Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Spiking Neural-Invariant Kalman Fusion for Accurate Localization Using Low-Cost IMUs

Yaohua Liu, Qiao Xu, Yemin Wang, Hui Yi Leong, Binkai Ou

TL;DR

The paper tackles accurate localization using low-cost MEMS IMUs by introducing a brain-inspired hybrid framework that fuses a spiking neural network (SNN) with an invariant extended Kalman filter (InEKF). The SNN denoises IMU signals and dynamically estimates the InEKF noise covariances, enabling robust dead reckoning with minimal computational overhead via a spiking Transformer backbone and surrogate-gradient training. Pseudo-measurements from nonholonomic constraints (lateral/up velocities) are incorporated into the InEKF, and the system is trained end-to-end with a Huber loss. Extensive KITTI-based experiments and real-field tests demonstrate improved localization accuracy and robustness to sensor noise and data loss, with favorable energy efficiency on neuromorphic hardware, indicating strong potential for real-world mobile robotics applications.

Abstract

Low-cost inertial measurement units (IMUs) are widely utilized in mobile robot localization due to their affordability and ease of integration. However, their complex, nonlinear, and time-varying noise characteristics often lead to significant degradation in localization accuracy when applied directly for dead reckoning. To overcome this limitation, we propose a novel brain-inspired state estimation framework that combines a spiking neural network (SNN) with an invariant extended Kalman filter (InEKF). The SNN is designed to extract motion-related features from long sequences of IMU data affected by substantial random noise and is trained via a surrogate gradient descent algorithm to enable dynamic adaptation of the covariance noise parameter within the InEKF. By fusing the SNN output with raw IMU measurements, the proposed method enhances the robustness and accuracy of pose estimation. Extensive experiments conducted on the KITTI dataset and real-world data collected using a mobile robot equipped with a low-cost IMU demonstrate that the proposed approach outperforms state-of-the-art methods in localization accuracy and exhibits strong robustness to sensor noise, highlighting its potential for real-world mobile robot applications.

Spiking Neural-Invariant Kalman Fusion for Accurate Localization Using Low-Cost IMUs

TL;DR

The paper tackles accurate localization using low-cost MEMS IMUs by introducing a brain-inspired hybrid framework that fuses a spiking neural network (SNN) with an invariant extended Kalman filter (InEKF). The SNN denoises IMU signals and dynamically estimates the InEKF noise covariances, enabling robust dead reckoning with minimal computational overhead via a spiking Transformer backbone and surrogate-gradient training. Pseudo-measurements from nonholonomic constraints (lateral/up velocities) are incorporated into the InEKF, and the system is trained end-to-end with a Huber loss. Extensive KITTI-based experiments and real-field tests demonstrate improved localization accuracy and robustness to sensor noise and data loss, with favorable energy efficiency on neuromorphic hardware, indicating strong potential for real-world mobile robotics applications.

Abstract

Low-cost inertial measurement units (IMUs) are widely utilized in mobile robot localization due to their affordability and ease of integration. However, their complex, nonlinear, and time-varying noise characteristics often lead to significant degradation in localization accuracy when applied directly for dead reckoning. To overcome this limitation, we propose a novel brain-inspired state estimation framework that combines a spiking neural network (SNN) with an invariant extended Kalman filter (InEKF). The SNN is designed to extract motion-related features from long sequences of IMU data affected by substantial random noise and is trained via a surrogate gradient descent algorithm to enable dynamic adaptation of the covariance noise parameter within the InEKF. By fusing the SNN output with raw IMU measurements, the proposed method enhances the robustness and accuracy of pose estimation. Extensive experiments conducted on the KITTI dataset and real-world data collected using a mobile robot equipped with a low-cost IMU demonstrate that the proposed approach outperforms state-of-the-art methods in localization accuracy and exhibits strong robustness to sensor noise, highlighting its potential for real-world mobile robot applications.
Paper Structure (17 sections, 29 equations, 10 figures, 2 tables)

This paper contains 17 sections, 29 equations, 10 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Low-Cost MEMS IMU and Dead Reckoning Modeling
  4. Low-Cost MEMS IMU Error Model
  5. Kinematic Model of the Mobile Robot
  6. Invariant Extended Kalman Filter
  7. Spiking Neurons and Surrogate Gradient
  8. Proposed Method
  9. The Hybrid State Estimation Strategy for IMU Dead Reckoning
  10. InEKF with Pseudo-measurements
  11. Proposed SNN Architecture
  12. Experimental Results and Analysis
  13. Dataset and Evaluation Metrics
  14. Experimental Setup
  15. Experimental Results on the KITTI Dataset
  16. ...and 2 more sections

Figures (10)

  • Figure 1: The structure illustration of the LIF neuron.
  • Figure 2: An overview of the proposed hybrid state estimation method for IMU dead reckoning.
  • Figure 3: The pipeline of SNN for IMU dead reckoning.
  • Figure 4: Test results on Seq.3 in the KITTI dataset.
  • Figure 5: The position results on Seq.3 in the KITTI dataset.
  • ...and 5 more figures