Abstract

Accelerometers are commonly used, yet the process of calibrating them and the influence this has on recorded accelerations is rarely reported. The aim of this study was to compare the accuracy of three simple gravity based calibration methods of accelerometers. Using a custom made rig, 16 Delsys Trigno sensors were simultaneously calibrated by positioning the sensors in 9 positions throughout the full range of orientations across three axes. Three calibration methods were used spanning a range of 1g (1G), 2g (2G), and 2g with optimisation (2Gopt). Errors were greatest in 1G (RMSD=3.1%) and equally as good for the 2G and 2Gopt (2.1%). Gravity based calibration of accelerometers can be achieved quickly, and calibration over a larger range provides more accurate results. This work provides recommendations of accelerometer use which help the applied practitioner to collect more reliable and valid data. Further investigation of factors, including those affecting the frequency of calibration, is required.