Five Degrees-of-freedom Pose-estimation Method for Smartphones using a Single Acoustic Anchor

In this paper, we describe a novel method for the five-degrees-of-freedom (5-DoF) pose estimation of a smartphone using the acoustic signal from only one external speaker. To obtain the smartphone’s 5-DoF pose using this speaker, we apply the concept of “mirror-image” speakers to effectively obtain a multiplicity of signals. Here, the primary reflected signals from the wall or floor are captured as signals transmitted from “virtual speakers” located at (wall or floor) mirror-image positions with respect to the real speaker’s location. By using the smartphone built-in stereo microphones, we can obtain not only its three-dimensional (3-D) position but also its yaw and pitch angles, giving a 5-DoF pose estimation. Holding the smartphone sideways, we can calculate arrival angles for each speaker using the two microphones, and identify which (real or mirror-image) speaker is the signal source. We can then robustly identify the 5-DoF pose by utilizing the information obtained from arrival angles and an acceleration-based pitch angle. For our experiments, we located the speaker and estimated the smartphone’s 5-DoF pose at several places along a 2-m wide corridor. Its measurement range was $2.0\times 2.0\times 1.9\,\,\text{m}^{3}$ . The results within the effective operating range gave 90th-percentile errors of less than 114.35 mm for the 3-D position, less than 7.96° for the yaw angle, and 7.09° for the pitch angle. We also clarified the effectiveness of the proposed method which can extend speakers’ baselines beyond the physical limit of the corridor via dilution-of-precision measures. Moreover, we identified limitations of the proposed method using simulations.