In lunar sampling missions,the slender and flexible configuration of the robotic arm results in significant nonlinear characteristics and uncertainty in the end-effector's pose.To support rapid and intelligent sampling operations,visual pose estimation technology is introduced to provide real-time and accurate decision-making support.However,the inherent depth perception limitations of monocular vision,compounded by image degradation caused by extreme lighting variations on the lunar surface,pose severe challenges to the accuracy and robustness of pose estimation.To address the aforementioned issues,this paper proposes the geometrically constrained pose estimation network(GCP-Net).By integrating a differentiable projection module,the network constructs the reprojection error and the contour algebraic error as additional loss terms,establishing a consistency constraint between image semantics and spatial pose,thereby effectively improving depth estimation accuracy.Simultaneously,to simulate image degradation caused by extreme occlusion and lighting changes in real lunar environments,this paper introduces stochastic noise injection and feature masking enhancement strategies based on GCP-Net,proposing an enhanced geometrically constrained pose estimation network(EGCP-Net).Simulation experiments demonstrate that the proposed method maintains good robustness even under conditions of partial feature loss.Validated by the publicly reported image data from Chang'e-6,the reprojection results of the pose estimated by the algorithm highly match the observed features,proving the reliability and feasibility of this method in aerospace engineering missions.
Under the deterministic assembly mode of aircraft structures,coaxiality deviations occur at the connecting holes of parts during assembly due to factors such as hole-making deviations and shape deviations.When these deviations are relatively large,they cause local assembly stress and weaken the mechanical properties of the structure.For the deterministic assembly of aircraft fuselage structures,this study combines the Jacobian-torsor method with the influence coefficient method to determine the matrix expressions of part shape deviations,final hole positioning deviations,and temporary fastener connections.It also proposes a coaxiality deviation calculation model that accounts for deviation accumulation and part deformation,while establishing an assembly experimental system for coaxiality deviation.By comparing the measured deviation data,the calculation results of the Jacobian-torsor model and those of the proposed model,it is found that the average error between the proposed model and the measured data is reduced by 8.89%.This improves the deviation calculation accuracy of the fuselage structure under the deterministic assembly mode and provides a more reliable theoretical basis for the optimization of assembly processes.
To investigate the vibration characteristics of a landing gear gear-driven nose wheel steering system under transmission clearance conditions,this paper first conducts a simplified theoretical analysis to reasonably design anti-sway damping.Subsequently,the landing gear vibration coupled rigid-flexible multi-body dynamics simulation model is validated using both theoretical analysis results and vibration test results.Then,nonlinear factors such as gear transmission clearance and torque arm axial and radial transmission clearance are simultaneously considered in the vibration dynamics analysis,achieving a coupled rigid-flexible nonlinear dynamics model for landing gear with transmission clearance.Based on this model,a multi-factor study on the differences in vibration response characteristics is conducted.The results reveal that gaps can cause the landing gear to oscillate with constant amplitude,with gear transmission clearance having a greater effect on wheel swing angle than torque arm clearance.Additionally,the gap size is approximately proportional to the wheel swing angle.Increasing vertical load helps suppress vibration,while taxi speed primarily affects the wheel swing frequency.These findings provide theoretical reference for the design of landing gear anti-sway systems.
To quantify the correlation between freeze-thaw resistance and its influencing factors in macroporous recycled concrete （MRC） and its mesoscopic constituents—recycled aggregate （RA）， new paste （NP）， and interfacial transition zone （ITZ）—saturated specimens of RA， NP， ITZ， MRC and paste-coated single aggregate were subjected to freeze-thaw tests in air. Multiple properties and parameters of each kind of specimen were measured under different numbers of freeze-thaw cycles， and grey relational analysis was employed to quantify the correlations among the test results. The relational degree rankings of influencing factors were identified as follows： For NP compressive strength， NP indentation elastic modulus > NP pore size distribution coefficient > NP average pore size； For ITZ shear strength， ITZ indentation elastic modulus > ITZ thickness； For old paste （OP） detachment rate， OP average pore size > ITZ thickness > OP pore size distribution coefficient > ITZ indentation elastic modulus > OP indentation elastic modulus； For MRC compressive strength， NP compressive strength ≈ ITZ shear strength > OP detachment rate. The relational degrees of all factors ranged from 0.503 to 0.955， indicating significant correlations. The strength of the NP and ITZ were strongly influenced by indentation elastic modulus， while the OP detachment rate was determined by weak points in the RA. When the porosity is around 25%， the influence of the material and the pore structure on the compressive strength of MRC reaches its peak and minimum， respectively.
The paper presents a configurational synthesis method for multi-mode parallel mechanisms based on planar closed-loop chains.Firstly,a planar closed-loop chain with two motion modes is introduced.This chain is incorporated into the realm of parallel mechanisms and serves as the moving platform of the parallel mechanism.By toggling its motion modes,the change in the degrees of freedom of the parallel mechanism can be achieved.Secondly,the definitions of the full configuration and sub-configurations in multi-mode parallel mechanisms are expounded.The screw theory is introduced into the configurational synthesis theory of multi-mode parallel mechanisms.Furthermore,the configurational synthesis principle of multi-mode parallel mechanisms based on the screw theory is put forward.Subsequently,based on the configurational synthesis principle of multi-mode parallel mechanisms and planar closed-loop chains,the configurational synthesis of a class of parallel mechanisms with two motion modes is conducted.Finally,the degrees of freedom of the synthesized multi-mode parallel mechanisms are analyzed by means of the screw theory.
The sampling and in-situ analysis of water ice in the lunar polar regions is currently one of the major international research focuses in the field of deep-space exploration.The use of space robots equipped with deep fluted augers to drill into and sample water-ice-bearing lunar regolith permafrost is considered a sampling method with significant engineering feasibility.However,from a scientific perspective,the ability of this method to accurately determine the depth of origin of lunar regolith samples has not yet been verified.To address the scientific need for establishing a complete profile of lunar regolith through sample analysis,this study verifies the depth-resolution performance of deep fluted auger drilling.Based on a theoretical analysis of the flow characteristics of lunar regolith in deep fluted augers,permissible ranges for drilling procedures are established.Furthermore,the depth-resolution characteristics of the deep fluted auger are investigated through experimental testing.The experimental results show that more than 85%of the samples in the drill rod's spiral flutes originate from the final 40 mm of drilling prior to rod retrieval,and this conclusion is largely unaffected by variations in drilling speed within the studied parameter range.Therefore,by employing step-by-step drilling and rod retrieval sampling within the same borehole,it is possible to establish a complete profile of the borehole.
The carrier-based aircraft must follow planned paths during taxiing operations due to the narrow and complex deck space. Therefore， taxiing paths have a significant impact on both safety and efficiency. Based on the advantages of the reinforcement learning method in solution efficiency， the proximal policy optimization （PPO） algorithms reinforcement learning method is adopted to establish carrier-based aircraft dispatch path planning method. A termination region is designed and the end-of-episode reward is improved， so that the path can satisfy the start and end point direction constraints at the same time. The obstacle detectors are introduced into the path planning agent and the random environment training are used to improve the adaptability of the agent in different environments. Finally， the rationality of the model and the effectiveness of the method are verified by simulation.
To address two major technical bottlenecks in lunar sampling missions,geometric reconstruction failure caused by in-orbit camera parameter drift and the lack of multi-scale comprehensive analysis integrating chassis passability with terminal operation constraints in sampling area evaluation,this paper proposes an integrated method for stereo digital elevation model(DEM)reconstruction and sampling area analysis.The method employs feature-guided robust epipolar rectification to adaptively compensate for in-orbit parameter drift,elevating epipolar alignment accuracy to sub-pixel level and providing a geometrically consistent DEM for terrain analysis.It further establishes a three-level progressive sampling area analysis framework,namely global coarse-grained screening,fine-grained parking evaluation,and terminal execution verification.This framework integrates chassis passability constraints,parking stability,and manipulator reachability into a unified cost evaluation framework,facilitating refined decision-making from passable regions to operable sampling points.Experimental results validate the effectiveness of the reconstruction method and the necessity of each cost component.This integrated approach connects the complete technical chain from raw image perception to sampling decision-making,providing important theoretical support and technical means for autonomous,safe,and reliable sampling in complex lunar environments.
To ensure the safe and precise collaboration between humanoid robotic arms and astronauts in the lunar environment,this paper investigates a six-degree-of-freedom(6-DOF)humanoid arm and proposes a dynamic parameter identification method based on its dynamic characteristics.First,the kinematic and dynamic models of the humanoid arm are established.For the extremely low-speed collaborative scenarios,the dynamic model is simplified to obtain the minimal set of composite gravitational parameters and its corresponding gravity regressor matrix.Second,the trajectory optimization objective is set to maximize the feasible parameter space of the gravity regression matrix,with constraints incorporating the humanoid arm's motion limits and collision avoidance.Finally,simulations and experiments are conducted in both a simulated lunar gravity environment and an earth gravity environment.The results show that the average percentage error of the calculated joint torques after identification is only 5.5%in the lunar simulation and the mean absolute error of the joint torques is significantly smaller than the average noise amplitude in the experimental validation.Results demonstrate that the dynamic model established by the proposed identification method can accurately reflect the dynamic characteristics of the humanoid arm during extremely low-speed motion.
A variable geometry turbofan engine component level model with adjustable compressor intermediate stage bleed air and low-pressure turbine guide vanes considering Reynolds number correction is established to address the significant impact of high-altitude low Reynolds number power extraction state on engine dynamic performance. A turbofan engine acceleration performance mitigation control method based on variable geometry composite adjustment is proposed for high-altitude low Reynolds number power extraction state. The simulation results show that under low Reynolds number flight conditions of H=11 km and Ma=0.8， increasing the power extraction by 50 kW increases the engine acceleration time from idle to intermediate state by 20%. However， the variable geometry composite adjustment acceleration control proposed in this paper reduces the acceleration time by 30.2% compared to conventional methods， effectively improving the engine acceleration performance under the condition of increased power extraction.
To address the lack of physical interpretability in existing data-driven vehicle dynamics models which limits their generalization performance under extreme handling conditions and makes them susceptible to fitting distortion and long-term prediction divergence,this paper proposes a conditional multi-expert Koopman(CME-Koopman)network incorporating spectral stability regularization.Architecturally,the network introduces a condition-aware gating mechanism that dynamically dispatches multiple local linear Koopman expert models based on real-time vehicle operating conditions,thereby effectively characterizing the dynamic evolution of the vehicle from the linear handling region to the nonlinear saturation region.To mitigate the inherent instability of multi-step predictions,a spectral stability regularization term based on the power iteration method is introduced into the loss function.By explicitly constraining the modulus of the Koopman operator's eigenvalues,this approach theoretically guarantees the boundedness and asymptotic stability of long-term predictions.Experiments conducted on the CarSim high-fidelity simulation platform demonstrate that,on a dataset containing 40%extreme handling conditions,the proposed method achieves a substantial reduction in long-term prediction error,namely root mean square error(RMSE)compared to the traditional deep extended dynamic mode decomposition(Deep EDMD)baseline.Furthermore,it accurately reproduces the asymptotic convergence characteristics of the vehicle at the critical point of instability.Closed-loop co-simulation with a model predictive controller(MPC)further validates the model's engineering potential in extreme environments:It not only achieves high-precision trajectory tracking under strongly coupled longitudinal and lateral variable conditions,but also provides precise prediction of dynamics for the control system during sudden drops in road adhesion coefficient(μ-Jump).This demonstrates its outstanding anti-divergence robustness and extreme sideslip prevention capabilities.
To investigate the effect of conjunction position on fatigue performance of laminated smart structures,a study of macro fiber composite(MFC)and glass fiber reinforced polymer(GFRP)laminated smart structure is presented.Experiments on two types of specimens with different conjunction positions are carried out,and the results indicate that the fatigue life of smart structure is higher when MFC is in the middle layer of structure than at the outermost layer.A finite element model(FEM)model is established and validated by the error within 7%comparing with the load-strain data during experiment.Fatigue life prediction using a linear S-N curve shows good agreement with experimental results.The effect of the MFC conjunction position on the fatigue life of the smart structure under different load is further analyzed.Under uniaxial load,the interfacial stress decreases and the fatigue life increases as the MFC conjunction position shifts from the outermost layer to the middle layer,and this improvement is dominated by a significant reduction in tensile stress.Besides,stress concentration occurs near the short edges of the rectangular MFC under uniaxial tensile load,and around the perimeter of the MFC under uniaxial torsional load.Under multiaxial load,compared to the outermost-layer conjunction,the middle-layer conjunction almost eliminates interlaminar tensile stress and reduces the effective shear stress to varying degrees,leading to enhanced fatigue life.The degree of fatigue life improvement is significantly affected by the load ratio,and the optimal improvement occurs at a load ratio of 0.273 4.
Teleoperated driving is an important approach for efficient lunar rover exploration.However,communication delays may cause a temporal mismatch between environmental perception and command execution,increasing safety risks during rover operation.To address this problem,this paper proposes an incremental lunar orthophoto map stitching and risk anticipation method for continuous teleoperated driving.Considering the near-field blind zones of lunar rovers,discontinuous environmental representation across successive fields of view,and the vulnerability of image features under weak texture and strong illumination variations,an incremental orthophoto mapping framework is developed.The framework integrates stereo reconstruction,local orthophoto generation,multi-frame point cloud registration,and incremental fusion to achieve coordinated stitching of terrain geometry and image texture,as well as continuous environmental map updating.Furthermore,terrain slope,roughness,and texture information are fused to construct a risk cost map,which is overlaid with rover trajectory and heading information to generate a motion-aware situational map.Experimental results show that the proposed method can stably generate orthophoto preview maps with good geometric consistency and spatial continuity in complex lunar surface environments,providing intuitive and continuous risk anticipation support for ground-based lunar rover teleoperation.
Based on the aerodynamic and thermodynamic conditions of the integrated afterburner， the influence of main structural parameters of the insulation screen air film cooling on the cooling performance is systematically studied through numerical simulation methods. This research results indicate that under the same inlet and outlet boundary conditions， an increase in the opening of the upstream injector increases the secondary flow rate， separating the heat shield from the main flow of gas. The increase in the aperture of the gas film increases the depth and momentum of the jet， and reduces the uniformity and continuity of the gas film coverage. The increase in the inclination angle of the gas film hole reduces the wall adhesion of the gas film. The increase in the axial spacing of the gas film holes enhances the superposition effect of the gas film， and increases the coverage range and uniformity of the gas film. The optimal structural parameters are as follows： The aperture of the hole d=1.2 mm， the tilt angle of the hole θ=30°， the axial spacing S=10 mm， and the corrugated plate height H=6 mm. This research uses a forked arrangement for the holes in the screen.
The work probes into the model design of low-level transit route （LLTR） planning in air defense operation scenarios and its related optimization algorithm. In the model design， route shortcut， radial velocity and relationship of routes are taken as objective functions， while air defense restriction， fighter performance， route range and route coordination are regarded as constraints. In the algorithm implementation， an improved LangEvin equation based evolutionary （LEE） algorithm is proposed. The algorithm is improved via Tent map-based chaotic initialization， hybrid dynamic perturbation fused with Lévy flight and Cauchy mutation and dynamic boundary elite opposition-based learning. Simulation experiments show that the proposed method can generate reasonable low-level transit route based on air defense operation requirements regardless of the number and deployment of surface‑to‑air missile positions， and both the constraint satisfaction rate and the convergence accuracy of the proposed algorithm reach 100%. Meanwhile， compared with existing swarm intelligence algorithms， the improved algorithm achieves the optimal convergence value on test functions， and the average computation time is reduced by 32.6% to 54.1%， which can provide a reference for air defense operation airspace planning.
Truck-beam landing gears usually face vertical velocity distortion of the drop test basket during traditional reduced-mass method drop tests,leading to deviations in dynamic response.Aiming to enhance test accuracy and engineering applicability,this study proposes a multibody dynamics model of the truck-beam landing gear.The drop response characteristics under both the reduced-mass method and the lift-loading method are compared and analyzed,revealing the inherent mechanism of the traditional method:The continuous post-touchdown acceleration and velocity overshoot due to the absence of lift simulation.An improved approach based on parameter feedback iteration is proposed,achieving simultaneous equivalence in velocity and energy through coordinated adjustment of the release mass and height.Simulation validation conducted on a main landing gear of a large civil aircraft demonstrates that the proposed method reduces the error in the basket's vertical velocity from 11.15%to 0.13%,confines the energy error within 2.34%,and narrows the discrepancy in key load indicators(e.g.,peak oil damping force)to 4.01%compared with the lift-loading method.The results indicate that the proposed method significantly enhances the accuracy and reliability of drop tests for truck-beam landing gears without introducing complex lift simulation systems,providing an efficient engineering solution for validating landing gear buffer performance.
The development of parallel robot technology puts forward higher requirements for its motion performance,in which the forward position solution problem is closely related to performance characterization,performance improvement and so on.It has received widespread attention since the 1980s.Due to the strong nonlinearity and multiple solutions of the forward kinematics equations,this problem has not been completely solved at present,and it is still the focus of research in the field of robotic mechanism.Firstly,from the perspective of method principles,this paper expounds the research status and characteristics of forward position solution at home and abroad,analyzes the core ideas of the solution algorithms,and excavates the fundamental problems faced in the research.Secondly,from the perspective of method characteristics,the advantages,disadvantages and applicability of the forward position solution methods are compared and analyzed.Finally,on the basis of summarizing the existing research methods of forward position solution of parallel robots,the future development direction is prospected,which provides a reference for the research in the field of robotic mechanism.
Current multimodal sentiment analysis(MSA)models predominantly employ cross-modal attention mechanisms to process feature information from different modalities.However,these approaches often overlook the inherent similarities and dissimilarities among modal features,which can easily lead to the generation of redundancy from modal similarities and an increase in noise during cross-modal interaction,thereby degrading model performance.To address these issues,this paper proposes a novel multimodal sentiment analysis model based on feature decoupling and cross-modal deep interaction enhancement(FD-CMDIE).Firstly,for feature extraction,NeoBERT is utilized to extract high-quality textual features,while stacked long short-term memory(LSTM)networks are employed for visual and acoustic features.Subsequently,common and private encoders are used to decouple the features of the three modalities into similar and dissimilar features.Contrastive learning is then applied,using the textual similar features as anchors,to pull similar features from different modalities closer in the feature space while pushing dissimilar features further apart.Finally,a cross-modal interaction enhancement network is designed for deep interaction and fusion of the decoupled features,and a gated attention pooling module is utilized to filter out noise generated during the interaction.Experiments conducted on two benchmark datasets demonstrate that our proposed method surpasses several state-of-the-art approaches across most metrics,validating its effectiveness.
Aiming at the problem of real‑time multi‑class complex task planning of heterogeneous multi‑unmanned aerial vehicles under obstacle environment， the distributed solution of task assignment model is studied. An improved consensus‑based bundling algorithm （CBBA） is proposed. The marginal cost function of task package construction is redesigned and the path planning cost calculation is coupled based on the rapidly‑exploring random tree star（RRT*） algorithm. The task selection operation is optimized during package construction， and a local on‑demand selection communication optimization conflict resolution mechanism is designed. A local task path set reconstruction mechanism is proposed for solving dynamic task assignment problems. The simulation results show that under the obstacle and dynamic environment， the convergence time of dynamic assignment of the improved algorithm is reduced by 38.1%， the total score is increased by 15.1%， and the total flight distance is reduced by 58.0%. Moreover， it satisfies multiple constraints such as task timing， time window， aircraft resource limitation， dynamic task addition， etc.， and achieves efficient task assignment with high feasibility and real‑time performance.
The seaplane， as a special type of aircraft capable of taking off and landing on water， relies critically on its hydrodynamic performance for navigation and safety. Inverse design of the hydrodynamic shape of a seaplane aims to rapidly generate hull geometries and operational parameters that meet target hydrodynamic performance requirements. However， significant variations in hydrodynamic performance and the relatively low diversity of hull parameters pose considerable challenges for learning effective inverse design models. This paper proposes an inverse design model for seaplanes based on a conditional diffusion model. The method takes the target hydrodynamic performance as a condition and utilizes a designed diffusion module to execute a reverse diffusion process. Starting from Gaussian random noise， the model iteratively denoises the samples step by step， ultimately reconstructing the hull bottom shape parameters corresponding to the given hydrodynamic performance. This study employs actual experimental data derived from a scaled model to validate the performance of the diffusion model. The validation results demonstrate that， in data‑sparse scenarios， the proposed method achieves higher accuracy compared to other machine learning models. This work provides an effective technical approach for the rapid design of seaplanes.
