A rod driving tensegrity unmanned aerial vehicle and a restricted space navigation control method thereof

Abstract

The application discloses a rod driving tension whole unmanned plane and a navigation control method thereof in a limited space. The unmanned plane is composed of a six-rod tension whole and an eight-rotor wing subsystem, and the unmanned plane is driven to deform through rod length extension and contraction and elastic cable tension, so that a safe distance from environmental obstacles is kept during flight.

Description

Technical Field

[0001] This invention relates to the field of deformable unmanned aerial vehicle (UAV) technology, specifically to a rod-driven tensioned integral UAV and its navigation and control method in confined spaces. Background Technology

[0002] Rotary-wing drones have been widely used in surveying, inspection, and emergency rescue due to their simple structure and low cost. However, these flight platforms are usually integral rigid structures, and when operating in environments with limited space or dense obstacles, they often cannot pass through narrow areas due to insufficient space clearance.

[0003] To improve space adaptability, existing deformable UAVs generally employ mechanisms such as telescopic arms, tilt rotors, or folding frames. However, these solutions typically suffer from high structural complexity, large additional mass, low coupling between deformation and attitude control, and insufficient impact resistance, making it difficult to balance maneuverability and reliability. In contrast, the tensioned integral structure, composed of compression rigid rods and tension elastic cables forming a self-stressed grid, possesses characteristics such as light weight, high strength, and deformability. It can actively buffer collision impacts during flight and improve the platform's space adaptability.

[0004] However, existing tensioning drones mostly rely on passive deformation or elastic cable drive, lacking an active drive mechanism to precisely adjust the length of the rods, and the deformation decision and navigation control are independent of each other, making it difficult to meet the safety requirements of flight in confined spaces. Summary of the Invention

[0005] In view of this, the present invention provides a lever-driven tensioning integral unmanned aerial vehicle and its confined space navigation and control method, which can realize the requirements of autonomous attitude transformation and obstacle avoidance of the lever-driven tensioning integral unmanned aerial vehicle.

[0006] The rod-driven tensioning integrated unmanned aerial vehicle of the present invention includes three sets of rigid rods and multiple elastic cables; each set of rigid rods includes two symmetrical and parallel rigid rods centered on the unmanned aerial vehicle;

[0007] Among them, the third set of rigid rods V and VI are located on the horizontal plane of the drone and fixed on both sides of the drone; the first set of rigid rods I and II are adjustable in length, perpendicular to the horizontal plane of the drone, and not rigidly connected to the drone; the second set of rigid rods III and IV are fixed in length, perpendicular to the other two sets of rigid rods, and not rigidly connected to the drone.

[0008] Each rod end is connected to the rigid rod ends of other groups on the same side by an elastic cable, and the elastic cable is prestressed.

[0009] The present invention also provides a control method for the above-mentioned rod-driven tensioning integral UAV, comprising:

[0010] S1 divides the UAV into two subsystems: the rigid rod subsystem, which refers to the first set of adjustable rigid rods I and II and the second set of fixed-length rigid rods III and IV; and the octocopter subsystem, which refers to the UAV and the third set of fixed-length rigid rods V and VI that are fixed to it; and establishes the Newton-Euler dynamic equations for each subsystem.

[0011] S2, Generate a collision-free initial path, and construct a convex polyhedron safe flight corridor along the initial path; Model the UAV as a polyhedron with the ends of rigid rods as vertices, and transform the UAV obstacle avoidance constraint into a vertex-plane linear inequality;

[0012] S3, the initial path is reconstructed using MINCO polynomials, B-splines, and other methods to reconstruct its spatiotemporal parameters; with the optimization objective of minimizing the weighted sum of trajectory smoothness cost, total time cost, dynamic penalty cost, and obstacle avoidance penalty cost, the polynomial coefficients of the trajectory and the time allocation of each segment are optimized under the constraints of boundary conditions and continuity between trajectory segments to obtain a smooth and feasible reference trajectory and the corresponding adjustable rigid rod length sequence.

[0013] S4, the controller is designed to track the reference trajectory and the corresponding adjustable rigid rod length sequence to achieve control of the UAV.

[0014] Preferably, in S1, the dynamic equation of the rigid rod system is:

[0015]

[0016] Where k = 1, 2, 3, 4, representing the first group of adjustable rigid rods I and II, and the second group of fixed-length rigid rods III and IV, respectively; p k v k For a rigid rod k, the center of mass is in the world coordinate system. Position and velocity in Q k From the rigid rod k-body coordinate system To the world coordinate system unit quaternion, ω k Let f be the angular velocity of the rigid rod k in body coordinates. s,k m is the resultant force exerted by the elastic cable on the rigid rod k; k Let g be the mass of the rigid rod k; g be the acceleration due to gravity; e3 = [0 0 1] T ;⊙ represents Hamiltonian quaternion multiplication; J k Here is the inertia matrix of the rigid rod k:

[0017]

[0018] Among them l k r is the real-time length of the rigid rod k; kLet be the radius of the rigid rod k; diag denotes a diagonal matrix;

[0019] The dynamic equations of the octagonal airfoil system are:

[0020]

[0021] Where, p db and v db The eight-rotor subsystem in the world coordinate system The position and velocity of the center of mass; Q db From the body coordinate system The unit quaternion to the world coordinate system W; ω db For the octagonal subsystem in the airframe coordinate system Angular velocity at the bottom; R db For the quaternion Q db The determined rotation matrix; f s,db With τ s,db These are the resultant force and resultant torque exerted by the elastic cable on the octagonal subsystem, respectively; f T and τ T These represent the total thrust and total torque generated by the propeller, respectively; J db is the inertial matrix of the octagonal subsystem.

[0022] A better approach is to first calculate the tension of each elastic cable according to Hooke's Law, and then calculate the resultant force and resultant torque exerted by the elastic cables on the rigid bar k / octagonal subsystem.

[0023] Preferably, in step S2, the JPS algorithm, A* algorithm, RRT algorithm, etc. are used to generate a collision-free initial path.

[0024] Preferably, in S2, a series of convex polyhedral safe flight corridors are constructed using methods such as RILS, IRIS, and FIRI.

[0025] Preferably, in step S3, the optimization objective and constraints are as follows:

[0026]

[0027] Where i = 1, 2, ..., M is the i-th initial path reconstructed using MINCO polynomials; T represents the s-th derivative of the initial path segment i; i ,T i-1 λ represents the duration of the ith segment and the (i-1)th segment of the trajectory, respectively; "||||2" represents the L2 norm; t represents time; λ T , λ f , λ c The weights for time duration, dynamic feasibility, and obstacle avoidance cost are respectively. ωi F represents velocity, acceleration, and angular velocity of the aircraft. max Its upper limit; p v This represents the position of the v-th node of the entire tensioning system. Let j be a point on the j-th safe flight corridor (SFC) polyhedron of the i-th trajectory segment, where j = 1, 2, ..., N. i N i Let i be the face of the SFC polyhedron corresponding to the i-th trajectory segment. Let be the unit normal vector of the j-th SFC polyhedron face of the i-th trajectory segment; function It only takes effect when constraints are violated (i.e., the penalty term is automatically activated when the system state exceeds the constraint boundary; otherwise, it remains at zero and does not affect the optimization process). Let be the d-th derivative vector of the i-th trajectory segment; The derivative states of the first trajectory segment at the initial moment are shown. For the final state; t M This is the final time of the entire trajectory; These are the SFC polyhedral constraint regions of the ith segment and the (i+1)th segment of the trajectory, respectively.

[0028] A better approach is to rewrite the original constrained optimization problem into an optimization model without explicit inequalities, containing only a smoothing penalty term; then, the L-BFGS algorithm is used to iteratively solve for the polynomial coefficients of the trajectory and the time allocation for each segment.

[0029] Preferably, in step S4, PID control, NMPC control, or manifold-based model predictive control are used to track the reference trajectory and the corresponding adjustable rigid bar length sequence.

[0030] Preferably, a manifold-based model predictive control framework is used to track the reference trajectory and the corresponding adjustable rigid rod length sequence, specifically:

[0031] The unmanned aerial vehicle (UAV) system state resides on a composite manifold:

[0032]

[0033] in, It is a composite state manifold, formed by the direct product of position space, velocity space, attitude space, and rod length space; x is the system state vector, which includes the UAV's center of mass position, center of mass velocity, body attitude, and adjustable rigid rod length; u is the system control input vector, which includes total thrust, body angular velocity, and rod length rate of change; dim() is the manifold dimension; Let be the real number field, representing the Euclidean space; SO(3) is a three-dimensional special orthogonal group; l = [l1 l2] T Let l be the real-time length of adjustable rigid rods I and II, where l = [l1 l2] TThe extension and retraction speeds of adjustable rigid rods I and II;

[0034] The controller selects local coordinate errors:

[0035]

[0036] in, Represents manifold subtraction; x d p is the desired reference state vector; d v d R d , l d , respectively, represent the desired position, desired velocity, desired rotation matrix, and desired rod length vector of the octagonal subsystem; p, v, and R represent the actual position, actual velocity, and actual rotation matrix of the octagonal subsystem, respectively. This is a logarithmic mapping from a rotation matrix to exponential coordinates, and the inverse mapping is an exponential mapping Exp(·).

[0037] Control input error is

[0038]

[0039] Among them, u d The desired control input is calculated based on the reference trajectory and the corresponding adjustable rigid rod length sequence, according to the differential flatness property of the rotary-wing UAV.

[0040] The tracking control problem of the reference trajectory and the corresponding adjustable rigid rod length sequence is formulated as the following constrained finite-time quadratic optimization problem:

[0041]

[0042] Where N represents the prediction time domain length; P κ ≥0, P N ≥0, H κ ≥0 represent the penalty matrices for the stage state, terminal state, and control input, respectively; x N To predict the state of the time-domain terminal; These represent the state transition Jacobian matrix and the control input Jacobian matrix of the system error state equation, respectively; x0 represents the actual system state at the current moment; x init This represents the initial state of the system at the current moment. As an input error interval constraint, u min u max These are the lower and upper physical limits of the control input vector, respectively; The expected control input reference value for the κ-th prediction step;

[0043] The optimal control sequence δu is obtained by solving in each prediction time domain. ＊ Take δu ＊ The first item With reference input The actual execution command is generated by superimposing the commands, i.e.:

[0044]

[0045] Beneficial effects:

[0046] 1. This invention proposes a rod-driven deformable UAV architecture consisting of a six-bar tensioned main body and an eight-rotor subsystem. By extending and retracting the rods and tensioning the elastic cables, the UAV is driven to deform, thereby maintaining a safe distance from environmental obstacles during flight.

[0047] 2. This invention first establishes a dynamic model based on the Newton-Euler equations, encompassing the coupling of rod length extension, elastic cable tension, and propeller thrust, providing an accurate mathematical foundation for describing the changes in inertia and force during deformation. Then, it introduces a planning framework combining a regular icosahedral safety envelope and a convex polyhedral flight corridor, transforming the obstacle avoidance constraints of the entire aircraft into vertex-plane linear inequalities. Based on MINCO polynomial trajectory optimization, it synchronously decides the flight path and rod length changes, thereby maintaining a safe distance from environmental obstacles throughout the entire deformation process.

[0048] 3. This invention employs an On-Manifold MPC method based on composite manifolds, which considers inertial changes caused by deformation in real time and combines the output thrust with the rod extension rate. This achieves high-precision tracking of the reference trajectory and ensures that all actuator inputs always meet physical constraints. Attached Figure Description

[0049] Figure 1 This is a schematic diagram of the rod-driven tensioning integral UAV mechanical mechanism of the present invention;

[0050] Figure 2 This is a schematic diagram of the integrated planning and control system framework for the deformable tensionable integral unmanned aerial vehicle (UAV) of the present invention.

[0051] Figure 3 This is a schematic diagram of the coordinate system of the rod-driven tensioning integral UAV of the present invention;

[0052] Figure 4 This is a schematic diagram of the active deformation sequence of the deformable tensioning integral UAV of the present invention in a narrow channel. Detailed Implementation

[0053] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0054] This invention provides a rod-driven tensioning integral unmanned aerial vehicle and its navigation and control method in confined space.

[0055] like Figure 1 As shown, the rod-driven tensioned integrated UAV of this embodiment consists of a central octagonal UAV and a six-rod tensioned integrated frame surrounding it. The frame consists of six compression rigid rods 1-6 and twenty-four tension elastic cables 7. The six rigid rods are divided into three groups, with two parallel rigid rods in each group and perpendicular to each other between different groups: the first group consists of adjustable rigid rods 1 and 2, with servo motor-synchronous belt drive mechanisms 8 on the sides of the two rigid rods, which can linearly extend and retract within a predetermined stroke and are not rigidly connected to the central octagonal UAV; the second group consists of fixed-length rigid rods 3 and 4, which are also not rigidly connected to the central octagonal UAV; the third group consists of fixed-length rigid rods 5 and 6, which are fixedly connected to the central octagonal UAV to form an "octagonal subsystem". The twenty-four tension elastic cables 7 cross-connect the ends of each rigid rod, and each rod end is flexibly connected to the four rod ends closest to it in space by four elastic cables, continuously applying prestress to maintain the geometry of the frame and absorb impact energy during collisions. By driving the first set of adjustable rigid rods to change their length in real time, the drone's dimensions can be adjusted during flight to meet the needs of passage and obstacle avoidance in narrow spaces.

[0056] like Figure 2 As shown, to achieve stable flight and safe navigation of the deformable UAV during its deformation process, it is necessary to accurately describe its motion laws and design corresponding control strategies. This embodiment first establishes a dynamic model of the system, and then designs motion planning and control methods based on this model.

[0057] Step 1: Dynamic Model of Deformable Tensioned Integral Unmanned Aerial Vehicle

[0058] To accurately describe the spatial motion of the deformable tensioned integral UAV under the influence of rigid rod extension and contraction and elastic cable tension, this embodiment divides the body into two types of rigid bodies: one is the "rigid rod," referring to the first set of adjustable rigid rods I and II and the second set of fixed-length rigid rods III and IV; the other is the "octocopter subsystem," referring to the central octocopter UAV and the third set of fixed-length rigid rods V and VI fixedly connected to it. Newton-Euler dynamic equations are established for each of these two types of rigid bodies.

[0059] An inertial reference frame fixed to the ground is defined as the world coordinate system. (Origin point: a fixed point on the ground, such as the takeoff point;) Axis: Pointing towards geographical north or a specific reference direction; Axis: pointing eastward, in line with The axes form a horizontal plane; Axis: pointing in the opposite direction to the Earth's center (i.e., vertically upward, conforming to the right-hand rule), the coordinate system fixed to the center of mass of the k-th rigid rod and moving rigidly with that rod is defined as the body coordinate system. (Origin: fixed to the center of mass of the rigid body; X-axis: points in the direction of movement of the rigid body; Y-axis: points to the left side of the rigid body; Z-axis: points above the rigid body, conforming to the right-hand rule). The state variables of the first set of adjustable rigid bars I and II are defined as follows:

[0060]

[0061] Where, p k v k For the adjustable rigid rod k, the center of mass is in the world coordinate system. Position and velocity in Q k ∈S 3 Represents the k-body coordinate system of the adjustable rigid rod To the world coordinate system unit quaternions, For volume coordinate angular velocity, l k Let l be the real-time length of the rigid rod. This embodiment assumes that the rotation of the adjustable rigid rod about its own longitudinal axis is negligible, and l k It extends and retracts at a constant speed via a servo motor.

[0062] The tension in the elastic cable j connected to both ends of the adjustable rigid rod satisfies Hooke's Law:

[0063] t j =max(0,K) j (||s j ||-L j ))

[0064] Where K j Let L be the stiffness coefficient of the j-th elastic cable. j Let j be the original length of the j-th elastic cable. Let f be the direction vector between the two endpoints of the elastic cable j, and "||||" denotes the L2 norm. The resultant force f s,k With resultant moment τ s,k The calculation formula is

[0065]

[0066] Where, n j =s j / ||s j || is the unit direction vector of the elastic cable j; R is the set of indices of all elastic cables connected to the k-th adjustable rigid rod; k From the adjustable rigid rod k-body coordinate system B k The rotation matrix to the world coordinate system W, the value of which is determined by the attitude quaternion Q. k Uniquely certain; This is the vector from the center of mass of the k-th adjustable rigid bar to the endpoint of the elastic cable. The dynamic equation of the adjustable rigid bar is:

[0067]

[0068] Where e3 = [0 0 1] T ⊙ represents Hamiltonian quaternion multiplication, and the inertia matrix of the k-th adjustable rigid rod is:

[0069]

[0070] Where m k and l k These represent the mass and length of the k-th adjustable rigid rod, respectively; r k Let be the radius of the k-th adjustable rigid rod; diag represents the diagonal matrix.

[0071] For the second group of non-adjustable rigid rods, the length l of the kth non-adjustable rigid rod is... k The state variable is constant at l0, does not contain any degree of freedom for scaling, and is defined as follows:

[0072]

[0073] The dynamic equation of the non-adjustable rigid rod is consistent with that of the adjustable rigid rod, except that the length-related term (l) is omitted. k (Item), the inertia matrix is ​​a constant matrix, and it is related to the adjustable rigid rod in l k The same applies under the condition of =l0, and the specific form will not be listed here.

[0074] In this embodiment, the octocopter subsystem is considered as a rigid body, such as Figure 3 As shown, the coordinate system fixed to the center of mass of the octagonal subsystem and moving with the subsystem is defined as the body coordinate system B. db (Origin: fixed to the center of mass of the rigid body; X-axis: points in the direction of movement of the rigid body; Y-axis: points to the left side of the rigid body; Z-axis: points above the rigid body, conforming to the right-hand rule), the state variables are:

[0075]

[0076] Satisfying the dynamic equations:

[0077]

[0078] Where, p db This represents the position of the center of mass of the octagonal airfoil subsystem in the world coordinate system W, v db For its center-of-mass velocity, Q db ∈S 3 The unit quaternion represents the octagonal subsystem from the body coordinate system B. dbAttitude transformation to world coordinate system W R is the angular velocity of the octagonal subsystem in the body coordinate system. db ∈SO(3) is a quaternion Q db The determined rotation matrix; g is the gravitational acceleration constant, f s,db With τ s,db These are the resultant force and resultant torque exerted by the elastic cable on the octagonal subsystem, f. T τ is the total thrust generated by the propeller. T J is the total torque generated by the propeller. db is the inertial matrix of the octagonal subsystem.

[0079] Step 2: Motion Planning of Deformable Tensioned Integral Unmanned Aerial Vehicle

[0080] This embodiment proposes a planning framework based on the integration of geometric parameters and dynamics, which unifies the modeling of UAV dimensions, rigid rod extension range, attitude constraints, and environmental safety distance. By solving the constraint optimization problem, a reference trajectory that balances safety and performance indicators is generated, providing reference input for subsequent attitude control and dynamic tracking.

[0081] In this embodiment, the front-end planning first generates collision avoidance paths in a 3D raster map using methods such as JPS, A*, and RRT*. Subsequently, a series of convex polyhedral safe flight corridors were constructed using methods such as RILS, IRIS, and FIRI. Each C i By N i Each hyperplane defines:

[0082]

[0083] in, Let the unit normal vector of the j-th hyperplane be denoted as . It is the reference point on the hyperplane.

[0084] like Figure 4 As shown, this embodiment models the tensioned integral aircraft as an icosahedron to achieve obstacle avoidance for the entire aircraft. An icosahedron consists of 20 triangular faces and 12 vertices, naturally reflecting the geometric characteristics of a six-bar tensioned integral structure. At time t, the position of the v-th vertex is defined as:

[0085]

[0086] in, In body coordinate system B db The direction vector from the center of mass of the octagonal subsystem to the v-th vertex. For the four vertices associated with the two adjustable rigid rods, Variation with the length of the rigid rod:

[0087]

[0088] Where l0 is the initial length of the non-adjustable rigid rod, l k Let be the real-time length of the k-th adjustable rigid rod.

[0089] In the backend optimization of this embodiment, trajectory classes such as MINCO polynomials, B-splines, and Bézier curves are used to reconstruct the spatiotemporal parameters of the initial path:

[0090]

[0091] Where p(t) represents a polynomial trajectory with M segments, dimension m, and order N = 2s⁻¹. This is a sequence of intermediate control points. The time allocation for each trajectory segment is given, where c is the polynomial coefficient.

[0092] Considering energy consumption, time, dynamic feasibility, and obstacle avoidance constraints, the optimization problem can be written as:

[0093]

[0094] Where, p (s) Let λ represent the s-order derivative of the trajectory. T , λ f , λ c These are the weights for time duration, dynamic feasibility, and obstacle avoidance cost, respectively. ω i F represents velocity, acceleration, and angular velocity of the aircraft. max Its upper limit; function The penalty term only takes effect when a constraint is violated; that is, it is automatically activated when the system state exceeds the constraint boundary. Otherwise, it remains at zero and does not affect the optimization process. Constraints (0 guarantees the continuity of the d = s⁻¹ derivative between segments) Ensure the trajectory always lies within an adjacent corridor and that the time strictly increases; boundary conditions The initial and final states are fixed.

[0095] After the cost function and constraints are determined, this embodiment follows the MINCO framework, rewriting the original constrained optimization problem into an optimization model without explicit inequalities, containing only a smoothing penalty term. Subsequently, the L-BFGS algorithm is used to iteratively update the control point parameters c and the piecewise duration T until the objective function converges and all constraints are satisfied. The final output piecewise duration vector and its corresponding polynomial coefficients together define the centroid reference state trajectory x of the octagonal subsystem. d (t), and also give the target length sequence l of the adjustable rigid rod. k(t), both of which together satisfy obstacle avoidance requirements and geometric deformation constraints, and can be directly called by the model predictive controller.

[0096] Step 3: Motion Control of the Deformable Tensioned Integral Unmanned Aerial Vehicle

[0097] This embodiment obtains the reference state trajectory x of the centroid of the octagonal subsystem. d (t) and the adjustable rigid rod length sequence l k After (t), based on the differential flatness property of the rotary-wing UAV, the desired thrust and attitude command are first calculated, and a nonlinear controller is designed to adjust the desired command, thereby accurately tracking the trajectory and shape.

[0098] This embodiment employs an On-Manifold MPC framework based on composite manifolds to apply the motion planning to the centroid reference state trajectory x of the eight-rotor subsystem. d (t) and the adjustable rigid rod length sequence l k (t) Achieve real-time and accurate tracking. PID, NMPC, and other control frameworks can also be used.

[0099] The deformable tensioned integral unmanned aerial vehicle system is located on a composite manifold:

[0100]

[0101] Where, l = [l1 l2] T Let l be the length of two adjustable rigid rods, where l = [l1 l2] T The extension and retraction speed of the rigid rod is adjustable.

[0102] To perform optimization in Euclidean vector space, the controller selects local coordinate errors:

[0103]

[0104] in The rotation matrix is ​​a logarithmic mapping to exponential coordinates, and the inverse mapping is an exponential mapping Exp(·). The control input error is defined as... Desired control input u d The calculation is performed based on the differential flatness property of rotary-wing UAVs.

[0105] Discretize the dynamics of a continuous system:

[0106]

[0107] symbol The manifold "addition" is represented. By linearizing the discrete system to the first order at the reference trajectory, the error dynamics are obtained:

[0108]

[0109] in and These are the Jacobian matrices of the system dynamics function at the reference point with respect to the state and input, respectively.

[0110] Ultimately, the trajectory tracking control problem is formulated as a constrained finite-time quadratic optimization problem:

[0111]

[0112] In the objective function, ||δx|| k To penalize state tracking errors (such as position, velocity, and attitude deviations), the goal is to bring the UAV's state as close to the desired value as quickly as possible; δu k ‖To penalize the magnitude of changes in the control input, ensuring a smooth control process and saving energy;‖δx N To predict the tracking error of the terminal during the predicted time period and ensure the stability of trajectory tracking.

[0113] In the constraints, δx k+1 As a dynamic constraint, it ensures that the optimized control commands must conform to the motion laws of the UAV (i.e., state equations), guaranteeing that the solution is physically feasible. δx0 is the initial condition constraint, and optimization must start from the actual state error measured at the current moment; δuk is the control input constraint, which represents the physical limits of the actuators (motor thrust, servo speed, etc.), ensuring that the calculated control commands are actually executable.

[0114] N represents the prediction time domain length; P κ ≥0, P N ≥0, H κ ≥0 represent the penalty matrices for the stage state, terminal state, and control input, respectively. An interval constraint is introduced for the input error. This quadratic programming approach satisfies the actuator's physical constraints regarding thrust amplitude and adjustable rigid rod drive rate during optimization. Within each prediction time domain, the MPC solution yields the optimal control sequence δu for the propeller's total thrust, angular velocity, and rod extension / retraction speed. ＊ Only the first item is taken. With reference input The actual execution command is generated by superimposing the commands, i.e.:

[0115]

[0116] This strategy achieves high trajectory tracking accuracy and automatically compensates for inertial changes caused by deformation, while ensuring that the actuator input is always within physical constraints.

[0117] In summary, the above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A control method for a lever-driven tensioning integral unmanned aerial vehicle, characterized in that, include: S1 divides the UAV into two subsystems: the rigid rod subsystem, which refers to the first set of adjustable rigid rods I and II and the second set of fixed-length rigid rods III and IV; and the octocopter subsystem, which refers to the UAV and the third set of fixed-length rigid rods V and VI that are fixed to it; and establishes the Newton-Euler dynamic equations for each subsystem. Each group of rigid rods includes two symmetrical, parallel rigid rods centered on the drone; the third group of rigid rods V and VI are located on the horizontal plane of the drone and fixed to both sides of the drone; the first group of rigid rods I and II have adjustable lengths, are perpendicular to the horizontal plane of the drone, and have no rigid connection to the drone; the second group of rigid rods III and IV have fixed lengths, are perpendicular to the other two groups of rigid rods, and have no rigid connection to the drone; each rod end is connected to the end of the rigid rods of the other group on the same side by an elastic cable, and the elastic cable is prestressed; S2, Generate a collision-free initial path, and construct a convex polyhedron safe flight corridor along the initial path; Model the UAV as a polyhedron with the ends of rigid rods as vertices, and transform the UAV obstacle avoidance constraint into a vertex-plane linear inequality; S3, reconstruct the spatiotemporal parameters of the initial path; with the optimization objective of minimizing the weighted sum of trajectory smoothness cost, total time cost, dynamic penalty cost and obstacle avoidance penalty cost, optimize the polynomial coefficients of the trajectory and the time allocation of each segment under the constraints of boundary conditions and continuity between trajectory segments, to obtain a smooth and feasible reference trajectory and the corresponding adjustable rigid rod length sequence. The optimization objectives and constraints are as follows: in, i =1,2,…, M For the first polynomial reconstructed using MINCO i Initial path of the segment; Indicates the first i Initial path of segment s First derivative; The first i Section and the i -1 represents the duration of the trajectory segment; "|| ||2" represents the L2 norm; t Indicates time; λ T , λ f , λ c The weights for time duration, dynamic feasibility, and obstacle avoidance cost are respectively. Indicates velocity, acceleration, and angular velocity of the body. F max Its upper limit; For the first tensioning of the whole v Each node position; For the first i The first segment of the trajectory j Points on the polyhedron of a Safe Flight Corridor (SFC), j =1,2,…, , For the first i The face of the SFC polyhedron corresponding to the segment trajectory. For the first i The first segment of the trajectory j Unit normal vectors of the faces of an SFC polyhedron; function It only takes effect when the constraint is violated; For the first i Segment trajectory d First derivative vector; The derivative states of the first trajectory segment at the initial moment are shown. This is the final state; This is the final time of the entire trajectory; The first i Section and the i +1 segment of trajectory SFC polyhedral constraint region; S4, the controller is designed to track the reference trajectory and the corresponding adjustable rigid rod length sequence to achieve control of the UAV; the controller adopts PID control, NMPC control or manifold-based model predictive control.

2. The control method as described in claim 1, characterized in that, In S1, the dynamic equation of the rigid rod system is: ； in, k =1,2,3,4, representing the first set of adjustable rigid rods I and II and the second set of fixed-length rigid rods III and IV, respectively; 、 rigid rod k Center of mass in world coordinate system Position and velocity in the middle, To from rigid rod k body coordinate system k To the world coordinate system unit quaternions, rigid rod k Volume coordinate angular velocity, For elastic cable to rigid bar k The resultant force applied; m k rigid rod k The quality; g It is the acceleration due to gravity; ; Hamiltonian quaternion multiplication; rigid rod k Inertia matrix: in l k rigid rod k The real-time length; rigid rod k The radius; diag represents a diagonal matrix; The dynamic equations of the octagonal airfoil system are: in, and The eight-rotor subsystem in the world coordinate system The position and velocity of the center of mass; From the body coordinate system db The unit quaternion in the world coordinate system W; For the octagonal subsystem in the airframe coordinate system db angular velocity at the bottom; For quaternions The determined rotation matrix; and These are the resultant force and resultant torque exerted by the elastic cable on the octocopter subsystem, respectively. and These represent the total thrust and total torque generated by the propeller, respectively. is the inertial matrix of the octagonal subsystem.

3. The control method as described in claim 2, characterized in that, First, calculate the tension of each elastic cable according to Hooke's Law, and then calculate the tension of the elastic cable on the rigid bar. k / Resultant force and resultant torque applied by the octocopter subsystem.

4. The control method as described in claim 1 or 2, characterized in that, In S2, the JPS algorithm, A* algorithm, or RRT algorithm is used to generate a collision-free initial path.

5. The control method as described in claim 1 or 2, characterized in that, In S2, a series of convex polyhedral safe flight corridors are constructed using the RILS method, IRIS method, or FIRI method.

6. The control method as described in claim 1, characterized in that, The original constrained optimization problem is rewritten as an optimization model without explicit inequalities, containing only a smoothing penalty term; then the L-BFGS algorithm is used to iteratively solve for the polynomial coefficients of the trajectory and the time allocation of each segment.

7. The control method as described in claim 1, characterized in that, A manifold-based model predictive control framework is used to track the reference trajectory and the corresponding adjustable rigid rod length sequence, specifically as follows: The unmanned aerial vehicle (UAV) system state resides on a composite manifold: in, It is a composite state manifold, which is formed by the direct product of position space, velocity space, attitude space and rod length space; The system state vector includes the UAV's center of mass position, center of mass velocity, body attitude, and adjustable rigid rod length. The system control input vector includes total thrust, body angular velocity, and rod length change rate; For manifold dimension; Let be the field of real numbers, and denote the Euclidean space; It is a special orthogonal group in three dimensions; This refers to the real-time lengths of adjustable rigid rods I and II. The extension and retraction speeds of adjustable rigid rods I and II; The controller selects local coordinate error: in, Represents manifold subtraction; The desired reference state vector; These are the desired position of the octagonal subsystem, the desired velocity of the octagonal subsystem, the desired rotation matrix of the octagonal subsystem, and the desired rod length vector, respectively. These are the actual position of the octagonal subsystem, the actual velocity of the octagonal subsystem, and the actual rotation matrix of the octagonal subsystem, respectively. The rotation matrix is ​​a logarithmic mapping to exponential coordinates, and the inverse mapping is an exponential mapping. ; Control input error is ， in, u d The desired control input is calculated based on the reference trajectory and the corresponding adjustable rigid rod length sequence, according to the differential flatness property of the rotary-wing UAV. The tracking control problem of the reference trajectory and the corresponding adjustable rigid rod length sequence is formulated as the following constrained finite-time quadratic optimization problem: in, N Indicates the length of the prediction time domain; P ≥0、 P N ≥0、 H ≥0 represent the penalty matrices for stage state, terminal state, and control input, respectively; To predict the state of the time-domain terminal; These are the state transition Jacobian matrix and the control input Jacobian matrix of the system error state equation, respectively. This represents the current actual state of the system. This represents the initial state of the system at the current moment. As an input error interval constraint, ; These are the lower and upper physical limits of the control input vector, respectively; For the first κ The expected control input reference value for each prediction step; The optimal control sequence is obtained in each prediction time domain. ;Pick The first item With reference input The actual execution command is generated by superimposing the commands, i.e.: 。 8. A rod-driven tensioning integrated unmanned aerial vehicle, characterized in that, It includes three sets of rigid rods and multiple elastic cables; each set of rigid rods contains two symmetrical, parallel rigid rods centered on the UAV; Among them, the third set of rigid rods V and VI are located on the horizontal plane of the drone and fixed on both sides of the drone; the first set of rigid rods I and II are adjustable in length, perpendicular to the horizontal plane of the drone, and not rigidly connected to the drone; the second set of rigid rods III and IV are fixed in length, perpendicular to the other two sets of rigid rods, and not rigidly connected to the drone. Each rod end is connected to the rigid rod ends of other groups on the same side by an elastic cable, and the elastic cable is prestressed. The drone is controlled by the method described in any one of claims 1 to 7.

Images