Abstract:Purpose The pedestrian dead reckoning (PDR) based on smartphones has been widely applied in continuous indoor positioning. However, when the position of the mobile phone and the walking patterns of the pedestrian are mixed, traditional PDR tends to become confused and thus degrade performance. To address this issue, this paper aims to propose an improved PDR scheme by focusing on gait pattern recognition and the impact of short-period but negative transitions on tracking. Design/methodology/approach The overall solution uses the inertial sensor integrated within the phone for positioning. A binary classifier-based change point detection algorithm is used to identify the transition points in pedestrian gait. Additionally, to enhance the accuracy of gait recognition, this paper presents a combined CNN-attention-based bi-directional long short-term memory(ABiLSTM) model, integrating convolutional neural networks (CNN), bi-directional long short-term memory (Bi-LSTM) and an attention mechanism, to recognize the current gait pattern. The outcomes of this gait pattern recognition are then applied to PDR. Based on distinct gait patterns, corresponding PDR strategies are devised to enable continuous tracking and positioning of pedestrians. Findings Through experimental verification, the CNN-ABiLSTM model achieves a gait recognition accuracy of 99.52% on the self-constructed data set. The pedestrian navigation estimation method proposed in this paper, which is based on gait recognition assistance, demonstrates a 32.56% improvement in accuracy over traditional positioning algorithms in multi-gait scenarios. Originality/value The improved PDR scheme algorithm significantly enhances the robustness and smoothness of pedestrian tracking, particularly during multiple gait transitions. This, in turn, provides strong support for the utilization of low-cost inertial sensors integrated within mobile phones for indoor positioning applications.

DePedo: Anti Periodic Negative-Step Movement Pedometer with Deep Convolutional Neural Networks

Estimating Walking Distance Based on Single Accelerometer

Early Detection of Parkinson's Disease Using Deep NeuroEnhanceNet With Smartphone Walking Recordings

Self-Adaptive Step Counting on Smartphones under Unrestricted Stepping Modes.

Novel Step Detection Algorithm for Smartphone Indoor Localization Based on CEEMDAN-HT

Smartphone-Based Unconstrained Step Detection Fusing a Variable Sliding Window and an Adaptive Threshold

Flexible, Wireless, Multifunctional Integrated Electronic System for Daily Wearable Gait Detection

Robust and Accurate Step Counting Based on Motion Mode Recognition for Pedestrian Indoor Positioning Using a Smartphone

A Novel Robust Step Detection Algorithm for Foot-Mounted IMU

DeepWalking: Enabling Smartphone-based Walking Speed Estimation Using Deep Learning

PDMove

Simple and efficient step detection algorithm for foot-mounted IMU

Multi-Activity Step Counting Algorithm Using Deep Learning Foot Flat Detection with an IMU Inside the Sole of a Shoe

An improved pedestrian dead reckoning method based on gait pattern recognition

Noise segmentation for step detection and distance estimation using smartphone sensor data

A Privacy-Protecting Step-Level Walking Direction Detection Algorithm based on Floor Vibration

In-home Gait Abnormality Detection through Footstep-Induced Floor Vibration Sensing and Person-Invariant Contrastive Learning

A step length estimation method based on frequency domain feature analysis and gait recognition for pedestrian dead reckoning

Accurate Step Count with Generalized and Personalized Deep Learning on Accelerometer Data

An Interpretable Deep Learning Optimized Wearable Daily Detection System for Parkinson's Disease

Static-Dynamic Temporal Networks for Parkinson's Disease Detection and Severity Prediction