Pedestrian activity detection in a multi-floor environment by a smart phone

Citations Over Time

Abstract

Indoor localization has attracted considerable attention recently. One approach is to use inertial sensors mounted on pedestrians to characterize the users' motions. However, few studies have focused on a multi floor environment where users' activities may include walking, running, and going up/down stairs. This paper proposes a lightweight activity detection system using inertial sensors on a smart phone to detect the behaviors of a pedestrian in the multi-floor indoor environment. The system first identifies strides using the accelerations values. It then uses the displacement, duration, and acceleration to classify their types. Our experimental results show that the stride detection accuracy is about 99%. In addition, the types of strides, namely walking, running, going upstairs, and going downstairs, can be detected with the accuracy of 94%, 91%, 95%, and 92%, respectively.

Related Papers

• → Pedestrian Stride Length Estimation from IMU Measurements and ANN Based Algorithm(2017)75 cited
• → Accurate Stride-Length Estimation Based on LT-StrideNet for Pedestrian Dead Reckoning Using a Shank-Mounted Sensor(2023)6 cited
• → Pedestrian Stride-Length Estimation Based on Bidirectional LSTM Network(2020)8 cited
• → ANN-based stride detection using smartphones for Pedestrian dead reckoning(2018)8 cited
• → A Stride Detection Algorithm Based on Triaxial Acceleration Characteristics of Pedestrians(2023)