Dealing with the Effects of Sensor Displacement in Wearable Activity Recognition

Abstract

Most wearable activity recognition systems assume a predefined sensor deployment that remains unchanged during runtime. However, this assumption does not reflect real-life conditions. During the normal use of such systems, users may place the sensors in a position different from the predefined sensor placement. Also, sensors may move from their original location to a different one, due to a loose attachment. Activity recognition systems trained on activity patterns characteristic of a given sensor deployment may likely fail due to sensor displacements. In this work, we innovatively explore the effects of sensor displacement induced by both the intentional misplacement of sensors and self-placement by the user. The effects of sensor displacement are analyzed for standard activity recognition techniques, as well as for an alternate robust sensor fusion method proposed in a previous work. While classical recognition models show little tolerance to sensor displacement, the proposed method is proven to have notable capabilities to assimilate the changes introduced in the sensor position due to self-placement and provides considerable improvements for large misplacements.

Feature extraction for robust physical activity recognition

“…This system includes 9 inertial measurement units generating and every unit contains four sensors, accelerometer (ACC), gyroscope (GYR), magnetometer (MAG) and quaternion (QUAD) sensor sensor, which generate 13 inertial signals: in total, 117 inertial signals are processed. For more details, the reader can refer to [24].…”

“…In this work, the HAR system has been mainly trained and tested using the REALDISP Activity Recognition dataset, available at the UCI Machine Learning Repository [24]. This dataset includes recordings from 17 subjects, seven females and ten males, with ages ranging from 22 to 37 years old.…”

“…The inertial signals are recorded by 9 inertial measurement units. Every unit contains four sensors, accelerometer (ACC), gyroscope (GYR), magnetometer (MAG) and quaternion (QUAD) [24]). The system architecture includes two steps: feature extraction and activity classification.…”

“…This dataset has permitted several HAR works focused on different aspects. One interesting aspect has been the degradation suffered on the HAR accuracy depending on the sensor placement or the number of displacements (wrong placements) [23][24][25]. Other analyzed aspects have been the window size [24,26], and the detection of activities transitions [27].…”

Segmentation Study of Aged Gait Based on FFT

“…The motion study based on the video is the mainstream and the relevant technologies are relatively matured, but the amount of data is so large that it is necessary to reduce the dimensions of the large data matrix [3][4][5][6]. This paper adopts the wearable sensor to gather the human motion data sequence.…”