As a further improvement of the technical solution, the method for the avoidance path planning module to establish a UAV obstacle avoidance operation path includes the following steps:

[0053] S30. Based on the identified obstacle type, obtain the relative obstacle speed set in S22.3 and the fixed obstacle speed set in S21.4 ;

[0054] S31. Adjust the speed based on the current flight status of the drone to a speed that does not belong to the relative obstacle speed set. Or a fixed set of obstacle speeds ;

[0055] S32. Get the expected speed of the drone as well as , adjust the current flight status according to the desired speed.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] In the obstacle avoidance system of the UAV based on obstacle behavior prediction, the obstacle state capture module is used to obtain the changing states of various obstacles in real time, predict the behavior logic of the obstacles, obtain the current operating speed and operating state of the UAV, predict the overlapping trajectory of the obstacle's operating route and the UAV's operating route, establish a UAV path avoidance model, establish a UAV obstacle avoidance operating path, the UAV performs operating state offset according to the operating path, avoids obstacles, distinguishes different types of obstacles in advance, predicts different types of obstacle operating rules, and formulates an adaptive avoidance path for the UAV, which can improve the avoidance efficiency while ensuring the avoidance success rate and reduce the response time of obstacle avoidance. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a flowchart of the overall structure of the present invention;

[0059] Figure 2 It is the global planning simulation diagram of the present invention;

[0060] Figure 3 A position relationship diagram between the initial location, regional points and destination of the present invention;

[0061] Figure 4 It is a schematic diagram of the shortest distance traversal of an array of the present invention;

[0062] Figure 5 It is a schematic diagram of the relative position rectangular coordinate system of the present invention;

[0063] Figure 6 It is a schematic diagram of the relative speed rectangular coordinate system of the present invention;

[0064] Figure 7 For the present invention Brush path simulation diagram;

[0065] Figure 8 The desired speed of the drone of the present invention Allowable range simulation diagram;

[0066] Fig. 9 It is a schematic diagram of a fixed position rectangular coordinate system of the present invention;

[0067] Fig.10 It is a schematic diagram of a fixed speed rectangular coordinate system of the present invention;

[0068] Fig.11 is the expected speed set of the drone of the present invention Simulation diagram.

[0069] The meaning of each number in the figure is:

[0070] 10. Route planning module;

[0071] 20. Obstacle type distinction module; 210. City 3D map simulation unit; 220. UAV real-time monitoring unit;

[0072] 30. Obstacle status capture module;

[0073] 40. Avoid path planning module. DETAILED DESCRIPTION

[0074] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0075] See also Figure 1 , a UAV obstacle avoidance system based on obstacle behavior prediction is provided, including a route planning module 10, an obstacle type differentiation module 20, an obstacle state capture module 30 and an avoidance path planning module 40;

[0076] The route planning module 10 obtains the location information of the initial location and the destination, and globally plans the flight route of the drone and the height above the ground during the flight;

[0077] The obstacle type distinguishing module 20 includes a city 3D map simulation unit 210 and a drone real-time monitoring unit 220. The city 3D map simulation unit 210 establishes a city 3D simulation map, and combines the flight route of the drone and the height above the ground during the flight to obtain objects in the flight route area that form obstacles to the drone and whose mobile area is fixed, and marks them as map obstacles. The drone real-time monitoring unit 220 uses real-time monitoring equipment to capture objects that are different from map obstacles during the drone's flight path, and marks them as unconventional obstacles.

[0078] The obstacle state capture module 30 obtains the changing state of various obstacles in real time, predicts the behavior logic of the obstacles, obtains the current running speed and running state of the drone, and predicts the overlapping trajectory of the obstacle's running route and the drone's running route;

[0079] The avoidance path planning module 40 establishes an obstacle avoidance operation path for the UAV, and the UAV performs an operation state deviation according to the operation path to avoid obstacles.

[0080] In specific use, when the drone performs a delivery task, the route planning module 10 is used to first obtain the location information of the initial location and the destination, and the flight route of the drone and the height above the ground during the flight are globally planned. At this time, the drone flies according to the flight route. In order to reduce the avoidance risk of the drone during the flight, it is necessary to determine the avoidance rules in advance, and the city 3D map simulation unit 210 is used to establish a city 3D simulation map. In combination with the flight route of the drone and the height above the ground during the flight, objects that form obstacles to the drone in the flight route area and have fixed moving areas are obtained and marked as map obstacles, that is, there are obstacles in the flight route area that overlap with the flight path of the drone, and the position changes of these obstacles are constant, such as buildings in fixed positions or doors and windows that are switched (although the positions of the doors and windows switches change, their movement trajectories are constant and reciprocating, so such obstacles are also specified as map obstacles);

[0081] As for some uncertain obstacles, since their locations and movement behaviors cannot be determined, they cannot be simulated in advance through the city 3D simulation map. Therefore, it is necessary to use a real-time monitoring device through the drone real-time monitoring unit 220. The real-time monitoring device is a camera, which captures objects different from map obstacles during the drone flight path and marks them as unconventional obstacles, such as flying birds or falling objects, etc. These objects are captured in real time by the monitoring device;

[0082] Since the trajectory of the captured object is uncertain, it may collide with the future flight area of ​​the drone, so its running behavior needs to be analyzed. At this time, the obstacle state capture module 30 is used to obtain the changing status of various obstacles in real time, and predict the behavior logic of the obstacles, obtain the current running speed and running status of the drone, and predict the overlapping trajectory of the obstacle's running route and the drone's running route. Finally, the avoidance path planning module 40 is used to establish the drone's obstacle avoidance running path, and the drone offsets its running state according to the running path to avoid obstacles.

[0083] In addition, map obstacles include fixed obstacles and obstacles with regularly changing positions, and unconventional obstacles include non-directional obstacles and abnormal obstacles;

[0084] Among them, fixed obstacles are obstacles whose position and height do not change;

[0085] Regularly changing position obstacles are obstacles that move back and forth in a specific area;

[0086] Non-directional obstacles are obstacles to non-directional movement in a specific area;

[0087] Abnormal obstacles are regular obstacles that are not within the flight path of the drone.

[0088] Since the behavior states of different obstacles change differently, when formulating an avoidance plan for the flight trajectory of the drone, it is necessary to formulate it according to the behavior state of the obstacle. For example, for obstacles whose position and height do not change, after the drone identifies the current obstacle, it automatically adjusts the flight state to avoid the current obstacle at one time. When encountering an obstacle that moves back and forth in a specific area, the current obstacle movement state is identified. For example, the current obstacle is an obstacle that moves back and forth up and down. When it moves downward, it will not change its behavior state before reaching the bottom, so it will not move up. At this time, it is only necessary to ensure that the drone is above the obstacle when it passes before the obstacle moves up to avoid the current obstacle. For obstacles that move in an undirected manner in a specific area and conventional obstacles that are not within the flight trajectory of the drone, since their movement behavior is uncertain, it is necessary to use monitoring equipment to predict their movement trends in real time for avoidance.

[0089] Furthermore, the method for the route planning module 10 to globally plan the flight route of the drone includes the following steps:

[0090] S1. Obtain the location information of the initial location and the destination, and locate the area points that the drone passes through from the initial location to the destination;

[0091] S2. Establish a shortest distance regional point set P, and use the set P to store the regional points with the shortest distance from the initial point;

[0092] S3. Establish a set Q of unknown area points, and store the remaining unknown area points through the set Q;

[0093] S4. Create an array , determine whether each regional point is in the unknown regional point set Q or in the shortest distance regional point set P;

[0094] When the array When the output is 1, it means that the region point is within the shortest distance region point set P;

[0095] When the array When the output is 0, it means that the region point is in the unknown region point set Q;

[0096] S5. Assume that the shortest distance from the initial point to itself is 0, and the distance of the regional points that the initial point can reach is expressed as , where A represents the initial location, N represents the code of the determined regional point, Z represents the distance between the two locations, and the distance of the regional point that the initial point cannot reach is represented as ∞;

[0097] S6. Select the regional point U closest to the initial point in the unknown regional point set Q. The distance value of the regional point U is Minimum, and added to the shortest distance area point set P;

[0098] S7, taking the regional point U as the starting point, obtaining the regional point V adjacent to the regional point U, and calculating the distance between the adjacent regional point V and the regional point U ,when Less than When Instead of the shortest distance area point set P ;

[0099] S8. Repeat step S7 until the unknown area points in the unknown area point set Q complete the distance calculation, and the final distance value in the shortest distance area point set P is the shortest distance from the initial point to the destination.

[0100] Depend on Figure 2 As shown, B, C, D, and E represent different regional points, A represents the initial location, and F represents the destination, wherein regional point B, regional point C, regional point D, regional point E, and destination F are stored in the unknown regional point set Q, which is represented by Figure 3 As shown, create an array to describe the distance relationship between the initial location, different regional points, and the destination, for example It means that the distance from the initial point A to the regional point B is 3, the distance between itself is 0, and the distance between unreachable positions is set to ∞. Then a one-dimensional array is created. To store the shortest distance from the initial point A to the rest of the area points, the distance from the initial point A to the rest of the area points is not known at the beginning, so the one-dimensional array All values ​​in are assigned to ∞, and the adjacent area points of the initial location A are traversed, such as Figure 4 As shown, the distances 3Z[A][B] and 10Z[A][C] from the adjacent vertices B and C are updated to Dis[B] and Dis[C], and an estimated value of the shortest distance from the initial point A to the remaining regional points is obtained and stored in the shortest distance regional point set P;

[0101] Next, find a vertex with the shortest distance from the initial point A. From the array Dis, we know that it is the regional point B. Regional point B has two outgoing edges, connecting regional points C and D respectively. The distance from the initial point A to regional point C via regional point B is 8Z[A][B]+ Z[B][C] = 3 + 5, which is less than the distance from the initial point A directly to regional point C, 10. Therefore, the value of Dis[C] is updated from 10 to 8. Similarly, the distance from the initial point A to regional point D via regional point B is 5Z[A][B] + Z[B][D] = 3 + 2, which is less than the initial value of infinity. Therefore, Dis[D] is updated to 5 and stored in the shortest distance regional point set P.

[0102] Among the remaining regional points C, D, E, and destination F, select the regional point D that is closest to the initial point A, and continue to calculate the distance of regional point D in the above manner to obtain the updated values ​​of Dis[E] and Dis[F], and store them in the shortest distance regional point set P;

[0103] Continue to select the regional point C that is closest to the initial point A from the remaining regional points C, regional points E, and destination F, and continue to calculate the distance of regional point C in the same way as above to obtain the updated value of Dis[E], which is stored in the shortest distance regional point set P;

[0104] Continue to select the regional point E closest to the initial point A from the remaining regional points E and the destination F, and continue to calculate all outgoing edges of regional point E in the same way as above to obtain the updated value of Dis[F], and store it in the shortest distance regional point set P;

[0105] Finally, all outgoing edges of the destination F are calculated. In this example, the destination F has no outgoing edges, so no processing is required. The values ​​of the distance from the initial location A in the one-dimensional array Dis have changed from "estimated values" to "determined values", which is the shortest distance from the initial location to the destination.

[0106] Furthermore, the obstacle state capturing module 30 acquires the change states of various obstacles in real time, including the change of obstacle size state, the change of obstacle speed state, the change of obstacle path state and the change of obstacle shape state.

[0107] Specifically, the obstacle state capture module 30 predicts the behavior logic of the obstacle including the following steps:

[0108] S20, capturing obstacles appearing in the flight trajectory through the configured camera, and classifying obstacle types according to map obstacle and unconventional obstacle distinction rules;

[0109] S21. For obstacles whose position and height do not change, a fixed obstacle avoidance algorithm is used to calculate the avoidance route of the UAV;

[0110] S22. For obstacles with regular position changes, a regular obstacle avoidance algorithm is used to calculate the avoidance route of the drone;

[0111] S23, using a configured camera to capture the final speed and direction of an obstacle in non-directional motion in a specific area that leaves the monitoring screen, predicting the obstacle behavior when the drone flies to the position closest to the point where the obstacle leaves the non-directional motion, predicting the obstacle collision range according to the shape and size of the obstacle in non-directional motion, and the drone determines the obstacle avoidance route according to the predicted obstacle collision range;

[0112] S24, using the configured camera to capture conventional obstacles that are not within the flight trajectory of the drone, and hovering to the current flight position, specifying a hovering monitoring unit time, and combining the captured image data to determine the type of conventional obstacles that are not within the flight trajectory of the drone;

[0113] When the obstacle is one with a regularly changing position, the regular obstacle avoidance algorithm in step S22 is used to calculate the avoidance route of the drone;

[0114] If the obstacle is non-directionally moving, repeat step S23 to determine the obstacle avoidance route.

[0115] In specific use, the obstacle avoidance route is determined in different ways for different obstacles. For example, for obstacles whose position and height do not change, since the position of the obstacle does not change, the collision responsibility is borne only by the drone. At this time, there is a fixed collision area in the flight path, and the fixed obstacle avoidance algorithm is used to calculate the drone avoidance route. The drone only needs to adjust the flight altitude or flight direction to avoid it. For obstacles with regular position changes, since the obstacle and the drone move relative to each other, in order to determine whether the two have collided, it is necessary to obtain the position change rules of the obstacle and the position during monitoring in advance, and calculate the drone avoidance route through the regular roadblock avoidance algorithm;

[0116] For obstacles with undirected motion, although their range of motion is determined, their direction and position cannot be determined. Therefore, it is impossible to formulate corresponding algorithms for calculation. We can only capture the final speed and direction of obstacles with undirected motion in a specific area when they leave the monitoring screen, predict the obstacle behavior when the drone flies to the position closest to the point where the obstacle leaves the undirected motion, and predict the obstacle collision range according to the shape and size of the obstacle. The drone determines the obstacle avoidance route according to the predicted obstacle collision range.

[0117] Finally, for conventional obstacles that are not within the flight trajectory of the drone, there are too many uncertain factors, such as their location, speed and size. Therefore, the configured camera can only be used to capture conventional obstacles that are not within the flight trajectory of the drone, and hover to the current flight position, specify the hovering monitoring unit time, and combine the captured image data to determine the type of conventional obstacles that are not within the flight trajectory of the drone;

[0118] When the obstacle is one with a regularly changing position, the regular obstacle avoidance algorithm in step S22 is used to calculate the avoidance route of the drone;

[0119] If the obstacle is non-directionally moving, repeat step S23 to determine the obstacle avoidance route.

[0120] In addition, the regular roadblock avoidance algorithm in S22 includes the following steps:

[0121] S22.1. Figure 5 As shown, the trajectory of the UAV is assumed to be a radius of The circular trajectory of , the current operating flight speed is ;

[0122] S22.2. The drone is transformed into a particle, and a rectangular coordinate system of relative position is established. The obstacle of the monitored position change rule is Z, and Z is the running radius. + The position of obstacle Z relative to the drone W is , the speed of the drone W relative to the obstacle Z is , is the running speed of the monitored obstacle Z;

[0123] S22.3. Figure 6 As shown, the unit time is set , according to S22.2, the relative position rectangular coordinate system is converted into the relative speed rectangular coordinate system, and the relative obstacle speed set is obtained according to the established relative speed rectangular coordinate system ;

[0124] When the relative speed between the drone W and the obstacle Z In the relative obstacle speed set Within, then in time Inside, the drone W and the obstacle Z will definitely collide;

[0125] S22.4. Set the running speed of obstacle Z The value range is the set (like Figure 7 ), get The Minkowski sum of , translate along the moving direction Unit distance, get The brush path, along the relative speed rectangular coordinate system, is the set of areas painted. (That is Figure 7 middle shaded area);

[0126] When the drone is running at a speed , the drone will not collide with obstacle Z.

[0127] Furthermore, the regular roadblock avoidance algorithm in S22 further includes the following steps:

[0128] S22.5, if Figure 8 As shown, the expected speed of the drone W is obtained. and the desired speed of obstacle Z ,satisfy , that is, at the expected speed, the drone W will collide with the obstacle Z;

[0129] S22.6. Calculation arrive ( Figure 8 The vector closest to the edge of the fan-shaped shaded area ,vector The direction of and the normal to the tangent plane Consistency;

[0130] S22.7, in the distance Find the Minkowski sum along the tangent line of the circle, where is the weight value, the specific value is 0.5, and the expected speed of drone W is obtained Allowed range, marked as ( Figure 8 The shaded area in the figure) is the expected speed of the drone W. In the allowed range , which can avoid collision with obstacles while maintaining the direction of the initial flight trajectory and avoid deviation from the original flight trajectory.

[0131] Furthermore, the obstacle avoidance algorithm in S21 includes the following steps:

[0132] S21.1, if Fig. 9 As shown, establish fixed position obstacles The fixed position rectangular coordinate system between the drone and the fixed position obstacle is marked as line segment O;

[0133] S21.2, let the UAV's trajectory be a radius of The circular trajectory of , the current operating flight speed is ;

[0134] S21.3. Establishing unit time , convert the fixed position rectangular coordinate system of S21.1 into the fixed speed rectangular coordinate system, and obtain the fixed obstacle speed set according to the established speed rectangular coordinate system ;

[0135] S21.4. Obtain the fixed speed rectangular coordinate system origin to the fixed obstacle speed set The shortest distance, marked as the desired speed of the drone ( Fig.10 The shaded area in the figure shows the speed set of the fixed obstacle. Make a tangent at the nearest point to obtain the expected speed set of the drone ( Fig.11 The shaded area in the figure), that is, when the desired speed of the drone Belongs to the speed set It can ensure that the UAV does not collide with obstacles while maintaining the current flight trajectory.

[0136] In addition, the method for the avoidance path planning module 40 to establish the obstacle avoidance operation path of the drone includes the following steps:

[0137] S30. Combine the identified obstacle types to obtain the relative obstacle speed set in S22.3 and the fixed obstacle speed set in S21.4 ;

[0138] S31. Adjust the speed based on the current flight status of the drone to a speed that does not belong to the relative obstacle speed set. Or a fixed set of obstacle speeds ;

[0139] S32. Get the expected speed of the drone as well as , adjust the current flight status according to the desired speed.

[0140] Since the drone avoids different obstacles in different ways during obstacle avoidance, it is necessary to first identify the obstacle type and obtain the S22.3 relative obstacle speed set. and the fixed obstacle speed set in S21.4 At this time, the simplest way to avoid is to adjust the flight speed of the drone. When the obstacle type is a regularly changing obstacle, the flight speed of the drone is adjusted to a speed that does not belong to the relative obstacle speed set. , it can ensure that the drone avoids the current obstacle. When the obstacle type is a fixed position obstacle, the drone's flight speed is adjusted to a speed that does not belong to the fixed obstacle speed set. , it can ensure that the drone avoids the current obstacle. Of course, at this time, the drone's flight speed selection is too much and does not meet the values ​​of the two speed sets. In order to reduce the adjustment range of the drone, it is necessary to obtain the expected speed of the drone. as well as , adjust the current flight state according to the desired speed, thereby further shortening the speed adjustment range of the drone while maintaining the initial flight trajectory.

[0141] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. The obstacle avoidance system for unmanned aerial vehicles based on obstacle behavior prediction is characterized by: It comprises a route planning module (10), an obstacle type distinguishing module (20), an obstacle state capturing module (30) and an avoidance path planning module (40); The route planning module (10) obtains the location information of the initial location and the destination, and globally plans the flight route of the drone and its height above the ground during the flight; The obstacle type distinguishing module (20) comprises a city 3D map simulation unit (210) and a drone real-time monitoring unit (220). The city 3D map simulation unit (210) establishes a city 3D simulation map and, in combination with the flight route of the drone and its height above the ground during flight, obtains objects in the flight route area that form obstacles to the drone and whose mobile area is fixed, and marks them as map obstacles. The drone real-time monitoring unit (220) uses real-time monitoring equipment to capture objects that are different from map obstacles during the drone's flight path, and marks them as unconventional obstacles. The map obstacles include fixed obstacles and obstacles with regularly changing positions, and the unconventional obstacles include non-directional obstacles and abnormal obstacles; Wherein, the fixed obstacle is an obstacle whose position and height do not change; The regularly changing position obstacle is an obstacle that reciprocates in a specific area; The non-directional obstacle is an obstacle to non-directional movement in a specific area; The abnormal obstacles are regular obstacles that are not within the flight trajectory of the drone; The obstacle state capturing module (30) acquires the changing states of various obstacles in real time, predicts the behavior logic of the obstacles, acquires the current running speed and running state of the drone, and predicts the overlapping trajectory of the running route of the obstacle and the running route of the drone; The obstacle state capturing module (30) predicts the behavior logic of the obstacle, including the following steps: S20, capturing obstacles appearing in the flight trajectory through the configured camera, and classifying obstacle types according to map obstacle and unconventional obstacle distinction rules; S21. For obstacles whose position and height do not change, a fixed obstacle avoidance algorithm is used to calculate the avoidance route of the UAV; S22. For obstacles with regular position changes, a regular obstacle avoidance algorithm is used to calculate the avoidance route of the drone; S23, using a configured camera to capture the final speed and direction of an obstacle in non-directional motion in a specific area that leaves the monitoring screen, predicting the obstacle behavior when the drone flies to the position closest to the point where the obstacle leaves the non-directional motion, predicting the obstacle collision range according to the shape and size of the obstacle in non-directional motion, and the drone determines the obstacle avoidance route according to the predicted obstacle collision range; S24, using the configured camera to capture conventional obstacles that are not within the flight trajectory of the drone, and hovering to the current flight position, specifying a hovering monitoring unit time, and combining the captured image data to determine the type of conventional obstacles that are not within the flight trajectory of the drone; When the obstacle is one with a regularly changing position, the regular obstacle avoidance algorithm in step S22 is used to calculate the avoidance route of the drone; If the obstacle is non-directionally moving, repeat step S23 to determine the obstacle avoidance route; The avoidance path planning module (40) establishes an obstacle avoidance operation path for the drone, and the drone performs an operation state deviation according to the operation path to avoid obstacles.

2. The obstacle avoidance system for unmanned aerial vehicles based on obstacle behavior prediction according to claim 1, characterized in that: The method in which the route planning module (10) globally plans the flight route of the drone comprises the following steps: S1. Obtain the location information of the initial location and the destination, and locate the area points that the drone passes through from the initial location to the destination; S2. Establish a shortest distance regional point set P, and use the set P to store the regional points with the shortest distance from the initial point; S3. Establish a set Q of unknown area points, and store the remaining unknown area points through the set Q; S4. Create an array , determine whether each regional point is in the unknown regional point set Q or in the shortest distance regional point set P; When the array When the output is 1, it means that the region point is within the shortest distance region point set P; When the array When the output is 0, it means that the region point is in the unknown region point set Q; S5. Assume that the shortest distance from the initial point to itself is 0, and the distance of the regional points that the initial point can reach is expressed as , where A represents the initial location, N represents the code of the determined regional point, Z represents the distance between the two locations, and the distance of the regional point that the initial point cannot reach is represented as ∞; S6. Select the regional point U closest to the initial point in the unknown regional point set Q. The distance value of the regional point U is Minimum, and added to the shortest distance area point set P; S7, taking the regional point U as the starting point, obtaining the regional point V adjacent to the regional point U, and calculating the distance between the adjacent regional point V and the regional point U ,when Less than When Instead of the shortest distance area point set P ; S8. Repeat step S7 until the unknown area points in the unknown area point set Q complete the distance calculation, and the final distance value in the shortest distance area point set P is the shortest distance from the initial point to the destination.

3. The unmanned aerial vehicle obstacle avoidance system based on obstacle behavior prediction according to claim 1, characterized in that: The obstacle state capturing module (30) acquires in real time the changing states of various obstacles, including obstacle size state changes, obstacle speed state changes, obstacle path state changes, and obstacle shape state changes.

4. The unmanned aerial vehicle obstacle avoidance system based on obstacle behavior prediction according to claim 1, characterized in that: The regular roadblock avoidance algorithm in S22 comprises the following steps: S22.1, let the UAV's trajectory be a radius of The circular trajectory of , the current operating flight speed is ; S22.

2. The drone is transformed into a particle, and a rectangular coordinate system of relative position is established. The obstacle of the monitored position change rule is Z, and Z is the running radius. + The position of obstacle Z relative to the drone W is , the speed of the drone W relative to the obstacle Z is , is the running speed of the monitored obstacle Z; S22.

3. Establishing unit time , according to the relative position rectangular coordinate system of S22.2, it is converted into a relative speed rectangular coordinate system, and the relative obstacle speed set is obtained according to the established relative speed rectangular coordinate system ; When the relative speed between the drone W and the obstacle Z In the relative obstacle speed set Within, then in time Inside, the drone W and the obstacle Z will definitely collide; S22.

4. Set the running speed of obstacle Z The value range is the set , get The Minkowski sum of , translate along the moving direction Unit distance, get The brush path, along the relative speed rectangular coordinate system, is the set of areas painted. ; When the drone is running at a speed , the drone will not collide with obstacle Z.

5. The unmanned aerial vehicle obstacle avoidance system based on obstacle behavior prediction according to claim 4, characterized in that: The regular roadblock avoidance algorithm in S22 further includes the following steps: S22.

5. Obtain the expected speed of the drone W and the desired speed of obstacle Z ,satisfy ; S22.

6. Calculation arrive The nearest vector to the edge ,vector The direction of and the normal to the tangent plane Consistency; S22.7, in the distance Find the Minkowski sum along the tangent line of the circle, where is the weight value, the specific value is 0.5, and the expected speed of drone W is obtained Allowed range, marked as , when the desired speed of the drone W In the allowed range .

6. The unmanned aerial vehicle obstacle avoidance system based on obstacle behavior prediction according to claim 1, characterized in that: The obstacle avoidance algorithm in S21 includes the following steps: S21.

1. Establish fixed position obstacles The fixed position rectangular coordinate system between the drone and the fixed position obstacle is marked as line segment O; S21.2, let the UAV's trajectory be a radius of The circular trajectory of , the current operating flight speed is ; S21.

3. Establishing unit time , convert the fixed position rectangular coordinate system of S21.1 into a fixed speed rectangular coordinate system, and obtain the fixed obstacle speed set according to the established speed rectangular coordinate system ; S21.

4. Obtain the fixed speed rectangular coordinate system origin to the fixed obstacle speed set The shortest distance, marked as the desired speed of the drone , at a fixed obstacle speed set Make a tangent at the nearest point to obtain the expected speed set of the drone .

7. The unmanned aerial vehicle obstacle avoidance system based on obstacle behavior prediction according to claim 4, characterized in that: The method for the avoidance path planning module (40) to establish a path for the drone to avoid obstacles comprises the following steps: S30. Based on the identified obstacle type, obtain the relative obstacle speed set in S22.3 and the fixed obstacle speed set in S21.4 ; S31. Adjust the speed based on the current flight status of the drone to a speed that does not belong to the relative obstacle speed set. Or a fixed set of obstacle speeds ; S32. Get the expected speed of the drone as well as , adjust the current flight status according to the desired speed.

Images