Air-ground amphibious platform autonomous landing planning method in unstructured environment

Abstract

The invention discloses a land-air amphibious platform autonomous landing planning method in an unstructured environment, and relates to the field of autonomous navigation and trajectory planning. The method comprises the following steps: establishing a platform autonomous landing stage kinetic model fused with a ground effect thrust gain sub-model; based on the model and starting point, terminal point and obstacle information, state sampling is carried out to generate an initial reference path, and a safe flight corridor is constructed; constructing a multi-constraint spatio-temporal joint trajectory optimization problem containing a ground effect penalty term by taking a safe flight corridor as a constraint; and solving a trajectory optimization problem to obtain an optimal landing trajectory and outputting the optimal landing trajectory. According to the method, active modeling and ground effect compensation are carried out on the planning layer, meanwhile, the problems of air obstacle avoidance and near-ground stability are cooperatively processed, and safe, stable and accurate autonomous landing of the air-ground amphibious platform under the complex rugged terrain is achieved.

Description

Technical Field

[0001] This application relates to the field of autonomous navigation and trajectory planning technology, and in particular to an autonomous landing planning method for amphibious platforms in unstructured environments. Background Technology

[0002] With the rapid development of unmanned systems technology, amphibious platforms, with their unique cross-domain mobility, have demonstrated enormous potential in fields such as reconnaissance in complex environments and disaster relief. One of the key aspects of achieving fully autonomous operation for such platforms is enabling them to safely and accurately land autonomously from the air into unknown, rugged, and unstructured land environments.

[0003] Unlike conventional aircraft that land on flat runways, amphibious platforms face complex terrain in their landing areas, presenting two core challenges: first, avoiding randomly distributed obstacles to ensure flight safety; and second, experiencing significant ground effects as they approach uneven ground, leading to abnormal lift and decreased stability, thus affecting landing accuracy and safety. Existing autonomous landing planning methods largely focus on path search and obstacle avoidance, or assume flat ground for trajectory optimization, lacking proactive consideration and compensation for near-ground aerodynamic effects (ground effects). A few methods that do consider complex terrain fail to systematically integrate ground effect models into the planning layer for collaborative optimization. This makes it difficult for existing methods to simultaneously guarantee air safety and near-ground stability during amphibious platform landings in unstructured environments, limiting their practical application effectiveness. Summary of the Invention

[0004] The purpose of this application is to provide an autonomous landing planning method for amphibious platforms in unstructured environments, which can solve the problem that existing technologies are unable to coordinate the protection of platform air safety and near-ground stability in unstructured environments, and achieve safe, stable and precise autonomous landing.

[0005] To achieve the above objectives, this application provides the following solution.

[0006] In one aspect, this application provides an autonomous landing planning method for amphibious platforms in unstructured environments, comprising the following steps.

[0007] A dynamic model for the autonomous landing phase of an amphibious platform is established. The dynamic model includes: a rigid body dynamic model, a near-ground model, and a differential flat dynamic model. The near-ground model includes: a track passive buffer mechanism, a terrain surface sub-model, and a ground effect thrust gain sub-model.

[0008] Acquire the initial status of the amphibious platform, target landing point information, and environmental obstacle information.

[0009] Based on the initial state, the target landing point information, and the environmental obstacle information, state sampling is performed to generate an initial reference path from the starting point to the ending point.

[0010] A safe flight corridor is constructed based on the initial reference path.

[0011] Based on the dynamic model of the autonomous landing phase of the amphibious platform, and taking the safe flight corridor as the spatial obstacle avoidance constraint, a multi-constraint spatiotemporal joint trajectory optimization problem including a ground effect penalty term is constructed.

[0012] Solving the trajectory optimization problem yields the optimal landing trajectory that satisfies both dynamic and terminal constraints. The dynamic constraints include velocity, acceleration, angular velocity, and thrust constraints. The terminal constraints include terminal velocity and terminal attitude constraints.

[0013] The optimal landing trajectory is output to control the amphibious platform to perform an autonomous landing.

[0014] In a second aspect, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the autonomous landing planning method for an amphibious platform in an unstructured environment as described above.

[0015] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the autonomous landing planning method for an amphibious platform in an unstructured environment as described above.

[0016] Fourthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the autonomous landing planning method for an amphibious platform in an unstructured environment as described above.

[0017] According to the specific embodiments provided in this application, this application has the following technical effects.

[0018] This application provides an autonomous landing planning method for amphibious platforms in unstructured environments. By establishing a dynamic model and terrain model that includes a ground effect model, a realistic physical basis is provided for planning. Through state sampling and safe corridor construction, random obstacles in the environment are quickly addressed, generating an initial safe passage and efficiently determining the flight feasible area in complex obstacle environments. In particular, by constructing a multi-constraint spatiotemporal joint optimization problem that includes a ground effect penalty term, the planner can actively predict and compensate for aerodynamic disturbances during near-ground flight when generating the trajectory, thereby synergistically optimizing airborne obstacle avoidance safety and near-ground landing stability. Finally, this method is the first to uniformly address the two major challenges of obstacle avoidance and ground effect resistance at the planning level, overcoming the shortcomings of the separation of these two aspects in existing technologies. It enables amphibious platforms to land safely, smoothly, and accurately in unstructured environments with rugged terrain and scattered obstacles, improving the platform's cross-domain operation capability and mission reliability. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is an application environment diagram of an autonomous landing planning method for a land-air amphibious platform in an unstructured environment, according to one embodiment of this application.

[0021] Figure 2 This is a flowchart illustrating an autonomous landing planning method for an amphibious platform in an unstructured environment, as provided in this application.

[0022] Figure 3 A fluid simulation result diagram of a land-air amphibious platform provided in an embodiment of this application.

[0023] Figure 4 This is a schematic diagram of an autonomous landing simulation process provided in one embodiment of this application.

[0024] Figure 5 A simulation diagram of an autonomous landing planning method provided in an embodiment of this application.

[0025] Figure 6 This is a diagram illustrating an autonomous landing algorithm architecture for an amphibious platform, provided as an embodiment of this application.

[0026] Figure 7 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0029] The autonomous landing planning method for amphibious platforms in unstructured environments provided in this application can be applied to, for example... Figure 1In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be set up independently, integrated into server 104, or placed in the cloud or on other servers. Terminal 102 can send the initial state of the amphibious platform, target landing point information, and environmental obstacle information to server 104. After receiving the initial state of the amphibious platform, target landing point information, and environmental obstacle information, server 104 establishes a dynamic model of the autonomous landing phase of the amphibious platform based on the initial state of the amphibious platform, target landing point information, and environmental obstacle information. The dynamic model includes: a rigid body dynamic model, a near-ground model, and a differential flat dynamic model. The near-ground model includes: a track passive buffer mechanism, a terrain surface sub-model, and a ground effect thrust gain sub-model. The system acquires the initial state of the amphibious platform, target landing point information, and environmental obstacle information; based on the... The initial state, target landing point information, and environmental obstacle information are sampled to generate an initial reference path from the starting point to the ending point. A safe flight corridor is constructed based on the initial reference path. Based on the dynamic model of the autonomous landing phase of the amphibious platform, and using the safe flight corridor as a spatial obstacle avoidance constraint, a multi-constraint spatiotemporal joint trajectory optimization problem with ground effect penalty terms is constructed. The trajectory optimization problem is solved to obtain the optimal landing trajectory that satisfies the dynamic constraints and terminal constraints. The dynamic constraints include velocity constraints, acceleration constraints, angular velocity constraints, and thrust constraints. The terminal constraints include terminal velocity constraints and terminal attitude constraints. The optimal landing trajectory is output to control the amphibious platform to perform autonomous landing. The server 104 can feed back the obtained optimal landing trajectory to the terminal 102. Furthermore, in some embodiments, an autonomous landing planning method for an amphibious platform in an unstructured environment can also be implemented separately by the server 104 or the terminal 102. For example, the terminal 102 can directly perform corresponding operations on the initial state, target landing point information, and environmental obstacle information of the amphibious platform, or the server 104 can obtain the initial state, target landing point information, and environmental obstacle information of the amphibious platform to be processed from the data storage system, and perform corresponding operations on the initial state, target landing point information, and environmental obstacle information of the amphibious platform to be processed.

[0030] The terminal 102 can be, but is not limited to, various desktop computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, and smart in-vehicle devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices. The server 104 can be implemented using a standalone server or a server cluster composed of multiple servers, or it can be a cloud server or an onboard computer.

[0031] In one exemplary embodiment, such as Figure 2 As shown, an autonomous landing planning method for amphibious platforms in unstructured environments is provided. This method is executed by computer equipment, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps 201 to 207.

[0032] Step 201: Establish a dynamic model for the autonomous landing phase of the amphibious platform; the dynamic model includes: a rigid body dynamic model, a near-ground model, and a differential flat dynamic model; the near-ground model includes: a track passive buffer mechanism, a terrain surface sub-model, and a ground effect thrust gain sub-model.

[0033] Step 202: Obtain the initial status of the amphibious platform, target landing point information, and environmental obstacle information.

[0034] Step 203: Based on the initial state, the target landing point information, and the environmental obstacle information, perform state sampling to generate an initial reference path from the starting point to the ending point.

[0035] Step 204: Construct a safe flight corridor based on the initial reference path.

[0036] Step 205: Based on the dynamic model of the autonomous landing phase of the amphibious platform, and taking the safe flight corridor as the spatial obstacle avoidance constraint, construct a multi-constraint spatiotemporal joint trajectory optimization problem that includes a ground effect penalty term.

[0037] Step 206: Solve the trajectory optimization problem to obtain the optimal landing trajectory that satisfies the dynamic constraints and terminal constraints; the dynamic constraints include velocity constraints, acceleration constraints, angular velocity constraints and thrust constraints; the terminal constraints include terminal velocity constraints and terminal attitude constraints.

[0038] Step 207: Output the optimal landing trajectory to control the amphibious platform to perform autonomous landing.

[0039] By implementing steps 201 to 207 above, this application enables the planner to possess active disturbance rejection capabilities by placing the ground effect model at the planning layer and as part of the optimization objective, fundamentally solving the near-ground landing instability problem. Simultaneously, the safety corridor ensures obstacle avoidance throughout the entire trajectory, and spatiotemporal joint optimization ensures the dynamic feasibility of the trajectory. This method achieves systematic and coordinated protection of air safety and near-ground stability in unstructured environments.

[0040] In another exemplary embodiment of this application, in order to accurately establish the dynamic model of the autonomous landing phase of the amphibious platform, the above step 201 is replaced by the following steps 301 to 303.

[0041] Step 301: Establish a rigid body dynamics model.

[0042] Step 302: Establish a near-ground model.

[0043] Step 303: Establish a differential flat dynamic model.

[0044] As an optional implementation, step 301 above specifically includes the following.

[0045] The amphibious platform's dual-modal design combines the mobility of both unmanned aerial vehicles (UAVs) and unmanned vehicles, allowing for seamless switching between air and ground modes. An inertial coordinate system is defined. and body coordinate system The amphibious platform model that lands in the air can be equivalent to a quadcopter drone model, and simplified into a rigid body dynamics model. The specific expression of the rigid body dynamics model is as follows.

[0046] .

[0047] Where x is the state variable, It is the rate of change of the state variable, and u is the control input. Differentiate for position, Take the second derivative for position. Differentiate for velocity, m It's the quality of the fuselage. p, v, a These are position, velocity, and acceleration in an inertial frame of reference. a T This indicates the thrust acceleration generated by the actuator. ω It represents the angular velocity of the organism, and [·]× denotes the antisymmetric matrix form of the cross product of vectors. g =[0, 0, 9.8] T It is the gravity vector in an inertial frame of reference. Let e3 represent the rotation matrix from the inertial frame to the machine frame, where e3 = (0, 0, 1). T It is a unit vector, and f represents the thrust along the body axis.

[0048] As an optional implementation, step 302 above specifically includes the following:

[0049] The near-ground model includes: a tracked passive buffer mechanism, a terrain surface sub-model, and a ground effect thrust gain sub-model. Through these three safeguards, the amphibious platform can make a smoother transition from the air to the land. Specifically, it includes the following:

[0050] This application introduces a passive tracked cushioning mechanism. The ducted tracked amphibious platform addressed in this application not only uses its track structure for ground movement but also acts as a highly efficient passive cushioning system during landing. It effectively absorbs impact energy passively, reducing the probability of platform damage during landing, increasing the ground contact area to disperse landing impact, assisting in adjusting the post-landing attitude, and seamlessly switching to ground-based motion modes, unlike other UAV landing gears which have limited functionality and are constrained by weight and complexity. The core value of the tracked walking mechanism during the landing phase is to improve landing tolerance and ensure ground mobility, allowing for more relaxed terminal attitude constraints and impact dynamics boundaries.

[0051] Construct a terrain surface sub-model; most current studies rely on the assumption of flat or sloping surfaces, so we first consider the landing surface of the amphibious platform in a complex terrain environment as an inclined plane, and perform planar fitting on the point cloud of the target landing ground.

[0052] Secondly, the lower side of the symmetrical amphibious platform is modeled as a disk. The platform is only constrained to the landing side when it falls into the disk's projection. The disk expression is as follows.

[0053] .

[0054] The position of any point at the bottom of the platform at any given time is represented as follows.

[0055] .

[0056] Where, p i (t) is the position, h c The distance from the platform's center of mass to the platform's bottom surface is represented by the z-axis in the fuselage coordinate system. b 'l' represents the platform width. This indicates the radius of the disk. Let r represent the mapping matrix from three-dimensional space to the bottom surface of a two-dimensional disk, and let r represent the scaling factor from the bottom center to the bottom edge, satisfying the following condition: and c(t) represents the center coordinates of the disk.

[0057] After modeling the bottom of the platform as a disk, accurately calculate the left and right track contact surfaces at the bottom of the computer body; the expressions for the left and right track contact surfaces are as follows.

[0058] .

[0059] in, p l，r ( s,t () represents the points on the left and right track contact surfaces. t Inertial frame coordinates at time 10:00 p 0 ( t The platform's core focus is on t Inertial frame coordinates at time 10:00 Let be the rotation matrix from the body coordinate system to the inertial coordinate system. s This is a normalized parameter along the track length direction. s ∈[-1, 1], l t Contact length of a single track, d t The distance from the track center to the platform's plane of symmetry. h c It is the vertical distance from the platform's center of gravity to the bottom surface of the track.

[0060] Finally, in order to evaluate the terrain slope, the terrain point cloud in the area below the left and right tracks needs to be fitted to a plane, and the corresponding covariance matrix is ​​as follows.

[0061] .

[0062] in, p i These are the inertial coordinates of a terrain point within that area. p m The coordinates are the average coordinates of the point cloud in the landing area, and the normal vector of the fitted plane in the track area is the eigenvector corresponding to the smallest eigenvalue of the covariance matrix.

[0063] The roll angle induced by the terrain of the machine body is calculated based on the normal vector of the fitted plane of the track area. ϕ i Terrain-induced pitch angle θ i Establish post-landing stability constraints, the stability constraints are as follows: ,in, ϕ max and θ max This represents the maximum permissible attitude angle.

[0064] By constructing a ground effect inference gain sub-model, the ducted amphibious platform can be regarded as a six-degree-of-freedom single rigid body system during flight. Its dynamic characteristics exhibit significant inter-channel coupling and strong nonlinearity. This nonlinearity is mainly reflected in the complex aerodynamic characteristics brought about by the modular duct design.

[0065] A key challenge during autonomous landing is managing the aerodynamic ground effect generated during near-ground flight. When the platform hovers or descends at low altitudes (especially within a rotor diameter), the rotor downwash impacts the ground, creating a reflected flow field that leads to vortex generation, shedding, and complex interactions between the vortex and the ground and fuselage. This phenomenon is known as the ground effect, which macroscopically manifests as follows: at the same rotational speed, the actual lift generated by the rotor is greater than its theoretical value in a free flow field. Specifically, when operating close to the ground, the ground effect causes disturbances in flight stability. For rotorcraft, accidents may occur when the platform leaves or enters the ground effect zone. Furthermore, the unstable dynamics affect the aircraft's trajectory at the final moments of landing, making it difficult for the aircraft to land safely and accurately at the target location on the surface.

[0066] Given that the ground effect involves complex unsteady fluid-structure interaction, it is currently impossible to establish a universally applicable and accurate analytical model for it using first-principles calculations. However, experimental observations show a strong correlation between thrust gain and platform height above ground. Therefore, this application constructs a semi-empirical ground effect thrust gain sub-model suitable for this platform configuration by combining computational fluid dynamics (CFD) numerical simulation with experimental data fitting. This sub-model is used to quantitatively predict and actively compensate for this effect at the planning level.

[0067] When using computational fluid dynamics (CFD) simulation in this application, without introducing complex flow field parameters, multiplicative correction is used to obtain the thrust affected by the ground effect when there is relative attitude angle disturbance. The specific expression of the thrust is as follows.

[0068] .

[0069] in, T ∞ It is ground effect thrust. H min This is the minimum normal distance from the rotor plane to the ground. θ It is the ground slope angle. R The rotor radius is... k and m This was determined by fitting simulation data.

[0070] The expression for ground effect thrust gain is defined as follows.

[0071] .

[0072] As an optional implementation, step 303 specifically includes the following:

[0073] The quadcopter UAV, with its equivalent dynamic characteristics when an amphibious platform is in flight, possesses the characteristic of differential flatness. Specifically, this characteristic allows trajectory planning to be transformed into optimization of a flat output within a linear space, without needing to solve for all the states and control variables of the original dynamics. The differential flatness characteristic is a prerequisite for the planning algorithm in this application, and the complete dynamic equations considering ground effects are given.

[0074] Actual trajectory planning is usually a nonlinear and non-convex nondeterministic polynomial problem. By applying the differential flatness property of quadrotor dynamics to formulate an optimization problem, the complexity of the problem can be significantly reduced by utilizing the inherent differential flatness properties such as variable decoupling and dimensionality reduction, and differential constraint removal.

[0075] For the amphibious platform affected by ground effects addressed in this application, the dynamic model maintains its differential flatness characteristics after introducing ground effects through modeling and equivalence methods, thus providing trajectory information and generating accurate control commands; that is, the basic planning characteristics of the system remain unchanged after considering ground effects. The full-state and full-control inputs are as follows.

[0076] .

[0077] Select flat output as .

[0078] Establish a differential flat dynamic model for .

[0079] in, , , These are the velocity / acceleration and jerk vectors in the inertial frame, respectively. and They are respectively states x and control u Differential flat mapping.

[0080] Within the Newton-Euler framework, considering rigid body dynamics and ground effects, the dynamic model for the autonomous landing phase of the amphibious platform is finally established as follows.

[0081] .

[0082] in, m J is the mass of the machine body, J is the moment of inertia of the machine body, a is the acceleration of the machine body, and ω is the angular velocity of the machine body. g It is gravitational acceleration. T and τ B It is the thrust and torque generated by all the rotors. f G It is the additional thrust generated by the ground effect. f DIt is resistance to forward movement. τ G It is the ground effect restoring torque. τ ext It is other unknown torques. z W =[0,0,1] is the direction of the Z-axis in the inertial coordinate system. z B It refers to the orientation of the Z-axis of the UAV's body coordinate system in the inertial coordinate system.

[0083] It should be noted that ground effect has a negative feedback effect on flight dynamics. For example, when the platform descends, the increased lift will resist the descent motion, a characteristic that is detrimental to achieving a stable and precise landing. In this application, we will address this complex aerodynamic effect through... f G and τ G The model is constructed as a predictable disturbance field. The planning algorithm utilizes this information to proactively predict disturbances and adjust the trajectory, thereby generating a landing trajectory more robust to ground effects and ultimately improving the stability and safety of the landing process.

[0084] Step 202: Obtain the initial status of the amphibious platform, target landing point information, and environmental obstacle information.

[0085] Step 203: Based on the initial state, the target landing point information, and the environmental obstacle information, perform state sampling to generate an initial reference path from the starting point to the ending point.

[0086] Specifically, in step 203, in order to quickly find a feasible path in a space full of obstacles, this application employs sparse informed response time (SRT). The algorithm employs a two-stage adaptive search strategy and online isometry to accelerate convergence and balance planning efficiency with path quality.

[0087] In the first stage, an expanded search radius and step size, along with target bias sampling, are used to quickly explore and obtain an initial feasible solution.

[0088] In the second stage, the standard search parameters are restored, and heuristic sampling is performed within the elliptical region with the current optimal path length as the major axis to refine the initial feasible solution, ultimately obtaining a set of state sampling points from the starting point to the ending point. and initial reference path ,in P k ( K =0,1,..., K ) represents the state point. K This is the maximum index of the state point.

[0089] Step 204: Construct a safe flight corridor based on the initial reference path.

[0090] Specifically, in step 204, the initial reference path is a line. To ensure sufficient obstacle avoidance space for the subsequently optimized trajectory, it needs to be widened into a safe space. Obstacle information is typically encoded as constraint terms in the optimization process, i.e., a safe flight corridor. The corridor is decomposed into convex components in the free space within a cluttered environment, and collision-free operation is verified using a semi-positive definite region expansion iteration method. This application constructs a safe flight corridor based on the initial reference path, determines a convex feasible region in the non-convex configuration space, and provides collision constraints for the optimization process. Represents aerial obstacles. To address the obstacle set, a safe corridor is employed to ensure the flight safety of the amphibious platform in complex obstacle environments, and collision avoidance constraints are established. .in, p The location of the drone. Represents the safe space after removing obstacles. For safe flight corridors.

[0091] The specific steps for constructing a safe flight corridor are as follows.

[0092] Using the line segments between adjacent points in the initial reference path as axes, generate several initial ellipsoids. .

[0093] Each initial ellipsoid is iteratively expanded until it contacts the obstacle space, and collisions are continuously detected to find the obstacle space. The contact points are used to generate several tangential planes at the contact points.

[0094] Construct a convex polyhedron using several tangent planes as a safe unit for the reference path.

[0095] Connect the safety units corresponding to all line segments on the reference path in sequence to form the safe flight corridor. .

[0096] in, M c Let be the number of convex polyhedra contained in the flight corridor. Indicates the first i An analytical expression for a convex polyhedron, A i b i It uses linear inequalities to represent the intersection of one side of the space of each face of a convex polyhedron. .

[0097] The advantage of the safe flight corridor generation method in this application is that the initial trajectory value is an equidistant polyline segment, and the length of the polyline can be customized by modifying the parameters to prevent the generated polyhedra from overlapping too much due to the line segments being too short. As a result, the number of each ellipsoid and convex polyhedron is smaller and the volume is larger. Furthermore, it can form a larger intersection space, which makes the optimization process more flexible.

[0098] In another exemplary embodiment of this application, in order to accurately construct a multi-constraint spatiotemporal joint trajectory optimization problem with ground effect penalty terms based on the dynamic model of the autonomous landing phase of the amphibious platform and with the safe flight corridor as the spatial obstacle avoidance constraint, step 205 is replaced by steps 401 to 403.

[0099] Step 401 involves parameterizing the trajectory using a piecewise polynomial. To achieve efficient numerical optimization, real-time planners typically utilize differential flatness properties and efficient trajectory parameterization to optimize the control sequence, thereby improving computational efficiency. The parameterization form of the Minimum Control Effort Polynomial (MINCO) is adopted. MINCO, as a novel polynomial parameterization form, allows for real-time synchronous spatiotemporal optimization and effectively reduces problem complexity by transforming some hard constraints into soft penalty terms. The expression for the Minimum Control Effort Polynomial is as follows.

[0100] .

[0101] in, p ( t ) is a use M Segment Continuous N polynomial expression of order m dimensional trajectory, where, N =2 s -1, no. i The segment trajectory is In the formula, c is the first... i The coefficient matrix of the segment polynomial t It is a natural base. T i It is the first i The duration of each segment. It's a transformation with linear complexity. , q i It's an intermediate waypoint. You can... q and T Evaluate the entire trajectory. This allows for any second-order continuous objective function... The gradient obtained therefrom is applicable to the equations derived from ... q and TThe expressed MINCO trajectory, and the corresponding objective function is calculated as follows: .

[0102] This application selects s=3, then N=5 is a fifth-order polynomial. It exhibits better numerical stability and physical realizability, thus satisfying the dynamics requirements of UAV systems. In contrast, the commonly used seventh-order polynomial Runge phenomenon leads to unstable trajectory generation, and the effect of parameter adjustment is limited.

[0103] Step 402: Construct the objective function. The objective function for autonomous landing planning on unstructured ground is set as a trade-off between minimum jerk, minimum time, and ground effect thrust gain constraints, which balances smoothness and aggression. The relevant expressions are as follows.

[0104] (a).

[0105] (b).

[0106] Wherein, formula (a) is the expression for the objective function. ρ >0 is the weighting parameter for the total flight time. In formula (b) T This represents the total duration of the trajectory.

[0107] Due to the nonlinear aerodynamic coupling caused by ground effects during the descent phase, this application introduces an additional soft constraint term in the trajectory optimization stage. This constraint term implicitly suppresses aggressive near-ground maneuvering behavior by penalizing the normalized thrust deviation caused by approaching the ground, while ensuring the feasibility of the sampling plan. The thrust boundary ensures feasibility, while the cost associated with ground effects only adjusts the optimal trajectory within the feasible range. Therefore, a ground effect thrust gain constraint term is introduced as a soft penalty to actively suppress aggressive maneuvers caused by approaching the ground during optimization. The expression for the ground effect penalty term is as follows.

[0108] .

[0109] in, These are the weighting coefficients; p g ( t ) represents the local ground reference point, and n represents the local terrain normal vector.

[0110] Step 403: The dynamic constraints, terminal constraints, and obstacle avoidance constraints are transformed into additional penalty terms in the objective function using the penalty function method, and the trajectory optimization problem is transformed into an unconstrained nonlinear optimization problem.

[0111] The optimization problem requires satisfying multiple hard constraints. This application employs a continuously differentiable penalty function method to transform these constraints into soft constraints, which are then integrated into the objective function, thus transforming the original constrained optimization problem into an unconstrained one. These constraints include terminal constraints, dynamic constraints, and obstacle avoidance constraints. Since these constraints are continuous and twice differentiable, they can be efficiently integrated into a polynomial trajectory optimization framework. Furthermore, the attitude deviation at the terminal landing stage is very small; compared to the highly correlated ground effects in a near-horizontal landing configuration, the attitude-induced deviation term can be ignored.

[0112] (1) Terminal constraints include terminal attitude constraints and terminal velocity constraints. Trajectory planning connects the current state and the target state to guide the UAV to complete the flight and landing mission, satisfying the terminal constraints. The expression is shown in formula (c).

[0113] (c).

[0114] Among them, t s and t g x represents the initial and terminal times, respectively. s and x g These represent the initial and final states, respectively. Boundary conditions at the starting point. The final pose is given by the platform's real-time motion status and constrained by the landing point and landing plane.

[0115] (2) Dynamic constraints include velocity and acceleration constraints (d), angular velocity constraints (e) and thrust constraints (f).

[0116] (d).

[0117] (e).

[0118] (f).

[0119] Among them, the velocity and acceleration of the dynamic constraints are used for dynamic feasibility. v max and a max It is the boundary between velocity and acceleration. It is the maximum angular velocity of the fuselage. It is the thrust limit.

[0120] (3) Obstacle avoidance constraint: The position trajectory is restricted to the safe flight corridor to achieve obstacle avoidance safety. The corridor is represented by a convex polyhedron, and a smooth corridor intrusion penalty function is constructed. When the trajectory point deviates from the corridor, a penalty is applied. The specific expression is as follows.

[0121] (g).

[0122] No.i The control energy cost and its gradient generated by the segment trajectory can be expressed by polynomial coefficients. c i and duration T i Analytical expression, gradient and The expression is as follows.

[0123] .

[0124] The resulting formulaic description, combining all constraint terms, includes equality and inequality constraints. Equality constraints limit the solution space, and in some cases, it may be impossible to find a trajectory that satisfies all constraints, leading to optimization failure. To avoid this, slack variables are introduced to relax the strictness of equality constraints. For all inequality constraints, penalty functions for constraint violations are designed.

[0125] The following methods can be used to eliminate the constraints mentioned above.

[0126] Regarding the time mapping of formula (b), the trajectory optimization... i Time length variable of segment trajectory T i (Physical time) usually needs to meet the following requirements T i >0, but direct optimization T i This could lead to numerical problems. Therefore, a logarithmic mapping is used to transform the unconstrained variables through time scaling. t Mapping to positive time optimization variables T i This simplifies the optimization process.

[0127] .

[0128] In the brake constraints (d), (e), and (f), the dynamic constraints ensure that the velocity and acceleration throughout the entire segment cannot exceed physical limits. Thrust constraints prevent motor saturation or stall, avoiding uncontrollable UAVs due to thrust overload or insufficiency. This is achieved by designing a second-order continuously differentiable relaxation function. Construct the penalty function for the brake constraint and add it to the objective function.

[0129] .

[0130] in, v ( t ) is the velocity variable. v max That is the maximum speed. a ( t ) is the acceleration variable. amax It is the maximum acceleration. ω ( t ) is the angular velocity variable. ω max It is the maximum angular velocity. τ ( t ) is the thrust variable. τ max It is the maximum thrust. τ min It is the minimum thrust.

[0131] Terminal constraints include terminal velocity constraints and terminal attitude constraints. The core of terminal velocity constraints is to resolve the conflict between ideal conditions (zero-velocity landing) and complex terrain. For a fifth-order polynomial trajectory in a flat output space, the boundary conditions for the endpoint should explicitly specify the location. to the second derivative All information is available. The endpoint location is specified by the user, and the endpoint speed under ideal conditions is also included. Since a hard constraint of strictly zero may lead to the absence of a solution space or an unsolvable optimization problem, especially in scenarios with uneven terrain, this application designs the endpoint velocity under non-ideal conditions as follows.

[0132] .

[0133] in, v n Given a normal velocity coefficient, this generates a small normal velocity component that allows the platform to make a smoother contact with the ground and reduces the risk of slippage on inclined landing surfaces. v t For the optimizable tangential velocity coefficient, the tangential velocity direction x t The landing direction x indicated by the user on the two-dimensional plane yaw Analytical calculation of the landing plane normal vector n. It is the identity matrix. It is the normal projection matrix.

[0134] To ensure the optimization problem has sufficient solution space, the tangential landing velocity should be slightly greater than 0, without being too large and causing airframe instability. Therefore, an adjustment term is introduced. g To minimize the tangential landing velocity: .

[0135] Terminal attitude constraints can be expressed using net thrust. Combining this with net thrust constraints, and then using a new optimization variable τ... e This is converted into terminal acceleration in the boundary conditions, causing the thrust to [ τ min , τ maxWithin the range.

[0136] .

[0137] in, , Therefore, by using the new optimal variable replace Terminal attitude and thrust variables were eliminated.

[0138] In obstacle avoidance constraint (g), a safe corridor is used, and a penalty function ensures that the trajectory remains within the corridor. The problem with hard constraint methods is that all safe areas are considered equivalent, which may lead to some parts of the trajectory being too close to obstacles. If the control unit does not perfectly follow the trajectory, a collision will still occur. Furthermore, hard constraint methods cannot optimize the gap between the trajectory and obstacles. By constructing a flight corridor, the position constraint is transformed into a convex space, making the optimization problem still convex and solvable by efficient methods.

[0139] Safety constraints strictly guarantee that the trajectory lies within the corridor, and its smoothness allows for the use of efficient second-order optimization algorithms. i The segment trajectory is restricted to the corresponding convex polyhedron. Internally, to ensure the platform's landing trajectory avoids aerial obstacles, a quadratic differentiable smooth C-axis is introduced. 2 The penalty function is then relaxed using a relaxation function. Construct penalty function as follows.

[0140] .

[0141] In summary, all constraints corresponding to the penalty functions should satisfy the entire trajectory. This application transforms the original multi-constraint nonlinear descent trajectory optimization problem into an unconstrained nonlinear optimization problem, the cost function of which is given by the following equation.

[0142] .

[0143] After obtaining all gradients of the optimization variables using the above method, analytical gradient values ​​are used to accelerate the optimization iteration, yielding the solution to the unconstrained nonlinear optimization problem, i.e., the numerical approximate solution to the optimal control problem, thus achieving rapid solution of trajectory planning. Unlike traditional trajectory planning methods, it eliminates the need to solve for the discretized time series of the full state and control variables, significantly reducing the dimensionality of the problem.

[0144] This application solves the ground effect thrust gain parameters of amphibious platforms at different altitudes and terrains, and evaluates the algorithm performance by comparing the planning method performance in a simulated environment.

[0145] (1) Numerical simulation analysis of terrain effects Before introducing planning methods, establishing a ground effect thrust gain model is crucial because it significantly impacts stability during near-ground landing. Typical empirical ground effect models for quadrotors are unsuitable for amphibious platforms for several reasons: first, they are primarily applied to open rotors, not ducted rotor structures; second, the landing support structures are inconsistent, including the track effect of amphibious platforms; and third, they only consider level, flat ground, neglecting the often-present sloped or uneven terrain.

[0146] Computational fluid dynamics (CFD) methods were employed to estimate the aerodynamic disturbances between an amphibious platform and the ground through numerical calculations. This enables the planner to have accurate trajectory planning capabilities and effectively resist aerodynamic disturbances. Specifically, this application conducted CFD analyses at different altitudes and ground tilt angles to comprehensively understand near-ground airflow disturbances during flight and landing in unstructured environments. Since performing CFD analyses on all possible scenarios is impractical, a series of settings representing typical real-world flight conditions were used. In each case, the four propellers of the amphibious platform were simplified to four actuator surface models, generating pressure jumps. A rectangular computational domain was set up using an unstructured polyhedral mesh, with local sizing and boundary layers added to capture near-wall flow and detailed meshes near the rotors. The bottom of the computational domain was the ground, while the other surfaces maintained atmospheric pressure. These cases were solved steadily using a pressure-based k-Omega SST model with enabled curvature correction and a coupled algorithm.

[0147] Numerical simulation methods using realistic models of dynamic units all suffer from low solution efficiency. Therefore, this application employs an equivalent disk, i.e., actuator surface model, based on momentum-blade element theory, to replace the steady momentum source method of the real blade for efficient solution. The momentum source method equates the rotational effect of the blade to a thin-body disk, calculates the force distribution on the disk based on blade element theory, and then transforms it into momentum source terms to be solved in the Navier-Stokes (NS) equations. The fluid simulation results for the amphibious platform are shown in the figure below. Figure 3 As shown, Figure 3 (a) Setting up fluid simulation, Figure 3 (b) is a pressure cloud map. Figure 3 (c) is a velocity contour map.

[0148] Figure 3 The diagram shows the propeller downwash airflow impacting the ground and spreading outwards. Some of the airflow in the central area does not have time to spread and even produces a backlash. The velocity contour plot shows that, due to the propeller, the airflow velocity increases significantly after passing the propeller disk, and due to the propeller's suction effect, the airflow velocity also increases in front of the propeller disk.

[0149] The first amphibious platform mThe aerodynamic force generated by the actuator disk is expressed as follows in the body coordinate system.

[0150] .

[0151] in, No. m The force vector of each disk in the body coordinate system The rotation transformation matrix from the inertial coordinate system to the body coordinate system, derived from the roll angle. ϕ Pitch angle θ Yaw angle ψ Definition: The pressure jump value set for the actuated disk model is a constant in a single steady-state solution. N f,m It is the first m The total number of discrete face cells of the actuator disk. It is the first m The first plate i The area vector of a face element is defined in an inertial coordinate system, with the direction being the outward normal of the face element and the size being the area.

[0152] Total thrust vector of the entire unmanned aerial vehicle system The force vectors of all actuated discs and the force vector of the machine body are given, and the expression for the total thrust vector is as follows.

[0153] .

[0154] The total thrust is obtained by vector calculation to obtain the inertial force, and then projected onto the body coordinate system through a rotation matrix. The calculated... The vertical component is the total lift required to balance gravity. Quantifying the ground effect serves as the modeling basis for the aerodynamic disturbance assessment model. This is determined through experimental fitting. k =0.25, m =1.5.

[0155] (2) Landing planning simulation analysis The simulation scenario involves uneven ground with randomly distributed columnar obstacles. The simulation requires the amphibious platform to autonomously plan a trajectory after being given a landing target point, ensuring a safe and stable landing. Specifically, after the algorithm starts, it reads map data and the amphibious platform model information. The aerial state of the amphibious platform is equivalent to a quadcopter model, and the map consists of uneven ground and randomly distributed columnar obstacles. The platform then flies to its initial position. p 0 = (-4.0, -4.0, 2.0) and hover, waiting to receive target location instructions. Issue target location command. p t =(x t y t (0.00), automatically changed according to terrain featuresp s =(x t y t , z s ).

[0156] The perceived state estimation is provided by the odometry sensor position data of the platform model. The platform is controlled by only one position controller, which also includes attitude control. The result of position control is the desired acceleration, and the result of attitude control is the desired angular acceleration. The angular acceleration is then passed through a mixer to finally obtain the desired motor speed, forming the entire closed-loop process. A schematic diagram of the autonomous landing simulation process is shown below. Figure 4 As shown.

[0157] Flight performance parameters are v max =2m / s, a max =4m / s 2 , ω max =3rad / s, τ max =17 / s 2 , τ min =5 / s 2 .

[0158] Figure 5 The simulation diagrams for the autonomous landing planning method are as follows: (a) the starting and ending points of the amphibious platform in the obstacle map; (b) the initial and final trajectories of the comparative method; and (c) the initial and final trajectories of the proposed method. The black broken line represents the initial trajectory, and the red line represents the final trajectory. The amphibious platform maintains stable flight throughout, without excessive high-order state abrupt changes, demonstrating good dynamic performance. During descent from the initial altitude, upon entering the ground effect zone, sensors are used to estimate the altitude and its rate of change, and the normalized thrust deviation caused by the ground effect is pre-adjusted in the planner. Compared to the comparative method, the proposed method can rapidly converge, shorten the planning time, and incorporate ground effect thrust constraints to make the terminal trajectory more stable, ultimately enabling the aircraft to land stably on the ground.

[0159] Figure 6This application provides an embodiment of an autonomous landing algorithm architecture for an amphibious platform. It also provides an application scenario where the aforementioned autonomous landing planning method for amphibious platforms in unstructured environments is applied. Sensors detect the environment in real time and return information, acquiring the platform's initial and target states. The planning algorithm intelligently samples in the state space and constructs a safe flight corridor, iteratively generating a multi-constraint spatiotemporally optimized trajectory. Finally, the trajectory server provides control input to the platform, thereby completing the autonomous landing. Specifically, the autonomous landing planning method provided in this embodiment can be applied to tasks such as field material delivery and post-disaster reconnaissance. An amphibious platform equipped with this method flies from the air to the mission area, uses onboard sensors to perceive the uneven landing zone below, which may contain obstacles such as loose rocks and fallen trees, and then automatically plans and executes a safe and stable landing trajectory, accurately landing in the designated area. It then switches to ground mode to perform subsequent tasks.

[0160] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 7 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs in the non-volatile storage media to run. The database stores the initial state of the amphibious platform, target landing point information, and environmental obstacle information. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements an autonomous landing planning method for amphibious platforms in unstructured environments.

[0161] Those skilled in the art will understand that Figure 7 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0162] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0163] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0164] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Moreover, the collection, use and processing of the relevant data are carried out in compliance with the relevant data protection laws and policies of the country where the location is located, and with the authorization granted by the owner of the corresponding device.

[0165] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0166] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0167] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0168] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. An autonomous landing planning method for an amphibious platform in an unstructured environment, characterized in that, The autonomous landing planning method for amphibious platforms in unstructured environments includes: A dynamic model for the autonomous landing phase of an amphibious platform is established. The dynamic model includes a rigid body dynamic model, a near-ground model, and a differential flat dynamic model. The near-ground model includes a track passive buffer mechanism, a terrain surface sub-model, and a ground effect thrust gain sub-model. Acquire the initial status of the amphibious platform, target landing point information, and environmental obstacle information; Based on the initial state, the target landing point information, and the environmental obstacle information, state sampling is performed to generate an initial reference path from the starting point to the ending point. Construct a safe flight corridor based on the initial reference path; Based on the dynamic model of the autonomous landing phase of the amphibious platform, and taking the safe flight corridor as the spatial obstacle avoidance constraint, a multi-constraint spatiotemporal joint trajectory optimization problem including a ground effect penalty term is constructed. Solving the trajectory optimization problem yields the optimal landing trajectory that satisfies both dynamic and terminal constraints. The dynamic constraints include velocity, acceleration, angular velocity, and thrust constraints. The terminal constraints include terminal velocity and terminal attitude constraints. The optimal landing trajectory is output to control the amphibious platform to perform an autonomous landing.

2. The method of claim 1, wherein, The rigid body dynamics model can be equivalent to a quadcopter UAV model, the amphibious platform is a ducted tracked amphibious platform, and the ground effect thrust gain model is established for the ducted rotor structure; the expression of the rigid body dynamics model is: ； Where x is the state variable, It is the rate of change of the state variable, and u is the control input. Differentiate for position, Take the second derivative for position. Differentiate for velocity, m It's the quality of the fuselage. p, v, a These are position, velocity, and acceleration in an inertial frame of reference. a T This indicates the thrust acceleration generated by the actuator. ω It represents the angular velocity of the organism, and [·]× denotes the antisymmetric matrix form of the cross product of vectors. g =[0, 0, 9.8] T It is the gravity vector in an inertial frame of reference. Let e3 represent the rotation matrix from the inertial frame to the machine frame, where e3 = (0, 0, 1). T It is a unit vector, and f represents the thrust along the axis of the aircraft.

3. The autonomous landing planning method for amphibious platforms in unstructured environments according to claim 1, characterized in that, The track passive buffering mechanism is as follows: it effectively absorbs the impact energy and reduces the probability of damage to the landing platform by passively absorbing the impact energy, increases the ground contact area to disperse the landing impact, assists in adjusting the attitude after landing, and seamlessly switches to ground mode motion. The establishment of the terrain surface sub-model is as follows: The complex landing surface is modeled as an inclined plane, and the point cloud of the target landing surface is fitted to a plane. Based on track geometry, the left and right track contact surfaces at the bottom of the computer body are defined; the expressions for the left and right track contact surfaces are as follows: ； in, p l，r ( s,t () represents the points on the left and right track contact surfaces. t Inertial frame coordinates at time 10:00 p 0 ( t The platform's core focus is on t Inertial frame coordinates at time 10:00 Let be the rotation matrix from the body coordinate system to the inertial coordinate system. s This is a normalized parameter along the track length direction. s ∈[-1, 1], l t Contact length of a single track, d t The distance from the track center to the platform's plane of symmetry. h c The vertical distance from the platform's center of gravity to the bottom surface of the track; The point clouds below the left and right contact areas are fitted to planes again to calculate the terrain-induced attitude angles and establish landing stability constraints. The establishment of the ground effect thrust gain sub-model specifically includes: Based on computational fluid dynamics, the near-ground flow field of the amphibious platform under different ground altitudes and ground tilt angles was simulated. Based on the simulation results, the thrust affected by the ground effect is obtained by fitting, and the thrust is: ； The ground effect thrust gain is: ； in, T ∞ It is ground effect thrust. H min This is the minimum normal distance from the rotor plane to the ground. θ It is the ground slope angle. R The rotor radius is... k and m To determine this through fitting simulation data; The dynamic model for the autonomous landing phase of the amphibious platform is as follows: ； in, m J is the mass of the machine body, J is the moment of inertia of the machine body, a is the acceleration of the machine body, and ω is the angular velocity of the machine body. g It is gravitational acceleration. T and τ B It is the thrust and torque generated by all the rotors. f G It is the additional thrust generated by the ground effect. f D It is resistance to forward movement. τ G It is the ground effect restoring torque. τ ext It is other unknown torques. z W =[0,0,1] is the direction of the Z-axis in the inertial coordinate system. z B It refers to the orientation of the Z-axis of the UAV's body coordinate system in the inertial coordinate system.

4. The autonomous landing planning method for amphibious platforms in unstructured environments according to claim 1, characterized in that, Based on the initial state, the target landing point information, and the environmental obstacle information, state sampling is performed to generate an initial reference path from the starting point to the ending point, specifically including: A sparse informed RRT based on a two-stage adaptive search strategy and online equidistant mapping is adopted. algorithm; In the first stage, an expanded search radius and step size, along with target bias sampling, are used to explore and obtain an initial feasible solution. In the second stage, the standard search parameters are restored, and heuristic sampling is performed within the elliptical region with the current optimal path length as the major axis to refine and optimize the initial feasible solution, thereby obtaining an initial reference path from the starting point to the ending point.

5. The autonomous landing planning method for amphibious platforms in unstructured environments according to claim 1, characterized in that, Constructing a safe flight corridor based on the initial reference path specifically includes: Using the line segments between adjacent points in the initial reference path as axes, generate several initial ellipsoids; Each initial ellipsoid is iteratively expanded until each initial ellipsoid comes into contact with the obstacle space, resulting in several tangent planes generated at the contact points; Construct a convex polyhedron using several tangent planes as a safe unit for the reference path; The safety units corresponding to all line segments on the reference path are connected sequentially to form the safe flight corridor.

6. The autonomous landing planning method for amphibious platforms in unstructured environments according to claim 1, characterized in that, Based on the aforementioned dynamic model and near-ground model, and using the safe flight corridor as a spatial obstacle avoidance constraint, a multi-constraint spatiotemporal joint trajectory optimization problem including a ground effect penalty term is constructed, specifically including: The trajectory is parameterized using piecewise polynomials; Construct an objective function, which is a trade-off between minimum jerk, minimum time, and ground effect thrust gain constraints; The dynamic constraints, terminal constraints, and spatial obstacle avoidance constraints are transformed into additional penalty terms in the objective function using the penalty function method. Based on the additional penalty term, the trajectory optimization problem is transformed into an unconstrained nonlinear optimization problem.

7. The method according to claim 6, characterized in that, The expression for the objective function is: ； in, Let be the objective function. ρ This is a weighting coefficient for the total flight time. p （3） ( t () is a polynomial locus. T The total duration of the trajectory. J GE This is a ground effect penalty term; The ground effect penalty item J GE Represented as: ； in, These are the weighting coefficients. p g ( t ) represents the local ground reference point, and n represents the local terrain normal vector.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that the processor executes the computer program to implement the autonomous landing planning method for an amphibious platform in an unstructured environment as described in any one of claims 1-7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the autonomous landing planning method for amphibious platforms in an unstructured environment as described in any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the autonomous landing planning method for amphibious platforms in an unstructured environment as described in any one of claims 1-7.

Images