This article addresses the critical task of visual geolocalization for unmanned aerial vehicles (UAVs) in GNSS-denied environments. It focuses on overcoming the limitations of classical feature-based methods, which are highly sensitive to photometric and textural variations in aerial images captured at different times and under different conditions. The paper proposes a comprehensive approach based on a novel convolutional neural network, CADE-Net, which introduces an integration of an adaptive architecture, a cross-attention mechanism, and a hybrid loss function. The core challenge lies in matching a current onboard aerial image against a georeferenced satellite or aerial map from a database despite significant changes caused by varying seasons, weather, and lighting conditions. The proposed CADE-Net architecture is specifically designed to tackle these challenges. It employs a two-stream structure with deformable convolutions to achieve geometric invariance to scale and rotation. Furthermore, a cross-attention mechanism is integrated between the streams to explicitly model the relationships between image pairs, enabling the network to focus on semantically correspondent regions despite their visual differences. The training process is enhanced by a hybrid loss function that combines metric learning principles with an adversarial approach. A key innovation is the intelligent mining strategy for hard negatives, which forces the model to learn fine-grained details by distinguishing between structurally similar but semantically different objects. Proposed approach can generate robust, invariant features that are resilient to complex, non-linear distortions. This work is of significant practical importance for developing fully autonomous and reliable UAV navigation systems capable of operating effectively without GNSS signals.
This article addresses the critical challenge of developing optimal control algorithms for ballistically linked groups (BLGs) of small satellites within broadband communication constellations. The research is driven by the global shift towards large-scale low Earth orbit (LEO) megaconstellations, where precise, long-term, and fuel-efficient maintenance of the orbital structure is paramount. The core difficulty stems from the severe mass, size, and power constraints of CubeSats, which render traditional high-propellant control strategies impractical. The primary objective is to design, compare, and validate algorithms for accurate formation flying while minimizing propellant consumption, thereby extending mission lifetime and reducing operational costs. The study uses model relative motion in a near-circular orbit. Transfer of a deputy spacecraft to a 100 km along-track separation serves as the test case. The first method applies Pontryagin’s maximum principle, solving the two-point boundary value problem via a Newton metod. The second employs a predictive control framework. The third strategy is based on parametric optimization of a predefined control input structure, with subsequent gradient-based correction. Numerical simulations for different transfer durations confirm the efficacy of all methods. The Newton method provides exceptional accuracy in meeting terminal state constraints at a fixed final time. The predictive controller demonstrates superior fuel economy by incorporating extended passive coasting arcs. The parametric optimization approach offers implementation flexibility, albeit with sensitivity to initial parameter guesses. This comparative analysis confirms the practical viability of these algorithms for the autonomous, fuel-conscious station-keeping and reconfiguration of future small-satellite communication swarms, a vital capability for next-generation global connectivity networks.
This paper presents methods for in-situ monitoring of holographic recording setups used for fabrication of volume holographic elements. The approach combines conventional analysis of the interference pattern in the recording plane with real-time phase diagnostics using a probe interferometer. Phase drift, vibration-induced noise and intensity fluctuations are converted into quantitative stability metrics and related to the expected diffraction efficiency and spatial uniformity of the recorded hologram.
An analysis of the current state of laser technology based on broadband solid-state active media with radiation in the range of 780-900 nm is presented. Methods of using Bragg gratings for radiation control and optical coupling of cavities of laser systems are presented.
The translational effects of gas streams, which form after the triple-shock configurations at Mach reflection of blast waves with normal main shock (so-called stationary Mach configurations), were analyzed. Unlike in the case of an elevated explosions of fuel as rocket starts in initially stagnant air, which is considered here as a private case, it was supposed that this shock-wave structure moves in a preceding flow with arbitrary velocity (and corresponding flow Mach number). Analyzing relations of the dynamic pressures across the slipstream, which emanates from the triple point of the Mach reflection, it was shown that the flows after the triple-shock configuration usually differ much in their translational action on surrounding objects. It was found and discussed that some configurations drag the objects initially situated above and below the triple-point trajectory in opposite directions. Moreover, the “trigger” structure was found that remains previous flow drag action on the object above the triple-point trajectory, but switches it to exactly opposite one, if the object is situated below the triple point.
