RU2794003C1 - Device and method for refining aircraft trajectory - Google Patents

Abstract

FIELD: aircraft (AC).

SUBSTANCE: device and a method for refining trajectory of an aircraft (AC). Device contains a display module, a reference trajectory generation module, a trajectory planning module. To refine the movement trajectory, a map of environmental objects is constructed based on a three-dimensional (3D) point cloud obtained from the 3D point cloud generation module contained in the AC, the distance from the AC to the surrounding objects is estimated in a certain way, a reference trajectory for the AC is generated by interpolating the trajectory constructed according to the flight task between the control points in a certain way, a refined trajectory is formed taking into account obstacle avoidance based on the reference trajectory, a map of environmental objects and data on the distance from the AC to surrounding objects, control signals are generated based on the refined trajectory.

EFFECT: increased accuracy of following the initial trajectory of the aircraft in presence of obstacles, improved smoothness of movement and increased flight safety.

11 cl, 4 dwg

Description

The field of technology to which the invention belongs

The invention relates to aircraft, in particular to the specification of the trajectory of an autonomously moving aircraft.

State of the art

Unmanned aerial vehicles (UAVs, UAVs) are gradually becoming a part of everyday life - they carry out cargo delivery, various studies, analysis of the surrounding space, territory monitoring, object control, photo and video shooting, are used in rescue operations and find many other applications.

The complication of the tasks that can be assigned to UAVs requires them to improve the quality of orientation and control in space, including near people. It is also necessary that aircraft have a high level of safety. At the same time, it is necessary that the aircraft perform their tasks in an optimal way.

UAV following the reference trajectory is used in various fields, for example, in cinematography or when landing on a moving platform. Various approaches have been proposed in the art for following a predefined reference trajectory - for example, in [1] T. Baca, D. Hert, G. Loianno, M. Saska, and V. Kumar, "Model predictive trajectory tracking and collision avoidance for reliable outdoor deployment of unmanned aerial vehicles," 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2018, pp. 6753-6760, [2] M.-E. Guerrero-Sanchez, O. Hernandez-Gonzalez, G. Valencia-Palomo, F.-R. Lopez-Estrada, A.-E. Rodnguez-Mata, and J. Garrido, "Filtered observer-based ida-pbc control for trajectory tracking of a quadrotor," IEEE Access, vol. 9, pp. 114 821-114 835, 2021, [3] O. Mechali, L. Xu, Y. Huang, M. Shi, and X. Xie, "Observer-based fixedtime continuous nonsingular terminal sliding mode control of quadrotor aircraft under uncertainties and disturbances for robust trajectory tracking: Theory and experiment," Control Engineering Practice, vol. 111, p. 104806, 2021.

Despite the efforts made, the designated problem remains open for research for several reasons: achieving real-time performance, avoiding nearby obstacles, being able to be used in various weather conditions, etc. As a result, the generation of a suboptimal control policy for maneuvering in an obstruction-rich, unknown and, moreover, changing environment is quite a challenge, especially from the point of view of real-time dynamic feasibility and flight safety constraints.

Partially, these problems could be solved by the method of control with a predictive model (Model Predictive Control, MPC) - for example, a method based on a linear model (LMPC) or a method based on a non-linear model (NMPC), however, in its pure form, such a method does not allow control in real time in most practical situations due to the limited available computing resources, due to the approaches used in this method to the formulation of the problem and to the processing of obstacles and input data. Moreover, the accuracy of the MRS is highly dependent on the detection capabilities available - for example, the field of view (FoV) of the camera, the detection range, and the way the surrounding free space and obstacles are represented. In addition, the more obstacles are located on a given reference trajectory or in its immediate vicinity, the higher the computational complexity of the MPS method, due to which, under conditions of limited computing resources, MPS can only produce a certain local minimum of the optimization function, which does not allow avoiding a collision with an obstacle.

The essence of the invention

The present invention is directed to devices and methods to overcome at least some of the above disadvantages of the prior art.

In particular, a device is proposed for refining the trajectory of an aircraft (LA) containing:

a display module configured to build a map of environmental objects based on a three-dimensional (3D) point cloud obtained from a 3D point cloud generation module contained in the aircraft, and estimate the distance from the aircraft to surrounding objects based on the map of environmental objects and aircraft location data received from the navigation module contained in the aircraft;

a reference trajectory generation module configured to generate a reference trajectory to be followed by the aircraft by interpolating the trajectory built according to the flight task between the control points through which the aircraft must pass, taking into account the preferred moments of their passage, using a uniform B-spline 3- order; And

a trajectory planning module configured to generate an updated trajectory of the aircraft movement around obstacles based on a reference trajectory, a map of environmental objects and data on the distance from the aircraft to surrounding objects, and generate control signals for the aircraft based on the refined trajectory using a control method with a nonlinear predictive model.

In one of the embodiments, the global planning module is configured to generate an updated trajectory of the aircraft movement around obstacles if it is determined that the reference trajectory passes through an environmental object or that the aircraft or the reference trajectory is at such a distance to the environmental object that is less than a predetermined avoidance distances.

In one embodiment, the location of the aircraft corresponds to the center of the aircraft, and the deviation distance is greater than the radius of the circle describing the aircraft.

In one embodiment, the control signals for the aircraft comprise velocity vectors in each of the x, y, z directions, as well as a yaw angle.

In one embodiment, the device further comprises a navigation module configured to determine the location, speed, and yaw angle of the aircraft.

In one embodiment, the device further comprises a 3D point cloud generation module configured to generate a 3D point cloud.

In one of the embodiments, the device further comprises a proportional-differentiating (PD) controller, configured to generate adjusted control signals for the aircraft and transmit them to the flight controller to control the aircraft actuators, and the PD controller uses data on the current state of the aircraft from the navigation module in as feedback and control signals from the trajectory planning module as an input signal.

In one of the embodiments, the global planning module is configured to generate an updated aircraft trajectory using the Broyden-Fletcher-Goldfarb-Shanno method with limited memory usage in a multidimensional cube.

In one of the embodiments, the global planning module is configured to identify on the reference trajectory a segment of the reference trajectory to be changed, containing a set of successive control points, at least some of which are within the environmental object or at such a distance to the environmental object that is less than a predetermined avoidance distance, dividing the free space around this segment into a plurality of raised fragments, and determining the locations of the control points of the refined segment within said raised free space fragments by applying a parallel convex decomposition to each point of the segment to be changed, and forming an updated trajectory of the aircraft passing through control points of the refined segment.

In one embodiment, if an attempt to generate a refined segment in a predetermined time fails, the global planning module is configured to represent the entire reference trajectory not passed at the moment as a segment to be changed and generate a refined trajectory around the obstacle using parallel convex decomposition .

A method for refining the trajectory of the aircraft movement is also proposed, containing the stages at which:

using the display module, a map of environmental objects is constructed based on a three-dimensional (3D) point cloud obtained from the 3D point cloud generation module contained in the aircraft, and the distance from the aircraft to surrounding objects is estimated based on the map of environmental objects and aircraft location data, received from the navigation module contained in the aircraft;

using the reference trajectory generation module, the generation of the reference trajectory to be followed by the aircraft is performed by interpolating the trajectory built according to the flight task between the control points through which the aircraft must pass, taking into account the preferred moments of their passage, using a homogeneous B-spline of the 3rd order; And

using the trajectory planning module, the formation of an updated trajectory of the aircraft movement around obstacles based on the reference trajectory, a map of environmental objects and data on the distance from the aircraft to surrounding objects, and the formation of control signals for the aircraft based on the refined trajectory using the control method with a nonlinear predictive model.

Technical result

The claimed device contains planning modules that allow you to plan in offline mode in unknown in advance, changing conditions, the most rational trajectory in terms of dynamic feasibility and safety in real time, also ensuring smooth running.

The technical result achieved by the invention is to increase the accuracy of following the initial trajectory of the aircraft in the presence of obstacles, improve the smoothness of movement and increase flight safety.

These and other advantages of the present invention will become apparent upon reading the following detailed description with reference to the accompanying drawings.

Brief description of the drawings

On FIG. 1 shows a block diagram of a device for optimizing the trajectory of the UAV.

On FIG. Figure 2 shows the visualization of the principles of optimal trajectory formation.

On FIG. 3 shows the relationship between the formed trajectories.

On FIG. 4 shows an example of an obstacle avoidance.

It should be understood that the figures may be represented schematically and not to scale and are intended primarily to improve understanding of the present invention.

Detailed description

An unmanned aerial vehicle (UAV) may contain a body, propellers, engines, a landing gear, an engine control module, a flight controller, a transceiver, a navigation module, an energy source, a camera, a set of sensors.

The UAV may also contain an inertial measurement unit.

The set of sensors may include, but is not limited to, a gyroscope, an accelerometer, a barometer, a magnetometer, and so on. Which sensors are installed on the UAV depends on the purpose of the UAV and the functions it performs.

The energy source can be a battery for supplying electrical power to the UAV onboard facilities. In other embodiments, the energy source may be kerosene, hydrogen, a fuel cell, gasoline, diesel fuel, or other fuel sources designed to power UAV propulsion systems and electronics.

Engines are designed to convert energy from an energy source into mechanical energy and can be made, for example, in the form of internal combustion engines (ICE) or electric motors. Electric motors can be made, for example, in the form of brushless motors.

The propellers can be in the form of propellers, in particular clockwise rotating propellers (pushing propellers) or counterclockwise rotating propellers (pulling propellers). At the same time, optionally, pushing and/or pulling propellers in any combination can be used in the UAV. Propellers can be made of two or more blades. Propellers can be made of metal, plastic or composite materials such as carbon fiber. The UAV can be made with a single propeller (for example, with a pusher propeller) or with multiple propellers.

A transceiver is a device responsible for receiving and transmitting radio signals sent to and from an aircraft. The transceiver may comprise a video transmission system including a video transmitter and a video receiver.

The engine control module controls the operation of the engine through actuators in order to increase efficiency and can be made in the form of an electronic speed controller, idle speed controller, valve timing controller, etc.

The navigation module may include a satellite navigation module that, based on data received from a global navigation satellite system (GNSS) such as GPS, GLONASS, BeiDou, Galileo, QZSS, determines the position (geographical coordinates), direction, speed and yaw angle of the UAV and synchronizes time.

The reference trajectory generation module generates the reference trajectory R. Various approaches may be used - for example, one embodiment proposes the following approach. According to the flight task, control points are set in three-dimensional space through which the UAV must fly, as well as time points at which the UAV should preferably be at each control point. Further, the trajectory between these control points, taking into account time, is interpolated using a homogeneous B-spline of the 3rd order based on the Cox-de Boer formulas. The reference trajectory R generated in this way is loaded into the UAV memory, and then, while performing its flight task, it seeks to follow this predetermined reference trajectory. If necessary, the reference trajectory can be generated not in advance, but when the UAV is already in flight. In this case, the current location of the UAV received from the navigation module can be taken as the initial control point, and the remaining control points and the moments of their passage are extracted from the flight task.

However, in practice, especially in a changing environment, it may happen that some sections of the reference trajectory at the time the UAV approaches them may be in the area of an obstacle (for example, a person or vehicle may have moved on the path, or the trajectory may initially be routed through an unknown environment that may contain stationary objects such as buildings, trees, etc.). Accordingly, it is required to correct the reference trajectory directly in flight, while striving to adhere to it as much as possible so that the flight task is completed, and on the other hand, so that the flight is carried out safely both for the UAV and for the environment.

Next, with reference to FIG. 1 describes the operation of the UAV to optimize the flight path. For this, the UAV may contain additional blocks, which are also mentioned in the following description.

To determine the position of the UAV in space, instead of the satellite navigation module or in addition to it, the navigation module may contain a visual odometry module, the algorithm of which includes the following steps:

- detection of key points of the image,

- matching or tracking of key points between two successive images,

- estimation of motion parameters based on the displacement of key points.

Accordingly, the UAV may include a visual odometry module. Visual odometers can be based on a monocular, stereo, or omnidirectional vision system. The odometer selects an object in a frame and tracks it on successive frames, calculating the orientation and distance to the object - for example, by triangulation. These data allow not only to determine at what point the UAV is currently located, but also to determine the speed of the UAV and the angle of yaw.

In addition, the UAV may contain a three-dimensional (3D) point cloud generation module implemented based on a scanner (for example, laser, such as lidar, based on radio waves, such as radar, etc.) or a camera (for example, monocular or stereo ).

The data from the navigation module and the data from the 3D point cloud generation module are transmitted to the display module, which in turn contains a mapping module and a distance estimation module.

Based on the received data, the mapping module builds a map of environmental objects, which can be considered, in essence, as an obstacle map, since the UAV must avoid collision with objects in flight. In one embodiment, the map of environmental objects can be formed as an octree based on a point cloud - for example, using the octomap library.

The distance estimation module estimates the distance from the UAV to the surrounding objects using the data on the UAV's location in space and a map of environmental objects. In one embodiment, the distance may be considered a Euclidean distance, and an EDTM (Euclidean Distance Matrix Transform Distance Mapping) method may be used for estimation.

The data obtained in the mapping module is transmitted to the trajectory planning module. In addition, the reference trajectory from the reference trajectory generation module is transmitted to the path planning module. Accordingly, having data on environmental objects, data on the location of the UAV in space, data on the distance from the UAV to environmental objects and data on the reference trajectory, the trajectory planning module in real time corrects the current trajectory of the UAV in order to avoid obstacles.

The trajectory planning module contains a global planning module and a local planning module. The proposed approach is based on the concept of two-level trajectory refinement, in which the global planning module attempts to refine the original reference trajectory by moving the reference trajectory away from known obstacles, while the local planning module generates a suboptimal UAV motion control policy along the refined trajectory.

In particular, the global planning module first attempts to generate a refined trajectory around obstacles based on the reference trajectory when it passes either through an obstacle or in close proximity to an obstacle, and if the attempt fails, calculates a backup refined trajectory. The global scheduling module mainly focuses on real-time performance and obstacle avoidance.

The local planning module, through predictive motion control based on a limited nonlinear model, ensures the safety, smoothing and dynamic feasibility of the refined trajectory, thereby forming the optimal trajectory and optimal control actions on the UAV.

At the output of the trajectory planning module, control signals are provided that allow the UAV to move along a certain optimal trajectory - in particular, the velocity vectors are set in each of the x, y, z directions, as well as the yaw angle around the z axis.

The data from the navigation module and the data from the trajectory planning module are transmitted to a proportional-differential controller, which outputs adjusted control signals by comparing the current state of the UAV (coordinates, speed and yaw angle) as feedback and optimal control signals as an input signal. In particular, the controller generates a control signal that is the sum of two terms, the first of which is proportional to the difference between the input signal and the feedback signal (error signal), and the second is the derivative of the error signal. Thus, discrepancies with the optimal control signal caused by external conditions (wind, rain, etc.), non-ideal physical actuators (motors, propellers) and signal processing delays are taken into account.

The adjusted control signals (velocity vectors in each of the x, y, z directions, as well as the yaw angle around the z axis) from the proportional-differentiating (PD) controller are transmitted to the flight controller, which controls the engines on their basis, and thereby the UAV performs the required moving along the optimal trajectory.

Thus, increased safety and accuracy of following the reference trajectory is provided, including in previously unknown, changing conditions.

Trajectory planning module

Since the main parameters of the trajectory planning module operation are trajectory smoothing, obstacle avoidance and dynamic feasibility, the objective function J performed by the trajectory planning module can be represented as follows:

where λ are weight coefficients defining the contribution of each of the function J of trajectory smoothing, obstacle avoidance, and dynamic feasibility.

As mentioned above, the global scheduling module and the local scheduling module have slightly different tasks, so the weighting factors in the operation of these modules individually may differ accordingly.

, при этом ввиду того, что поле зрения датчиков БПЛА является ограниченным, карта объектов окружающей среды также ограничена соответствующей областью вокруг БПЛА (показана в виде круга с центром в местоположении Q_p БПЛА). Остальное окружающее пространство неизвестно (или недостаточно известно) БПЛА, оно обозначено символом

.The visualization of the work of the trajectory planning module is shown in Fig. 2. At time t _n the UAV itself is at point Q _p. The map of environmental objects built by the mapping module, which is input to the trajectory planning module, is marked with the symbol

, while due to the fact that the field of view of the UAV sensors is limited, the map of environmental objects is also limited to the corresponding area around the UAV (shown as a circle centered on the location Q _p of the UAV). The rest of the surrounding area is unknown (or insufficiently known) to the UAV, it is indicated by the symbol

.

. Затемненная область O_u обозначает неизвестное препятствие, то есть препятствие, которое существует на самом деле, но о наличии которого БПЛА не знает, так как оно расположено за пределами карты

, в неизвестной области

. Области препятствий окружены участками D_z, которые обусловлены расстоянием уклонения, то есть минимальным расстоянием, допустимым между БПЛА и препятствием для предотвращения столкновения.The shaded area O _m denotes a known obstacle, i.e. an obstacle that the UAV knows about because it is located inside the map

. The shaded area O _u denotes an unknown obstacle, i.e. an obstacle that actually exists, but the presence of which the UAV does not know, since it is located outside the map

, in an unknown area

. The obstacle areas are surrounded by sections D _z , which are determined by the avoidance distance, that is, the minimum distance allowed between the UAV and the obstacle to prevent collision.

, свободное от объектов (препятствий). Область F_u обозначает неизвестное свободное пространство, то есть пространство за пределами карты

, в неизвестной области

, свободное от объектов (препятствий). Соответственно, можно считать, что пространство в пределах карты

состоит из областей O_m (известных препятствий) и F_m (известного свободного пространства), а пространство в неизвестной области

состоит из областей O_u (неизвестных препятствий) и F_u (неизвестного свободного пространства). В свою очередь, карта

и неизвестная область

вместе составляют общее трехмерное пространство R³.Area F _m denotes the known free space, i.e. the space inside the map

, free from objects (obstacles). Area F _u denotes unknown free space, i.e. space outside the map

, in an unknown area

, free from objects (obstacles). Accordingly, we can assume that the space within the map

consists of regions O _m (known obstacles) and F _m (known free space), and the space in the unknown region

consists of areas O _u (unknown obstacles) and F _u (unknown free space). In turn, the map

and unknown area

together make up the common three-dimensional space R ³ .

. Количество контрольных точек в пределах

обозначено как N_r. Конечная контрольная точка, показанная на Фиг. 2 как точка с_е, определяется моментом окончания уточненной траектории и может иметь индекс е от s+1 до N_c. Оптимальная траектория для момента времени t_n, которую формирует модуль локального планирования, обозначена как

.The original reference trajectory R (marked with frequent dots) contains a set of control points c _i , where i=0, ..., N _c , while the number N _c specifies the total number of control points in the reference trajectory. The initial control point (with index 0) is located at the beginning of the trajectory. The path planning initialization point in FIG. 2 is denoted by time t ₀ . The checkpoint corresponding to the current point in time (denoted by time t _n ) is shown in FIG. 2 as point c _s . On the one hand, this is the point where the UAV should be located according to the reference trajectory at the moment t _n . On the other hand, this is the point that should be the starting (initial) point for the further movement of the UAV along the reference trajectory, if we consider the movement starting from the moment t _n . It should be understood that, depending on time, the nearest starting checkpoint c _s may have an index s from 0 to N _c -1. The refined trajectory for the moment of time t _n , which is generated by the global planning module, is designated as

. Number of control points within

denoted as N _r . The end checkpoint shown in Fig. 2 as a point with _e , is determined by the end of the refined trajectory and may have an index e from s+1 to N _c . The optimal trajectory for the moment of time t _n , which is formed by the local planning module, is denoted as

.

и оптимальной траекторией

, демонстрируемое предложенным способом в реальных условиях, показано также на Фиг. 3. Как можно видеть, уточнение траектории выполняется в пределах части предстоящей опорной траектории, тогда как оптимальная траектория ищется в пределах части уточненной траектории. Это связано как с ограниченностью поля обзора датчиков и вычислительных ресурсов БПЛА, так и с тем, что уточнение и оптимизация траектории на большом отдалении от БПЛА может не иметь практического смысла в условиях изменяющейся и/или насыщенной препятствиями окружающей среды, что может приводить к растрате ресурсов обработки, притом что необходимо обеспечивать обработку в режиме реального времени.Relationship between reference trajectory R, refined trajectory

and optimal trajectory

, demonstrated by the proposed method in real conditions, is also shown in Fig. 3. As can be seen, the trajectory refinement is performed within a portion of the upcoming reference trajectory, while the optimal trajectory is sought within a portion of the refined trajectory. This is due both to the limited field of view of the sensors and computing resources of the UAV, and to the fact that the refinement and optimization of the trajectory at a great distance from the UAV may not make practical sense in a changing and/or saturated environment with obstacles, which can lead to a waste of resources. processing, while it is necessary to provide processing in real time.

Returning to Fig. 2, the estimated time difference between two successive test points, i.e. between c _i and c _i+1 , denoted as δt _d . The actual time taken to complete the path from the moment t ₀ of the initialization of trajectory planning to the starting checkpoint c _s is denoted as δt _c . Accordingly, in practice, the values of δt _c and δt _d may be the same or may differ from each other.

Next, the operation of the global scheduling module and the local scheduling module will be described in more detail.

Global planning module

The main task of the global planning module is to avoid obstacles, therefore, in the representation (1) mentioned above, specifically for this module, among the weight coefficients λ _smoothing , λ _obstacle and λ _feasibility , the coefficient λ _obstacle should have the highest value, thereby determining the importance of the contribution of the obstacle avoidance function to the overall result of the global planning module. Determining the Direction of the Toolpath Offset

Some control points in the reference trajectory R may be in the obstacle area O _m and they must be moved to the obstacle-free area F _m . The displacement of one control point leads to the fact that the sections of the trajectory adjacent to this point are also displaced. An example of a path offset is shown in FIG. 4.

. Этот сегмент состоит из набора контрольных точек c_j, их число составляет

.Here, a number of control points c _j on the reference trajectory R (shown by the dotted line) are in the area of the obstacle. Accordingly, they must be shifted to a free area. Throughout the entire trajectory of the UAV, there may be several such segments that need to be shifted, especially in conditions of an environment saturated with obstacles and in conditions when environmental objects can move and overlap the trajectory originally built along obstacle-free areas, i-th segment of the trajectory to be changed is indicated in FIG. 4 as

. This segment consists of a set of control points c _j , their number is

.

. Если видоизменить пример на Фиг. 4 так, что в нем одна точка c_j-1 находилась бы в свободном пространстве, а предыдущие и последующие контрольные точки по-прежнему оставались бы внутри препятствия, то это означало бы, что условие по предварительно заданному количеству последовательных контрольных точек за пределами препятствия не выполнено, и все равно изменению подлежал бы сегмент, включающий в себя все эти 8 контрольных точек.The selection of segments on the trajectory to be changed is proposed to be carried out as follows: if there is an obstacle on the reference trajectory, then it is checked whether the trajectory before this obstacle and after it has a predetermined number of consecutive control points located in free space. In the example in FIG. 4 this number is 2, i.e. there is an obstacle on the trajectory, there are 2 adjacent control points in free space before the obstacle on the trajectory, 4 control points are located inside the obstacle on the trajectory, and further after the obstacle on the trajectory there are 2 more adjacent control points in free space . Accordingly, the path segment including these 8 control points is considered as the segment to be modified.

. If we modify the example in Fig. 4 in such a way that one point c _j-1 would be in free space, and the previous and subsequent control points would still remain inside the obstacle, this would mean that the condition for a predetermined number of successive control points outside the obstacle does not completed, and the segment that includes all of these 8 control points would still be subject to change.

, в свободное пространство может повлечь за собой также смещение некоторых соседних контрольных точек - например, точки c_j-1. Тем самым, формируется уточненный сегмент траектории (показан сплошной линией, обозначен

). В примере на Фиг. 4 уточненный сегмент также содержит 8 контрольных точек.Offset of control points c _j included in the segment

, into free space may also entail the displacement of some neighboring control points - for example, point c _j-1 . Thus, a refined segment of the trajectory is formed (shown by a solid line, denoted by

). In the example in FIG. 4 refined segment also contains 8 control points.

The direction of displacement of each control point is determined by solving the following convex optimization problem for each segment:

This problem can be solved, for example, using the MOSEK solver.

Where

(здесь начало подлежащего изменению сегмента для удобства обозначено с₀), как показано на Фиг. 4 в виде набора полупрозрачных прямоугольников;A and b represent the area of free space F _m in the form of a convex polyhedron from ₀ to c _n in

(here, the beginning of the segment to be changed is denoted with ₀ for convenience), as shown in FIG. 4 as a set of translucent rectangles;

, соответствующего подлежащему изменению сегменту

, используемые для определения направления градиента каждой контрольной точки;p ₀ , …, p _n are control points of the refined segment

corresponding to the segment to be changed

, used to determine the direction of the gradient of each control point;

t ₁ is the limit offset of the initial control point (c ₀ to p ₀ );

t ₂ is the limit offset of the end control point (from c _n to p _n );

t ₃ is the limit total distance between adjacent intermediate control points (that is, the sum of the distances between points p _k and p _k+1 , where k takes values from 1 to n-1), which, to some extent, can be considered as the limit length of the refined segment trajectories on the section between intermediate control points;

λ ₁ , λ ₂ and λ ₃ are regularization parameters that determine the contribution of each of the limits t ₁ , t ₂ and t ₃ to the overall result. The regularization parameters can be set to provide more bias at the start and end breakpoints than intermediate breakpoints - for example, λ ₁ =0.8, λ ₂ =0.8, and λ ₃ =0.6.

Parallel convex decomposition

) на выпуклые части. Как только определены контрольные точки с₀, …, c_n, находящиеся в

, необходимо проверить, имеются ли среди них промежуточные контрольные точки, которые находятся в свободном пространстве F_m. Выпуклая декомпозиция применяется к таким последовательным контрольным точкам параллельно, что в результате дает свободное пространство в виде множества решений конечной системы линейных алгебраических уравнений и неравенств, соответствующих крайним опорным гиперплоскостям (Н-rep), Ах≤b для каждого

.To reduce the computation time, a parallel version of the convex decomposition can be used. A convex decomposition is a division of an object (in this case, the free space F _m adjacent to the segment to be modified

) into convex parts. As soon as the control points with ₀ , …, c _n , which are in

, it is necessary to check whether there are intermediate control points among them, which are in the free space F _m . The convex decomposition is applied to such successive control points in parallel, resulting in a free space in the form of a set of solutions to a finite system of linear algebraic equations and inequalities corresponding to the extreme support hyperplanes (H-rep), Ax≤b for each

.

находится в свободном пространстве и является началом первого прямоугольника (выпуклого многоугольника). Все следующие контрольные точки, которые находятся в свободном пространстве между с₀ и препятствием, также попадают в первый прямоугольник. Конечная контрольная точка c_n сегмента

находится в свободном пространстве и является концом второго прямоугольника. Первый прямоугольник и второй прямоугольник частично перекрываются между собой. Участок сегмента

между препятствием и контрольная точка c_n, в котором находится еще одна контрольная точка, не попадает во второй прямоугольник, поэтому строится третий прямоугольник, который охватывает этот участок, а также контрольную точку c_n, то есть частично перекрывается со вторым прямоугольником. Таким образом, внутри первого, второго и третьего прямоугольников, которые огибают препятствие и находятся в свободном пространстве, можно построить уточненную траекторию движения.In this case, the starting control point from the segment

is in free space and is the beginning of the first rectangle (convex polygon). All of the following control points that are in the free space between c ₀ and the obstacle also fall within the first box. End control point c _n segment

is in free space and is the end of the second rectangle. The first rectangle and the second rectangle partially overlap each other. segment plot

between the obstacle and the control point c _n , which contains another control point, does not fall into the second rectangle, so a third rectangle is built that covers this area, as well as the control point c _n , that is, it partially overlaps with the second rectangle. Thus, inside the first, second and third rectangles that go around the obstacle and are in free space, it is possible to build a refined trajectory.

Computing gradients

(см. Фиг. 2) в момент времени t_n может быть разбита на множество N^seg сегментов

(где i=0, …, N^seg), которые находятся в области препятствий и которые необходимо передвинуть в область, свободную от препятствий. Как упоминалось выше, в результате формируется множество уточненных сегментов

Refined trajectory

(see Fig. 2) at time t _n can be divided into many N ^seg segments

(where i=0, …, N ^seg ) that are in the area of obstacles and which need to be moved to the area free of obstacles. As mentioned above, the result is a set of refined segments

находящейся в области препятствия, зададим вектор v₁ (см. Фиг. 4), который представляет собой аппроксимированный вектор направления движения по траектории вдоль сегмента

около точки c_j. Вектор v₁ можно представить путем движения от точки c_j-1 до точки c_j+1, которые являются соседними с точкой c_j; то есть v₁=c_j+1-c_j-1. Также зададим контрольную точку p_k на уточненном сегменте

Вектор v₂ указывает направление смещения точки c_j к точке p_k, проходя через них и пересекая как уточненный сегмент

так и подлежащий изменению сегмент

(а именно, для упрощения вычисления - вектор v₁). Соответственно, вектор v₂ можно представить путем движения от точки c_j до точки p_k, то есть v₂=p_k-c_j.To find the gradient direction vector for the control point c _j of the segment

located in the area of the obstacle, we set the vector v ₁ (see Fig. 4), which is an approximate vector of the direction of movement along the trajectory along the segment

near the point c _j . Vector v ₁ can be represented by moving from point c _j-1 to point c _j+1 that are adjacent to point c _j; i.e. v ₁ =c _j+1 -c _j-1 . We also set a control point p _k on the refined segment

The vector v ₂ indicates the direction of displacement of the point c _j to the point p _k , passing through them and intersecting as a refined segment

and the segment to be changed

(namely, to simplify the calculation - the vector v ₁ ). Accordingly, the vector v ₂ can be represented by moving from the point c _j to the point p _k , that is, v ₂ =p _k -c _j .

An optimal value of k indicating the index of the reference point p _k of the refined segment into which the current point c _j should be shifted. is determined by calculating the angle between v ₁ and v ₂ :

Thus, the gradient vector corresponding to c _j can be fully defined as:

то есть при инициализации значение k сначала может быть равно половине числа

точек в уточненном сегменте (при необходимости округленно).In this case, in the optimization process, the value of k can first be set as corresponding to the middle of the refined segment

that is, during initialization, the value of k may initially be equal to half the number

points in the refined segment (rounded if necessary).

Then, k is iteratively incremented or decremented according to whether the current iteration, compared to the previous one, has optimized the angle between v ₁ and v ₂ .

After the determination of the gradient vector for each i-th segment, the function J _obstacle , mentioned above in expression (1), is determined as a continuously differentiable penalty function:

As mentioned above, D _z is a predetermined avoidance distance, that is, the minimum distance allowed between the UAV and an obstacle to avoid collision. In the present invention, the UAV is considered to be at the reference point when its center is at the reference point. Accordingly, in order to avoid a collision with an obstacle, the control point must be more than the radius of the UAV (taking into account the rotation of the blades) from the obstacle. As a non-limiting example, the evasion distance D _z may be set to 0.8 m.

Dead zone

и такая траектория

может оказаться динамически неосуществимой. Следовательно, требуется защитный механизм, который позволил бы построить уточненную траекторию в обход препятствия, даже если вышеуказанный подход выдал неудовлетворительный результат. Для таких случаев предлагается следующий механизм восстановления, при котором вся траектория рассматривается целиком как один-единственный сегмент, который необходимо сместить в свободное пространство:When the depth sensor has a small field of view, the mapping may not be completely accurate, since there may be an obstacle immediately outside the field of view, and the area D _z around the obstacle may be mistakenly considered an area free of obstacles. In addition, the EDTM method takes a significant amount of time when the environment is cluttered (saturated with obstacles), and due to the fact that there is a calculation time limit (processing must be carried out in real time), the method may not have time to work out. In such situations, segmentation of free space may not provide the correct set of restrictions A, b, and the refined trajectory may lead the UAV Q _p to the area D _z . Thus, the approach proposed above with segment shift and search for control points of the refined segment may not properly determine the control points p _k , i.e. estimated control points may end up at the ends (with ₀ , c _Ni ^seg ) of the segment

and such a trajectory

may not be dynamically feasible. Therefore, a protective mechanism is required that would allow building an improved trajectory around the obstacle, even if the above approach gave an unsatisfactory result. For such cases, the following recovery mechanism is proposed, in which the entire trajectory is considered as a whole as a single segment that must be shifted into free space:

This problem can be solved, for example, using the MOSEK solver.

- это контрольные точки, подлежащие перемещению, при этом

=1, N_r, a N_r - это количество контрольных точек в момент времени на траектории

.Where

are the control points to be moved, while

=1, N _r , and N _r is the number of control points at a time on the trajectory

.

The stopped breakpoints are defined using

A _r and b _r are determined by the method of parallel convex decomposition.

Smoothing

In order to determine the _smoothing function J mentioned above in expression (1), in order to avoid vibration effects, in general, in addition to speed, other derivatives of the higher order position, such as acceleration, jerk, etc., must be taken into account. The present invention is aimed at optimizing the trajectory when the UAV moves at a relatively low speed (up to 1.5-2 m/s), so such a derivative as jerk and higher order derivatives are negligible and can be ignored.

Thus, to ensure smoothness, the smoothing function (J _smoothing ) can only be formulated as a minimization of the acceleration components:

where ∂a _i /∂c _i =1, ∂a _i /∂c _i+1 =-2, ∂a _i /∂c _i+2 =1,

a _i is the acceleration (a _i =c _i+2 -2c _i+1 +c _i ) at the i-th index

v _i is the speed (v _i =c _i+1 -c _i ) at the i-th index

c _i is the breakpoint at the i-th index

Feasibility

In order for the refined trajectory to be dynamically feasible (i.e., so that the UAV can actually perform a maneuver based on its technical capabilities), the corresponding feasibility function (see J _feasibility in expression (1) above) imposes a penalty on the velocity and acceleration components only when they exceed the minimum and maximum limits, as follows:

определяется следующим образом:where operator

is defined as follows:

in this case, the allowable maximum speed and acceleration limits are set through v _max and a _max , respectively, and the minimum limits - through -v _max and -a _max . When the speed and acceleration are within the allowable range, no penalty is applied.

Calculation of the objective function

Thus, by completing the above steps, it is possible to determine the objective function J (see expression (1)), which includes the functions of trajectory smoothing (J _smoothing ), obstacle avoidance (J _obstacle ) and dynamic feasibility (J _feasibility ). to solve the objective function J, for example, the L-BFGS-B method (Broyden - Fletcher - Goldfarb - Shanno algorithm with limited memory usage in a multidimensional cube) can be used.

Local planning module

In the present invention, local scheduling is based on solving a non-linear optimization problem, in particular NMPC. NMPC calculation time increases when the number of constraints increases, that is, when the trajectory passes near or through obstacles. Such a situation can lead to falling into a local minimum and failure in the search for the optimal UAV control policy, which ultimately can result in a collision with an obstacle. However, the present invention does not have this disadvantage, since the global planning module refines the original reference trajectory, moving it away from the obstacle area.

In addition, the fact that the trajectory is divided into many segments from the checkpoint to the checkpoint, between which the UAV passes a relatively small distance in a relatively short period of time, the restrictions on movement imposed by such segments are individually not very complex, as a result of which the method NMPC can be successfully applied to solving optimization problems on such segments in real time.

Accordingly, the local scheduling module in the present invention can focus on the tasks of providing smoothing, dynamic feasibility, and accuracy of the refined trajectory. As a result, in the representation (1) mentioned above, specifically for this module, among the weight coefficients λ _smoothing , λ _obstacle and λ _feasibility , the coefficients λ _smoothing and λ _feasibility should have the highest value, thereby determining the importance of the contribution of the trajectory smoothing functions and dynamic feasibility to the objective function J.

для заданного горизонта планирования, то есть для того количества N_p контрольных точек, которое подлежит обработке модулем локального планирования.At time t _n, the local planning module generates the optimal trajectory

for a given planning horizon, that is, for the number N _p of checkpoints to be processed by the local planning module.

where Q _p is the current location of the UAV, and c are the next control points on the trajectory that the UAV must pass.

The local planning module generates the optimal control for UAV maneuvering subject to the imposed constraints, where w=[u _tn , …, u _tn+Np-1 , x _tn , …, x _tn+Np ] denotes the set of decision variables to be minimized. Here, x _tn is a state vector determined considering the x, y, z directions and yaw angle for each control point on the refined trajectory received from the global planning module, and u _tn are control inputs (control actions). Accordingly, the task of the local scheduling module is to simultaneously optimize both input control signals and state vectors. Such a task (objective function J _P (x,u) _tn ) can be formed using the multiple shooting method as follows:

, где

изменяется в пределах от 0 до N_p. v_min и v_max обозначают минимальную и максимальную линейную и угловую скорости, допустимые для движения данного БПЛА.The function g ₁ (w) indicates the restrictions imposed on the movement by the dynamics of the system, and the function g ₂ (w) indicates the restrictions imposed on the movement by approaching obstacles. x _min and x _max denote the minimum and maximum bound for the state vector

, Where

varies from 0 to N _p. v _min and v _max denote the minimum and maximum linear and angular speeds allowed for the movement of this UAV.

где v обозначает скорость в каждом направлении х, у, z, а со обозначает угол рыскания вокруг оси z.The local planning module, like the global planning module, works iteratively. When one iteration with trajectory recalculation is completed, the next one starts immediately or after a certain period of time. At each scheduling iteration, the local scheduling module takes Q _p , x _tn and u _tn as input and produces the optimal control policy as output, i.e.

where v denotes the speed in each direction x, y, z, and co denotes the yaw angle around the z-axis.

Thus, as a result of processing performed in the global planning module and in the local planning module, an optimal trajectory is formed that the UAV should follow, and control signals that allow the UAV to move exactly along the given trajectory.

Data from the visual odometry module and data from the trajectory planning module are transmitted to the proportional-derivative controller. The end result of the processing associated with trajectory optimization is the supply of adjusted control signals (velocity vectors in each of the x, y, z directions, as well as the yaw angle around the z axis) from the proportional-differential controller to the flight controller, which, on their basis, controls the engines , and thus the UAV performs the required movements along the optimal trajectory.

Synchronization

Data processing in each module is performed iteratively. When one iteration is completed, the next one begins immediately or after a certain period of time.

In the present invention, a plurality of modules are used to obtain the final result. Under conditions of an unknown and/or changing environment and consequently changing input data, a single module performs processing during different iterations at different times. Accordingly, it is impossible to determine in advance exactly how long it will take a particular module to perform processing.

In one embodiment, the iterations in each module are performed continuously. The result of processing the last iteration is stored in memory. When the module completes processing, it immediately proceeds to the next iteration, while receiving input data not directly from the module whose data it needs to process, but from memory, extracting from it the result of processing the last iteration of the required module. This provides a simple data update mechanism in which the data is updated continuously. The disadvantage of this approach is that the processing time of one module can be very different from the processing time of another module, due to which, in the same period of time, the first module can work once, and the second module, which supplies data for the first - for example , 8 times, of which 7 will be useless, which causes a waste of computing resources and energy. In addition, in practice, the data stored in memory sometimes lose their relevance, which can adversely affect the accuracy of determining the trajectory and flight safety.

In another embodiment, one or more modules iterate at predetermined intervals. For example, the local scheduling module starts a new iteration with a period of 80 ms. If processing completed in 50ms, then the next iteration will start in 30ms. If the selected processing method allows you to set a limit on the processing time, then the limit can be set to a processing time that is equal to or less than the processing period. If the selected processing method does not allow a limit on the processing time, then the processing period can be set to a value equal to the maximum or average maximum processing time demonstrated by this module during tests for similar conditions of use. If the processing time is exceeded and there is no current output, the next module uses as input the result of processing the last iteration of this module.

Accordingly, knowing the maximum processing interval among all modules, it is possible to set the same processing interval for all modules. For example, if the maximum processing interval among all modules has a path planning module and its value is 100 ms, then a value of 100 ms can be assigned as a processing interval for all modules, and all modules can start processing at the same time. Thereby, it is ensured that resources are not wasted due to differences in processing time.

In another embodiment, knowing which module has the maximum processing interval among all modules, one can set the same processing interval for those modules that only supply data to that module. For example, the reference path generation module only supplies data to the path planning module, so its processing interval can be set equal to the processing interval of the path planning module, while the navigation module supplies data to several modules at once, so its processing interval can be set to the smallest of the processing intervals. those modules to which it supplies data - for example, the processing interval of a PD controller. Unlike the previous option, this option provides more up-to-date data, which reduces the processing delay and the errors of deviation from the optimal trajectory caused by this, and also improves flight safety.

In another embodiment, the operation schedule of each module can be generated so that by the time the module that consumes information from this module starts working, this module has time to process and produce a relatively up-to-date result. For example, the local scheduling module has an RMS maximum processing time of 70 ms, according to which its processing period is set; the global scheduling module has an RMS maximum processing time of 50 ms. Knowing the moment when the local scheduling module starts a new iteration, you can set the start time of the global scheduling module so that it finishes its processing just before the local scheduling module starts processing (that is, with a lead of 50-55 ms) to give it more up-to-date data than if the iteration started at the same time, and after the completion of processing by the global scheduling module, the data would wait for processing by the local scheduling module at the next iteration for about 20 ms. In this way, an additional increase in the relevance of the data is provided, which reduces the processing delay and the deviation errors caused by this deviation from the optimal trajectory, and also improves flight safety.

Distance

Above with respect to FIG. 2 it was indicated that the estimated time difference between two successive control points on the trajectory is denoted as δt _d . This value is preset depending on the requirements of the particular application. For example, the value of δt _d may be set to 0.05 s. Accordingly, the distance that the UAV must overcome during this time may differ between different neighboring points on the trajectory. The shorter the time of flight between two successive control points, the more points the trajectory contains and the closer the optimal trajectory will be to the reference one, but the higher the computational complexity will be.

The present invention proposes to set the time-of-flight between adjacent checkpoints to be approximately equal to or somewhat greater than the average maximum processing time of the global scheduling module. Thus, in one iteration, the global planning module manages to process the current environmental conditions when approaching the next control point in order to correct the trajectory if necessary. This allows each successive control point to be taken into account without data loss and severe deviation from the trajectory, and on the other hand, without excessive processing. Thus, an increase in processing accuracy and flight safety is ensured.

In practice, when flying over obstacles, the distance between checkpoints can increase, and the predetermined constant processing period can cause the iteration start to be at a different distance from the nearest point, which may cause loss of synchronization. To avoid this, in one embodiment, processing is triggered by an event. For example, when approaching a checkpoint at a predetermined distance (or predetermined approach time), at least one module (eg, a global planning module) may start a new iteration if the processing of the previous iteration has already been previously completed. This makes it possible to take into account each successive control point without data loss and severe deviation from the trajectory, as well as improve adaptability to changing environmental conditions. Thus, an increase in processing accuracy and flight safety is ensured.

When describing the operation of the local planning module, the expressions that are calculated during processing in this module were indicated above. Based on these expressions and the well-known features of the NMPC method, it would be logical to assume that the larger the planning horizon, that is, the number of control points that must be processed by the local planning module, the more difficult the processing and the longer one iteration takes. However, tests of the proposed invention unexpectedly showed that with planning horizons equal to about 4.5-6 (preferably 5) UAV diameters (counting from the current location Q _p , that is, from the center of the UAV), compared with smaller and larger horizons planning provides an increase in the frequency of successful (without failures) execution of a flight task and even a decrease in the average processing time while maintaining a similar total distance traveled by the UAV from the start to the end point.

Example

For testing the present invention, a DL Matrice 100 UAV was used, which contained the following modules: a module for generating a 3D point cloud based on a Velodyne Lite 16 lidar attached to the UAV for environmental analysis, display modules, generation of a reference trajectory, trajectory planning and a PD controller based on an attached Velodyne Lite 16 to the UAV of the Nvidia Jetson NX processor, the GPS navigation module built into the UAV and the DL A3 flight controller. The maximum UAV speed was set to 1.2 m/s, planning horizon 4 m, time between fixes 0.05 s, avoidance distance 0.8 m, initial reference trajectory 54.3 m. obstacles on the way was 79.8 m, the total flight time was about 150 s, the average deviation from the trajectory was less than 1 m during the entire flight.

Application

Devices, systems and methods according to the present invention can be used for autonomous movement along the reference trajectory of UAVs performing cargo delivery, various studies, analysis of the surrounding space, object control, photo and video filming, rescue operations, autonomous monitoring for the tasks of protecting territories, farmland, cartography , remote chemical-physical analysis, control of germination and ripeness of the crop, chemical treatment, etc.

It is also possible to use the present invention in manned aircraft in the mode of autonomous movement (autopilot).

Additional Implementation Features

Various illustrative blocks and modules described in connection with the disclosure herein may be implemented or executed by a general purpose processor, digital signal processor (DSP), application specific integrated circuit (ASIC), field programmable gate array (FPGA), or other programmable logic device (PLD), discrete logic element or transistor logic, discrete hardware components, or any combination of the foregoing, designed to perform the functions described in this document. A general purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. A processor may also be implemented as a combination of computing devices (eg, a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors together with a DSP core, or any other similar configuration).

Some blocks or modules, individually or collectively, may be, for example, a computer, and include a processor that is configured to call and execute computer programs from memory to perform method steps or functions of blocks or modules in accordance with embodiments of the present invention. According to embodiments, the device may further include a memory. The processor may call and execute computer programs from memory to execute the method. The memory may be a separate device independent of the processor or may be integrated with the processor. The memory may store code, instructions, commands, and/or data for execution on a set of one or more processors of the device described. Codes, instructions, commands may direct the processor to perform the steps of a method or device function.

The functions described in this document may be implemented in hardware, software running on one or more processors, firmware, or any combination of the foregoing, as applicable. The hardware and software implementing the functions may also be physically located in different locations, including such a distribution that parts of the functions are implemented in different physical locations, that is, distributed processing or distributed computing may be performed.

If the amount of data is large, multi-threaded data processing can be performed, which in a simple representation can be expressed in the fact that the entire set of data to be processed is divided into a set of subsets, and each processor core performs processing in relation to its assigned data subset.

The above memory may be volatile or non-volatile memory, or may include both volatile and non-volatile memory. One of ordinary skill in the art will also understand that when referring to memory and storage of data, programs, codes, instructions, instructions, and the like, it is understood that there is a machine-readable (or computer-readable, processor-readable) storage medium. A computer-readable storage medium can be any available medium that can be used to carry or store a desired piece of program code in the form of instructions or data structures, and that can be accessed by a computer, processor, or other general purpose or special purpose processing device. .

By way of example, and not limitation, computer-readable media may include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electronically erasable programmable read-only memory (EEPROM), flash memory. , Random Access Memory (RAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDR SDRAM) , increased speed synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct access bus random access memory (DR RAM), and the like.

The information and signals described herein may be represented using any of a variety of different technologies. For example, the data, instructions, commands, information, signals, bits, symbols, and elementary signals that may be exemplified in the above description may be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combinations of the above.

The processor may include one or more processors. At the same time, one or more processors may be a general purpose processor such as a central processing unit (CPU), an application processor (AP) and the like, a graphics processing unit such as a graphics processing unit (GPU), a visual processing unit (VPU) etc.

It should be understood that although terms such as "first", "second", "third" and the like may be used herein to describe various elements, components, regions, layers and/or sections, these elements, components, areas, layers and/or sections should not be limited by these terms. These terms are only used to distinguish one element, component, region, layer, or section from another element, component, region, layer, or section. Thus, a first element, component, region, layer, or section may be referred to as a second element, component, region, layer, or section without departing from the scope of the present invention. As used herein, the term "and/or" includes any and all combinations of one or more of the respective listed positions. Elements mentioned in the singular do not exclude the plurality of elements, unless otherwise specified.

The functionality of an element specified in the description or claims as a single element may be practiced by means of several components of the device, and conversely, the functionality of elements indicated in the description or claims as several separate elements may be practiced by means of a single component.

In one embodiment, the elements/blocks/modules of the proposed device are in a common housing, can be placed on the same frame/structure/printed circuit board/crystal and are structurally connected to each other through assembly (assembly) operations and functionally through communication lines. The mentioned communication lines or channels, unless otherwise indicated, are typical communication lines known to specialists, the material implementation of which does not require creative efforts. The communication link may be a wire, a set of wires, a bus, a track, a wireless link (inductive, RF, infrared, ultrasonic, etc.). Communication protocols over communication lines are known to those skilled in the art and are not disclosed separately.

The functional connection of elements should be understood as a connection that ensures the correct interaction of these elements with each other and the implementation of one or another functionality of the elements. Particular examples of functional communication may be communication with the ability to exchange information, communication with the ability to transmit electric current, communication with the ability to transmit mechanical motion, communication with the ability to transmit light, sound, electromagnetic or mechanical vibrations, etc. The specific type of functional connection is determined by the nature of the interaction of the mentioned elements, and, unless otherwise indicated, is provided by well-known means, using principles well-known in the art.

Although the exemplary embodiments have been described in detail and shown in the accompanying drawings, it should be understood that such embodiments are illustrative only and are not intended to limit the present invention, and that the present invention should not be limited to the particular arrangements and structures shown and described, since a person skilled in the art on the basis of the information set forth in the description and the knowledge of the prior art may be obvious various other modifications and embodiments of the invention, without going beyond the essence and scope of this invention.

Claims (17)

1. A device for clarifying the trajectory of an aircraft (LA), containing: a display module configured to build a map of environmental objects based on a three-dimensional (3D) point cloud obtained from a 3D point cloud generation module contained in the aircraft, and estimate the distance from the aircraft to surrounding objects based on the map of environmental objects and aircraft location data received from the navigation module contained in the aircraft; a reference trajectory generation module configured to generate a reference trajectory to be followed by the aircraft by interpolating the trajectory built according to the flight task between the control points through which the aircraft must pass, taking into account the preferred moments of their passage, using a uniform B-spline 3- order; And a trajectory planning module configured to generate an updated trajectory of the aircraft movement around obstacles based on a reference trajectory, a map of environmental objects and data on the distance from the aircraft to surrounding objects and generate control signals for the aircraft based on the refined trajectory using a control method with a nonlinear predictive model. 2. The device according to claim 1, in which the global planning module is configured to generate an updated trajectory of the aircraft around obstacles if it is determined that the reference trajectory passes through the environmental object or that the aircraft or the reference trajectory is at such a distance from the environmental object , which is less than the preset avoidance distance. 3. The device according to claim 2, in which the location of the aircraft corresponds to the center of the aircraft, and the avoidance distance is greater than the radius of the circle describing the aircraft. 4. The device according to claim 1, in which the control signals for the aircraft contain velocity vectors in each of the x, y, z directions, as well as the yaw angle. 5. The device according to claim 1, further comprising a navigation module configured to determine the location, speed and yaw angle of the aircraft. 6. The apparatus of claim 1, further comprising a 3D point cloud generation module configured to generate a 3D point cloud. 7. The device according to claim 1, additionally containing a proportional-differential (PD) controller, configured to generate adjusted control signals for the aircraft and transmit them to the flight controller to control the aircraft actuators, and the PD controller uses data on the current state of the aircraft from the module navigation as feedback and control signals from the path planning module as input. 8. The device according to claim. 1, in which the global planning module is configured to generate an updated trajectory of the aircraft using the Broyden - Fletcher - Goldfarb - Shanno method with limited memory use in a multidimensional cube. 9. The device according to claim 2, in which the global planning module is configured to detect on the reference trajectory a segment of the reference trajectory to be changed, containing a set of successive control points, at least some of which are within the environmental object or at such a distance to the object of the environment that is less than a predetermined evasion distance, dividing the free space around this segment into a plurality of bulge fragments, and determining the locations of the refined segment control points within said bulge free space fragments by applying a parallel convex decomposition to each point of the segment to be modified, and generating a refined trajectory movement of the aircraft passing through the checkpoints of the refined segment. 10. The device according to claim 9, in which, if an attempt to generate a refined segment in a predetermined time is completed unsuccessfully, the global planning module is configured to present the entire reference trajectory not passed at the moment as a segment to be changed and generate an updated trajectory bypassing the obstacle using parallel convex decomposition. 11. A method for clarifying the trajectory of the movement of an aircraft, containing the steps, in which: using the display module, a map of environmental objects is constructed based on a three-dimensional (3D) point cloud obtained from the 3D point cloud generation module contained in the aircraft, and the distance from the aircraft to surrounding objects is estimated based on the map of environmental objects and aircraft location data, received from the navigation module contained in the aircraft; using the reference trajectory generation module, the generation of the reference trajectory to be followed by the aircraft is performed by interpolating the trajectory built according to the flight task between the control points through which the aircraft must pass, taking into account the preferred moments of their passage, using a homogeneous B-spline of the 3rd order; And using the trajectory planning module, the formation of an updated trajectory of the aircraft movement around obstacles based on the reference trajectory, a map of environmental objects and data on the distance from the aircraft to surrounding objects, and the formation of control signals for the aircraft based on the refined trajectory using the control method with a nonlinear predictive model.

Images