Deep Learning-based Inertial Odometry for Pedestrian Tracking using Attention Mechanism and Res2Net Module

TL;DR

This paper introduces a deep learning approach using Res2Net and attention modules for pedestrian inertial odometry, significantly reducing translation errors compared to traditional and existing deep learning methods.

Contribution

It presents a novel deep neural network architecture that effectively estimates velocity from inertial data, improving accuracy in pedestrian tracking.

Findings

Reduces absolute translation error by 76%-86% compared to traditional methods.

Improves upon the state-of-the-art deep learning method RoNIN by 6%-31.4%.

Validated on multiple public inertial odometry datasets.

Abstract

Pedestrian dead reckoning is a challenging task due to the low-cost inertial sensor error accumulation. Recent research has shown that deep learning methods can achieve impressive performance in handling this issue. In this letter, we propose inertial odometry using a deep learning-based velocity estimation method. The deep neural network based on Res2Net modules and two convolutional block attention modules is leveraged to restore the potential connection between the horizontal velocity vector and raw inertial data from a smartphone. Our network is trained using only fifty percent of the public inertial odometry dataset (RoNIN) data. Then, it is validated on the RoNIN testing dataset and another public inertial odometry dataset (OXIOD). Compared with the traditional step-length and heading system-based algorithm, our approach decreases the absolute translation error (ATE) by 76%-86%. In addition, compared with the state-of-the-art deep learning method (RoNIN), our method improves its ATE by 6%-31.4%.