Real-Time Performance Analysis of Multi-Fidelity Residual Physics-Informed Neural Process-Based State Estimation for Robotic Systems

2025-11-11

AI Summary

To address the unreliable data-driven model prediction error bounds in real-time nonlinear state estimation for safety-critical robotic systems, this paper proposes the **Multi-Fidelity Residual Physics-Informed Neural Process (MF-Res-PINP)** framework. Our method integrates residual modeling based on low-order physics, structural prior embedding via Physics-Informed Neural Processes (PINPs), and distribution-free uncertainty quantification using Split Conformal Prediction—operating within a hybrid online learning paradigm to enable high-confidence real-time estimation. Experimental results demonstrate that MF-Res-PINP significantly outperforms the Unscented Kalman Filter (UKF) and Deep Kalman Filter (DKF) in estimation accuracy, robustness, and—crucially—the reliability of predicted error bounds. Notably, it maintains strong generalization under dynamic disturbances and data-scarce conditions, thereby enhancing safety guarantees in real-world deployment.

Technology Category

• Intelligent Robots: State Estimation
• Machine Learning: Calibration & Uncertainty Quantification
• Natural Language Processing: Safety and Robustness

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Various neural network architectures are used in many of the state-of-the-art approaches for real-time nonlinear state estimation. With the ever-increasing incorporation of these data-driven models into the estimation domain, model predictions with reliable margins of error are a requirement -- especially for safety-critical applications. This paper discusses the application of a novel real-time, data-driven estimation approach based on the multi-fidelity residual physics-informed neural process (MFR-PINP) toward the real-time state estimation of a robotic system. Specifically, we address the model-mismatch issue of selecting an accurate kinematic model by tasking the MFR-PINP to also learn the residuals between simple, low-fidelity predictions and complex, high-fidelity ground-truth dynamics. To account for model uncertainty present in a physical implementation, robust uncertainty guarantees from the split conformal (SC) prediction framework are modeled in the training and inference paradigms. We provide implementation details of our MFR-PINP-based estimator for a hybrid online learning setting to validate our model's usage in real-time applications. Experimental results of our approach's performance in comparison to the state-of-the-art variants of the Kalman filter (i.e. unscented Kalman filter and deep Kalman filter) in estimation scenarios showed promising results for the MFR-PINP model as a viable option in real-time estimation tasks.

Problem

• Addressing model-mismatch in robotic state estimation
• Learning residuals between low and high-fidelity dynamics
• Providing robust uncertainty guarantees for safety-critical applications

Innovation

• Multi-fidelity residual physics-informed neural process
• Learning residuals between low and high fidelity dynamics
• Robust uncertainty guarantees via split conformal prediction