Research On Pedestrian Dead Reckoning With Unconstrained Smartphones

In recent years, indoor positioning plays a more and more important role in many fields. Meanwhile, as the prompt development of micro-electro-mechanical systems (MEMS), smartphones are equipped with various inertial sensors, including accelerometers, gyroscopes, magnetometers and etc. Thus, the pedestrian dead reckoning (PDR) technique based on smartphones has become a hot research topic in the field of indoor positioning. To realize accurate PDR, a primary challenge in PDR studies and applications is that some restrictions must be imposed on smartphone placements. However, postures and placements of smartphones during pedestrians’ walking are arbitrary in reality. To address this challenge, the paper studies the following three aspects.Firstly, a six-order Butterworth low-pass filter is utilized to process raw signals from the accelerometer, gyroscope and magnetometer of a smartphone. These processed signals are fused to infer attitudes of the smartphone, and then accelerations and angular velocities in the smartphone reference frame are transformed to the counterpart in the Earth reference frame by formulating a rotation matrix.Secondly, to accurately detect step counts with an unconstrained smartphone, the vertical component of accelerations in the Earth reference frame is utilized to perform peak detection and zero crossing. The experimental result shows that the proposed method can achieve the successful rate of 94.38%when the smartphone is totally unconstrained.Finally, to estimate the pedestrian heading, a novel pedestrian heading estimation algorithm based on a least-squares method is presented by using the horizontal component of angular velocities in the Earth reference frame. The experimental result shows that the proposed method can achieve the mean absolute error of 5.1 degrees when the smartphone is carried in the swinging hands and 7.1 degrees when it is carried in the front trousers’pocket both in indoor environments.In conclusion, the proposed method turns out to be more effective and more robust than the existing PDR algorithms, and will contribute to the popularity of PDR in practice.