Accurately Counting Steps of the Pedestrian with Varying Walking Speeds
Authors:
- Xiaomin KangXKXiaomin Kang
- Baoqi HuangBHBaoqi Huang
- Runze YangRYRunze Yang
- Guodong QiGQGuodong Qi
- 1 October 2018
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 679-686
Abstract
With the popularity and development of smartphones, mobile services and applications have gained much attention from both academia and industries. Since walking is one of the most ordinary daily activities and contributes to human health, grasping the knowledge of walking steps helps to evaluate the corresponding energy consumption so as to give advice for future. Besides, it also assists in many other applications, such as pedestrian localization and navigation, behavior recognition, and etc. In this paper, by taking into account the fact that the pedestrian walking speeds suffer from dynamics due to individual emotions, walking habits, or environmental influences, two adaptive and efficient algorithms are proposed to accurately count pedestrian steps by only using gyroscope. Specifically, the realtime walking frequency is estimated based on the spectrum of gyroscope measurements obtained through fast Fourier transform (FFT) for the purpose of step detection, and meanwhile, the step count can be further inferred by making use of the realtime frequency. Extensive experiments are conducted by involving 12 participants with typically different walking speeds and 2 smartphone placements. It is shown that the average accuracies of the proposed algorithms are as high as 97.24% and the standard deviations are as low as 2.18 with respect to all considered scenarios and participants, which are significantly superior to several well-known counterparts and commercial products in terms of both accuracy and stability.Keywords
This publication has 30 references indexed in Scilit:
- Recognizing Human Activities User-independently on Smartphones Based on Accelerometer DataInternational Journal of Interactive Multimedia and Artificial Intelligence, 2012
- Accelerometry-Based Classification of Human Activities Using Markov ModelingComputational Intelligence and Neuroscience, 2011
- Activity identification using body-mounted sensors—a review of classification techniquesPhysiological Measurement, 2009
- Accelerometry: A technique for quantifying movement patterns during walkingGait & Posture, 2008
- Foot mounted inertial system for pedestrian navigationMeasurement Science and Technology, 2008
- A hidden Markov model-based stride segmentation technique applied to equine inertial sensor trunk movement dataJournal of Biomechanics, 2007
- Automatic Step Detection in the Accelerometer Signal2007
- Health benefits of physical activityCanadian Medical Association Journal, 2006
- Feature Selection and Activity Recognition from Wearable Sensors2006
- How Many Steps/Day Are Enough?Sports Medicine, 2004