Vision-based object detectors are a crucial basis for robotics applications as they provide valuable information about object localisation in the environment. These need to ensure high reliability in different lighting conditions, occlusions, and visual artifacts, all while running in real-time. Collecting and annotating real-world data for these networks is prohibitively time consuming and costly, especially for custom assets, such as industrial objects, making it untenable for generalization to in-the-wild scenarios. To this end, we present Synthetica, a method for large-scale synthetic data generation for training robust state estimators. This paper focuses on the task of object detection, an important problem which can serve as the front-end for most state estimation problems, such as pose estimation. Leveraging data from a photorealistic ray-tracing renderer, we scale up data generation, generating 2.7 million images, to train highly accurate real-time detection transformers. We present a collection of rendering randomization and training-time data augmentation techniques conducive to robust sim-to-real performance for vision tasks. We demonstrate state-of-the-art performance on the task of object detection while having detectors that run at 50-100Hz which is 9 times faster than the prior SOTA. We further demonstrate the usefulness of our training methodology for robotics applications by showcasing a pipeline for use in the real world with custom objects for which there do not exist prior datasets. Our work highlights the importance of scaling synthetic data generation for robust sim-to-real transfer while achieving the fastest real-time inference speeds. Videos and supplementary information can be found at this URL: https://sites.google.com/view/synthetica-vision.
This paper addresses autonomous racing by introducing a real-time nonlinear model predictive controller (NMPC) coupled with a moving horizon estimator (MHE). The racing problem is solved by an NMPC-based off-line trajectory planner that computes the best trajectory while considering the physical limits of the vehicle and circuit constraints. The developed controller is further enhanced with a learning extension based on Gaussian process regression that improves model predictions. The proposed control, estimation, and planning schemes are evaluated on two different race tracks.
Accurate state estimates are required for increasingly complex systems, to enable, for example, feedback control. However, available state estimation schemes are not necessarily real-time feasible for certain large-scale systems. Therefore, we develop in this paper, a real-time feasible state-estimation scheme for a class of large-scale systems that approximates the steady state Kalman filter. In particular, we focus on systems where the state-vector is the result of discretizing the spatial domain, as typically seen in Partial Differential Equations. In such cases, the correlation between states in the state-vector often have an intuitive interpretation on the spatial domain, which can be exploited to obtain a significant reduction in computational complexity, while still providing accurate state estimates. We illustrate these strengths of our method through a hyperthermia cancer treatment case study. The results of the case study show significant improvements in the computation time, while simultaneously obtaining good state estimates, when compared to Ensemble Kalman filters and Kalman filters using reduced-order models.
This paper introduces "invisible manipulation," an innovative cyber-attack mechanism achieved through strategically timed stealthy false data injection attacks (SFDIAs). By stealthily manipulating measurements of a critical asset prior to the target time period, the attacker can subtly guide the engineering system toward a predetermined operational state without detection. Using the battery energy management system (BEMS) as a case study, we employ deep reinforcement learning (DRL) to generate synthetic measurements, such as battery voltage and current, that align closely with actual measurements. These synthetic measurements, falling within the acceptable error margin of residual-based bad data detection algorithm provided by state estimation, can evade detection and mislead Extended Kalman-filter-based State of Charge estimation. Subsequently, considering the deceptive data as valid inputs, the BEMS will operate the BESS towards the attacker desired operational states when the targeted time period come. The use of the DRL-based scheme allows us to covert an online optimization problem into an offline training process, thereby alleviating the computational burden for real-time implementation. Comprehensive testing on a high-fidelity microgrid real-time simulation testbed validates the effectiveness and adaptability of the proposed methods in achieving different attack objectives.
Being able to predict battery internal states that are related to battery degradation is a key aspect to improve battery lifetime and performance, enhancing cleaner electric transportation and energy generation. However, most present battery management systems (BMSs) use equivalent-circuit models (ECMs) for state of charge (SOC) and state of health (SOH) estimation. These models are not able to predict these aging-related variables, and therefore, they cannot be used to limit battery degradation. In this paper, we propose a method for electrode-level SOC (eSOC) and electrode-level SOH (eSOH) estimation using an electrode-level ECM (eECM). The method can produce estimates of the states of lithiation (SOL) of both electrodes and update the eSOH parameters to maintain estimation accuracy through the lifetime of the battery. Furthermore, the eSOH parameter estimates are used to obtain degradation mode information, which could be used to improve state estimation, health diagnosis and prognosis. The method was validated in simulation and experimentally.
When tracking maneuvering targets, model-driven approaches encounter difficulties in comprehensively delineating complex real-world scenarios and are prone to model mismatch when the targets maneuver. Meanwhile, contemporary data-driven methods have overlooked measurements' confidence, markedly escalating the challenge of fitting a mapping from measurement sequences to target state sequences. To address these issues, this paper presents a deep maneuvering target tracking methodology based on target state space projection. The proposed methodology initially establishes a projection from the target measurement sequence to the target state space by formulating the probability density function of measurement error and samples the distribution information of measurement noise within the target state space as a measurement representation. Under this representation, the sequential regression task of target state estimation can be transmuted into a task of detecting the target location in the state space. Subsequently, a deep detection network is devised to accomplish target location detection in the target state space. Finally, a loss function is designed to facilitate the network's training for attaining the desired network performance. Simulation experiments suggest that the proposed method can maintain satisfactory tracking performance even when the target maneuvers, and can rapidly converge and achieve higher estimation accuracy compared with existing methods after the target maneuvers.
In this paper, we study the collaborative state fusion problem in a multi-agent environment, where mobile agents collaborate to track movable targets. Due to the limited sensing range and potential errors of on-board sensors, it is necessary to aggregate individual observations to provide target state fusion for better target state estimation. Existing schemes do not perform well due to (1) impractical assumption of the fully known prior target state-space model and (2) observation outliers from individual sensors. To address the issues, we propose a two-stage collaborative fusion framework, namely \underline{L}earnable Weighted R\underline{o}bust \underline{F}usion (\textsf{LoF}). \textsf{LoF} combines a local state estimator (e.g., Kalman Filter) with a learnable weight generator to address the mismatch between the prior state-space model and underlying patterns of moving targets. Moreover, given observation outliers, we develop a time-series soft medoid(TSM) scheme to perform robust fusion. We evaluate \textsf{LoF} in a collaborative detection simulation environment with promising results. In an example setting with 4 agents and 2 targets, \textsf{LoF} leads to a 9.1\% higher fusion gain compared to the state-of-the-art.
Discrete Empirical Interpolation Method (DEIM) estimates a function from its pointwise incomplete observations. In particular, this method can be used to estimate the state of a dynamical system from observational data gathered by sensors. However, when the number of observations are limited, DEIM returns large estimation errors. Sparse DEIM (S-DEIM) was recently developed to address this problem by introducing a kernel vector which previous DEIM-based methods had ignored. Unfortunately, estimating the optimal kernel vector in S-DEIM is a difficult task. Here, we introduce a data-driven method to estimate this kernel vector from sparse observational time series using recurrent neural networks. Using numerical examples, we demonstrate that this machine learning approach together with S-DEIM leads to nearly optimal state estimations.
Despite extensive research in robust visual-inertial navigation systems~(VINS) in dynamic environments, many approaches remain vulnerable to objects that suddenly start moving, which are referred to as \textit{abruptly dynamic objects}. In addition, most approaches have considered the effect of dynamic objects only at the feature association level. In this study, we observed that the state estimation diverges when errors from false correspondences owing to moving objects incorrectly propagate into the IMU bias terms. To overcome these problems, we propose a robust VINS framework called \mbox{\textit{DynaVINS++}}, which employs a) adaptive truncated least square method that adaptively adjusts the truncation range using both feature association and IMU preintegration to effectively minimize the effect of the dynamic objects while reducing the computational cost, and b)~stable state recovery with bias consistency check to correct misestimated IMU bias and to prevent the divergence caused by abruptly dynamic objects. As verified in both public and real-world datasets, our approach shows promising performance in dynamic environments, including scenes with abruptly dynamic objects.
Autonomous vehicles performing navigation tasks in complex environments face significant challenges due to uncertainty in state estimation. In many scenarios, such as stealth operations or resource-constrained settings, accessing high-precision localization comes at a significant cost, forcing robots to rely primarily on less precise state estimates. Our key observation is that different tasks require varying levels of precision in different regions: a robot navigating a crowded space might need precise localization near obstacles but can operate effectively with less precision elsewhere. In this paper, we present a planning method for integrating task-specific uncertainty requirements directly into navigation policies. We introduce Task-Specific Uncertainty Maps (TSUMs), which abstract the acceptable levels of state estimation uncertainty across different regions. TSUMs align task requirements and environmental features using a shared representation space, generated via a domain-adapted encoder. Using TSUMs, we propose Generalized Uncertainty Integration for Decision-Making and Execution (GUIDE), a policy conditioning framework that incorporates these uncertainty requirements into robot decision-making. We find that TSUMs provide an effective way to abstract task-specific uncertainty requirements, and conditioning policies on TSUMs enables the robot to reason about the context-dependent value of certainty and adapt its behavior accordingly. We show how integrating GUIDE into reinforcement learning frameworks allows the agent to learn navigation policies that effectively balance task completion and uncertainty management without explicit reward engineering. We evaluate GUIDE on various real-world robotic navigation tasks and find that it demonstrates significant improvement in task completion rates compared to baseline methods that do not explicitly consider task-specific uncertainty.
State Estimation is a crucial task in power systems. Graph Neural Networks have demonstrated significant potential in state estimation for power systems by effectively analyzing measurement data and capturing the complex interactions and interrelations among the measurements through the system's graph structure. However, the information about the system's graph structure may be inaccurate due to noise, attack or lack of accurate information about the topology of the system. This paper studies these scenarios under topology uncertainties and evaluates the impact of the topology uncertainties on the performance of a Temporal Graph Convolutional Network (TGCN) for state estimation in power systems. In order to make the model resilient to topology uncertainties, modifications in the TGCN model are proposed to incorporate a knowledge graph, generated based on the measurement data. This knowledge graph supports the assumed uncertain system graph. Two variations of the TGCN architecture are introduced to integrate the knowledge graph, and their performances are evaluated and compared to demonstrate improved resilience against topology uncertainties. The evaluation results indicate that while the two proposed architecture show different performance, they both improve the performance of the TGCN state estimation under topology uncertainties.
The paper deals with measures of nonlinearity. In state estimation, they are utilized i) to select a suitable state estimation algorithm by assessing the nonlinearity of a system model, ii) to adapt the estimation algorithm structure or parameters, or iii) to indicate the possible effect of strong nonlinearity that leads to estimate credibility loss. This paper summarizes the state of the art of nonlinearity measures, focusing on the mean-square-error-based measure of nonlinearity. Its weak point related to unit selection is illustrated, and based on this, requirements for a new measure of nonlinearity are formulated. A new nonlinearity measure that is both unitless and normalized is designed. Its properties are demonstrated using numerical tracking experiments.
The Kalman filter (KF) and its variants are among the most celebrated algorithms in signal processing. These methods are used for state estimation of dynamic systems by relying on mathematical representations in the form of simple state-space (SS) models, which may be crude and inaccurate descriptions of the underlying dynamics. Emerging data-centric artificial intelligence (AI) techniques tackle these tasks using deep neural networks (DNNs), which are model-agnostic. Recent developments illustrate the possibility of fusing DNNs with classic Kalman-type filtering, obtaining systems that learn to track in partially known dynamics. This article provides a tutorial-style overview of design approaches for incorporating AI in aiding KF-type algorithms. We review both generic and dedicated DNN architectures suitable for state estimation, and provide a systematic presentation of techniques for fusing AI tools with KFs and for leveraging partial SS modeling and data, categorizing design approaches into task-oriented and SS model-oriented. The usefulness of each approach in preserving the individual strengths of model-based KFs and data-driven DNNs is investigated in a qualitative and quantitative study, whose code is publicly available, illustrating the gains of hybrid model-based/data-driven designs. We also discuss existing challenges and future research directions that arise from fusing AI and Kalman-type algorithms.
This paper addresses the optimal state estimation problem for dynamic systems while preserving private information against an adversary. To dominate the adversary's estimation accuracy about private information in the mean square error (MSE) sense, the Cram\'er-Rao lower bound (CRLB) is employed to evaluate privacy level. The problem is formulated as a constrained optimization, which minimizes the MSE of the state estimate with a constraint on privacy level, achieving a trade-off between privacy and utility. To solve the constrained optimization problem, an explicit expression for CRLB is first provided using the information inequality. To overcome the increasing sizes of the involved matrices over time, a low-complexity approach is then proposed to achieve online calculation for CRLB, significantly reducing computational complexity. Next, the optimization problem is relaxed to a semi-definite programming problem, and a relaxed solution is provided. Finally, a privacy-preserving state estimation algorithm with low complexity is developed and proved to achieve differential privacy. Two illustrative examples, including a practical case of building occupancy, demonstrate the effectiveness of the proposed algorithm.
Training large models with millions or even billions of parameters from scratch incurs substantial computational costs. Parameter Efficient Fine-Tuning (PEFT) methods, particularly Low-Rank Adaptation (LoRA), address this challenge by adapting only a reduced number of parameters to specific tasks with gradient-based optimizers. In this paper, we cast PEFT as an optimal filtering/state estimation problem and present Low-Rank Kalman Optimizer (LoKO) to estimate the optimal trainable parameters in an online manner. We leverage the low-rank decomposition in LoRA to significantly reduce matrix sizes in Kalman iterations and further capitalize on a diagonal approximation of the covariance matrix to effectively decrease computational complexity from quadratic to linear in the number of trainable parameters. Moreover, we discovered that the initialization of the covariance matrix within the Kalman algorithm and the accurate estimation of the observation noise covariance are the keys in this formulation, and we propose robust approaches that work well across a vast range of well-established computer vision and language models. Our results show that LoKO converges with fewer iterations and yields better performance models compared to commonly used optimizers with LoRA in both image classifications and language tasks. Our study opens up the possibility of leveraging the Kalman filter as an effective optimizer for the online fine-tuning of large models.
Despite substantial progress in recent years, probabilistic solvers with adaptive step sizes can still not solve memory-demanding differential equations -- unless we care only about a single point in time (which is far too restrictive; we want the whole time series). Counterintuitively, the culprit is the adaptivity itself: Its unpredictable memory demands easily exceed our machine's capabilities, making our simulations fail unexpectedly and without warning. Still, dropping adaptivity would abandon years of progress, which can't be the answer. In this work, we solve this conundrum. We develop an adaptive probabilistic solver with fixed memory demands building on recent developments in robust state estimation. Switching to our method (i) eliminates memory issues for long time series, (ii) accelerates simulations by orders of magnitude through unlocking just-in-time compilation, and (iii) makes adaptive probabilistic solvers compatible with scientific computing in JAX.
A fundamental challenge in developing data-driven approaches to ecological systems for tasks such as state estimation and prediction is the paucity of the observational or measurement data. For example, modern machine-learning techniques such as deep learning or reservoir computing typically require a large quantity of data. Leveraging synthetic data from paradigmatic nonlinear but non-ecological dynamical systems, we develop a meta-learning framework with time-delayed feedforward neural networks to predict the long-term behaviors of ecological systems as characterized by their attractors. We show that the framework is capable of accurately reconstructing the ``dynamical climate'' of the ecological system with limited data. Three benchmark population models in ecology, namely the Hastings-Powell model, a three-species food chain, and the Lotka-Volterra system, are used to demonstrate the performance of the meta-learning based prediction framework. In all cases, enhanced accuracy and robustness are achieved using five to seven times less training data as compared with the corresponding machine-learning method trained solely from the ecosystem data. A number of issues affecting the prediction performance are addressed.
Many commercial and defense applications involve multisensor, multitarget tracking, requiring the fusion of information from a set of sensors. An interesting use case occurs when data available at a central node (due to geometric diversity or retrodiction) allows for the tailoring of state estimation for a target. For instance, if a target is initially tracked with a maneuvering target filter, yet the target is clearly not maneuvering in retrospect, it would be beneficial at the fusion node to refilter that data with a non-maneuvering target filter. If measurements can be shared to the central node, the refiltering process can be accomplished by simply passing source measurements through an updated state estimation process. It is often the case for large, distributed systems, however, that only track information can be passed to a fusion center. In this circumstance, refiltering data becomes less straightforward as track states are linearly dependent across time, and the correlation needs to be properly accounted for before/during refiltering. In this work, a model-based temporal decorrelation process for state estimates with process noise will be studied. A decorrelation procedure will be presented based on a linear algebraic formulation of the problem, and process noise estimates will be created that ensure a conservative system state estimate. Numerical examples will be given to demonstrate the efficacy of the proposed algorithm.
In this paper we study the question of how to reconstruct the state of a power system using Phasor Measurement Units (PMUs). In our previous research we proved that this question has an affirmative answer imposing some rather strict structural assumptions: namely, neglecting the generator rotors saliency and assuming that the stator resistance of the synchronous generator is zero. It was shown in simulations that the performance of the proposed observer was sensitive to these assumptions, observing a transient quality degradation for realistic simulations not imposing these assumptions. Moreover, it was assumed in our previous work that the mechanical power and the field voltage are available for measurement, a scenario that it is not always realistic. In this paper we accomplish two ambitious objectives. First, we propose a new observer that does not impose the simplifying assumptions on the generator model. Secondly, we consider the more realistic scenario where only mechanical power is available for measurement. That is, we solve a problem of state reconstruction of a nonlinear system with partially known input measurements -- that is well-known to be a very challenging task. The design of the first observer relies on two recent developments proposed by the authors, a parameter estimation based approach to the problem of state estimation and the use of the Dynamic Regressor Extension and Mixing (DREM) technique to estimate these parameters. The use of DREM allows us to overcome the problem of lack of persistent excitation that stymies the application of standard parameter estimation designs. On the other hand, the observer for the partial input measurement scenario relies on the clever exploitation of the systems model. Simulation results illustrates the good performance of the proposed observers.
Many scientific fields collect longitudinal count compositional data. Each observation is a multivariate count vector, where the total counts are arbitrary, and the information lies in the relative frequency of the counts. Multiple authors have proposed Bayesian Multinomial Logistic-Normal Dynamic Linear Models (MLN-DLMs) as a flexible approach to modeling these data. However, adoption of these methods has been limited by computational challenges. This article develops an efficient and accurate approach to posterior state estimation, called $\textit{Fenrir}$. Our approach relies on a novel algorithm for MAP estimation and an accurate approximation to a key posterior marginal of the model. As there are no equivalent methods against which we can compare, we also develop an optimized Stan implementation of MLN-DLMs. Our experiments suggest that Fenrir can be three orders of magnitude more efficient than Stan and can even be incorporated into larger sampling schemes for joint inference of model hyperparameters. Our methods are made available to the community as a user-friendly software library written in C++ with an R interface.