PedestrianDiffusion: Multimodal Generative Denoising and Dense State Estimation for Inertial Navigation

2026-07-03

AI Summary

This work addresses the limitations of consumer-grade inertial navigation systems, which suffer from MEMS sensor noise and the inability of conventional neural networks to simultaneously preserve high-frequency motion fidelity and numerical stability. The authors propose a frequency-domain generative diffusion model that reformulates 6D state estimation as a continuous conditional denoising process. They introduce spectral covariance constraints to stabilize diffusion trajectories—a first in this domain—and integrate vision-language embeddings to enable zero-shot semantic conditioning. By employing a single-step deterministic probability flow ODE solver, the method supports real-time batch trajectory refinement on edge devices. Evaluated on OxIOD, RIDI, RoNIN, and TLIO benchmarks, the approach achieves state-of-the-art performance, demonstrating significantly enhanced robustness against impulsive disturbances and 6D coupled drift, thereby validating the efficacy of generative modeling for inertial navigation.

Abstract

The accuracy of consumer-grade inertial navigation is bottlenecked by the stochastic noise of Micro-Electro-Mechanical Systems (MEMS). Traditional deterministic neural architectures often succumb to ``estimation jittering,'' sacrificing high-frequency kinematic fidelity for numerical stability. We propose PedestrianDiffusion, a multimodal spectral-domain generative framework reformulating dense 6D state estimation as a continuous conditional denoising process. By operating in the frequency domain, our formulation bounds the spectral covariance, acting as a mathematical preconditioner to stabilize the reverse diffusion trajectory. Furthermore, we introduce a zero-shot semantic conditioning mechanism leveraging vision-language embeddings as categorical priors to generalize across heterogeneous sensor noise profiles. To address the computational intractability of generative tracking, we deploy a single-step deterministic probability flow ODE solver ($T=1$). This yields high-capacity asynchronous batch trajectory refinement, establishing the viability of generative architectures for asynchronous batch trajectory refinement on edge hardware. Extensive evaluations on the OxIOD, RIDI, RoNIN, and TLIO benchmarks demonstrate that PedestrianDiffusion achieves state-of-the-art performance, exhibiting unprecedented robustness to impulse perturbations and coupled 6D kinematic drift. This work provides a rigorous algorithmic blueprint for next-generation Neural Inertial Measurement Units (N-IMUs).

Problem

• inertial navigation
• MEMS noise
• estimation jittering
• 6D state estimation
• kinematic drift

Innovation

• generative denoising
• spectral-domain modeling
• zero-shot semantic conditioning
• asynchronous trajectory refinement
• neural inertial navigation