Adaptive drift calibration of accelerometers with direct velocity measurements

The accelerometer has become one of the most popular sensors in recent years due to its low cost and the widespread availability of smart phones that now contain three axis accelerometers. This paper proposes an adaptive drift calibration technique for accelerometer signals, correcting higher sampling rate accelerometers using lower sampling rate velocity and position measurements. Specifically, this study made use of 40Hz sampled accelerometer signals captured by smart phones, and corrected them using two different 1Hz sampled velocity reference signals: a vehicle speed sensor and velocity from a Global Positioning System position sensor. The paper compares the performance of two error correction algorithms based on step and ramp shaped error correction delta functions. The ramp function was found to be susceptible to oscillation caused by high frequency noise, while the step function remains stable. The paper also shows that the GPS velocity signal has better performance over the dashboard vehicle velocity signal due to the higher frequency noise within the direct velocity signal.