Tag
This paper introduces Unscented KalmanNet (UKN), a hybrid deep learning filter that augments the Unscented Kalman Filter with learned components to improve state estimation accuracy and covariance calibration under unknown noise statistics and model mismatch. Experiments show significant RMSE reductions over UKF and other KalmanNet variants.
Proposes Generative Bayesian Filtering (GBF), a framework that replaces restrictive observation models with pretrained conditional generative models for state estimation, improving accuracy and robustness in synthetic and real-world applications.
STOCKTAKE is a 26-week supply-chain benchmark using a POMDP with a fair oracle to separately measure failures of perception and action in LLM agents. Results show that agents often correctly diagnose hidden state changes but fail to act appropriately, indicating a gap between stated beliefs and costly actions.
A short introduction to quadcopter modeling, state estimation, motion planning, and control, providing a jumping-off point for newcomers.
This paper presents CA-NKCF, a novel distributed latent state estimator combining partial domain knowledge with deep neural networks, achieving robust performance without noise statistics knowledge, outperforming traditional filters in linear, chaotic, and wireless tracking environments.
This paper investigates using reinforcement learning to train observable control policies that enable estimation of an agent's state from its actions, with applications in multiagent coordination and monitoring under communications constraints.
This paper introduces a structured parameterization for noise models in ELTO-based Kalman filters, enabling dynamic adaptation to non-stationary processes and improving state estimation performance in noisy, time-varying environments.
This paper presents a theoretical framework interpreting Transformer components (attention, residual connections, normalization) as arising from a spherical state estimation problem using Radial-Tangential SDEs.
The article describes a personal project to solve range-only relative localization using two microcontrollers with IMUs and UWB, allowing devices to find each other using distance and motion data, inspired by the game Marco Polo.
This paper introduces Christoffel-DPS, a distribution-free framework for optimal sensor placement in diffusion posterior sampling that outperforms classical Gaussian-based methods. It provides theoretical guarantees and practical improvements for reconstructing states from complex, non-Gaussian distributions using generative models.