Research Analyzer
← Back ICRA 2026

Continuous Gaussian Process Pre-Optimization for Asynchronous Event-Inertial Odometry

Zhixiang Wang, Xudong Li, Yizhai Zhang, Fan Zhang, Panfeng Huang

PDF

AI summary

Key figure (auto-extracted from paper)
GPO enables efficient, highly accurate asynchronous event-inertial odometry by modeling inertial preintegration as continuous local Gaussian process trajectories with constant-time querying.
event-inertial fusion continuous-time estimation Gaussian process preintegration asynchronous odometry robotic navigation

Problem

Existing continuous-time event-inertial odometry methods either neglect high-frequency motion information between keyframes or suffer from high computational costs and slow query times, making efficient asynchronous sensor fusion challenging.

Approach

The authors propose GPO, which models raw IMU measurements as two local Temporal Gaussian Process trajectories and uses a lightweight two-step optimization to infer continuous pseudo-measurements, enabling constant-time state queries and analytical Jacobian propagation.

Key results

  • Linear solving and constant-time querying for continuous preintegration
  • Analytical Jacobian propagation strategy for arbitrary timestamps
  • Tightly-coupled asynchronous event-inertial odometry pipeline
  • Superior accuracy and efficiency over state-of-the-art GP methods on public and real-world datasets

Why it matters

Provides a computationally efficient and accurate foundation for real-time ego-motion estimation in high-speed, high-dynamic-range robotic applications.

Abstract

Event cameras, as bio-inspired sensors, are asyn- chronously triggered with high-temporal resolution compared to intensity cameras. Recent work has focused on fusing the event measurements with inertial measurements to enable ego-motion estimation in high-speed and HDR environments. However, exist- ing methods predominantly rely on IMU preintegration designed mainly for synchronous sensors and discrete-time frameworks. In this paper, we propose GPO, a continuous-time preinte- gration framework that can efficiently achieve tightly-coupled fusion of fully asynchronous sensors. Concretely, we model the preintegration as two local Temporal Gaussian Process (TGP) trajectories and leverage a light-weight two-step optimization to infer the continuous preintegration pseudo-measurements. We show that the Jacobians of arbitrary queried states can be naturally propagated using our framework, which enables GPO to be involved in the asynchronous fusion. Our method realizes a linear and constant time cost for optimization and query, respectively. To further validate the proposal, we leverage GPO to design an asynchronous event-inertial odometry and compare with other asynchronous fusion schemes. Experiments conducted on both public and own-collected datasets demonstrate that the proposed GPO offers significant advantages in terms of accuracy and efficiency, outperforming existing approaches in handling asynchronous sensor fusion. The code of GPO can be found at https://github.com/NPU-RCIR/GPO-Preint.

Index terms

Sensor Fusion Visual-Inertial SLAM Vision-Based Navigation

Related papers