This work establishes a simple algorithm to recover an information vector from a predefined database available every time. It is considered that the information analyzed may be incomplete, damaged, or corrupted. This algorithm is inspired by Hopfield Neural Networks (HNN), which allows the recursive reconstruction of an information vector through an energy-minimizing optimal process, but this paper presents a procedure that generates results in a single iteration. Images have been chosen for the information recovery application to build the vector information. In addition, a filter is added to the algorithm to focus on the most important information when reconstructing data, allowing it to work with damaged or incomplete vectors, even without losing the ability to be a non-iterative process. A brief theoretical introduction and a numerical validation for recovery information are shown with an example of a database containing 40 images.

This paper addresses the data-driven identification of open-chain robot morphology from finite windows of heterogeneous signals, including commanded joint references, measured joint states, and end-effector pose observations. Unlike conventional calibration procedures that assume a known kinematic topology, the proposed formulation estimates both discrete structural quantities and continuous kinematic coordinates: the number of active joints, the revolute/prismatic token sequence, Product-of-Exponentials (POE) screw axes, and the home pose of the end effector. A temporal transformer encoder is used as the main estimator and compared with a gated recurrent unit (GRU) baseline on the same dataset, with the same output heads and a multitask physics-aware objective. The continuous target is expressed in POE coordinates rather than as a Denavit–Hartenberg table because POE directly represents spatial joint axes and avoids several frame-assignment ambiguities. Simulated results on a noisy benchmark of 48 serial-robot families show that both sequence models recover the discrete structure on the tested in-library trajectories, while their continuous reconstruction errors reveal different trade-offs in screw-axis, home-pose, and trajectory reconstruction accuracy. The study also discusses inactive-slot masking, out-of-library behavior, synthetic-to-real limitations, persistent excitation, and the role of the learned model as an initialization for subsequent calibration refinement.

This paper presents a fault-diagnosis methodology for detecting progressive faults in quadrotor drones using a controlled Neural ODE network and a GRU recurrent baseline. Training and evaluation data were obtained from a numerical simulation of the physical system, including vehicle dynamics, battery electrical behavior, motor thermal response, and randomly scheduled long-duration degradations. The fault modes considered are battery degradation, loss of motor effectiveness, and overheating. Simulated telemetry was corrupted with measurement noise and slow sensor drift to approximate realistic operating conditions. The analysis compares a continuous latent-space formulation with a discrete recurrent architecture under the same temporal-window representation, class definitions, and evaluation criteria. The proposed framework provides a reproducible basis for studying early incipient fault diagnosis in unmanned aerial vehicles prior to experimental validation using flight logs.

This study develops a sparse-sensor physics-informed neural network (PINN) to reconstruct a transient two-dimensional thermalfield and identify thermal diffusivity, volumetric heat loss, and source amplitude from noisy point measurements. Sixteen interiorsensors provide 5424 observations for a moving-source benchmark. The nominal model achieves a mean root-mean-square error(RMSE) of 3.52 × 10−3 normalized temperature units and a global relative L2 error of 2.87 × 10−2. The corresponding relativeerrors in the three identified coefficients are 5.04%, 24.54%, and 7.90%. Under matched data, architecture, initialization, anditeration budget, the physics-informed model reduces mean RMSE by 96.03% relative to a conventional data-only neural networkand by 95.56% relative to spatiotemporal interpolation. One-factor-at-a-time tests indicate that spatial sensor scarcity and lossweighting are more influential than temporal subsampling over the ranges examined. The results show that governing-equationconstraints provide strong regularization for sparse thermal reconstruction; however, the sensitivity results are diagnostic ratherthan statistical, and experimental validation remains necessary before deployment