A tutorial on human activity recognition using body-worn inertial sensors

Abstract

The last 20 years have seen ever-increasing research activity in the field of human activity recognition. With activity recognition having considerably matured, so has the number of challenges in designing, implementing, and evaluating activity recognition systems. This tutorial aims to provide a comprehensive hands-on introduction for newcomers to the field of human activity recognition. It specifically focuses on activity recognition using on-body inertial sensors. We first discuss the key research challenges that human activity recognition shares with general pattern recognition and identify those challenges that are specific to human activity recognition. We then describe the concept of an Activity Recognition Chain (ARC) as a general-purpose framework for designing and evaluating activity recognition systems. We detail each component of the framework, provide references to related research, and introduce the best practice methods developed by the activity recognition research community. We conclude with the educational example problem of recognizing different hand gestures from inertial sensors attached to the upper and lower arm. We illustrate how each component of this framework can be implemented for this specific activity recognition problem and demonstrate how different implementations compare and how they impact overall recognition performance.

Human activity recognition for emergency first responders via body-worn inertial sensors

“…Human activity recognition based on inertial data has been a fruitful line of enquiry for more than a decade [3], [4]. Although much of the work during this time has been dedicated to recognising activities associated with daily living, especially for monitoring the elderly or other populations at risk, there have been attempts to extend HAR to more dynamic activities and environments.…”

A generic framework to support participatory surveillance through crowdsensing

“…The reason behind this decision lies in the fact that in this evacuation exercise, identification of human activities, e.g. walking or running, was the main goal and in this respect the accelerometer has been the sensor most widely used for this purpose in related research works [29].…”

Human-Robot Motion: An attention-based navigation approach

“…The interaction/activity spaces are broadly defined by the convex hull of the person at hand and the people or the objects he interacts with. In this framework, human detection and human activity recognition are central in the definition of said social spaces (which explains why activity recognition has grown into an important area of research especially in computer vision [6], [7]). Having characterized the social spaces corresponding to the current situation, the most common approach in HRM is then to define costmaps on such social spaces: the higher the cost, the less desirable it is for the robot to be at the corresponding position.…”

“…Each person in the environment has his/her corresponding intention and distraction vectors.Ĩ is typically the output of a human activity recognition module [6], [7]. As far asD is concerned, it is a function of the distraction of the different elements, other persons and objects, that surround the person considered.…”

“…It is henceforth assumed thatĨ j is constantly estimated thanks to a human activity recognition module [6], [7]. Examples of field-of-distractions and field-of-perceptions are presented in Section V-A.…”