Precise near-ultrasonic tilt determination for smartphones

Modern smartphones incorporate numerous sensors that enable relative positioning utilizing the measurement data from the internal inertial measurement unit. Their motion can thus be derived from the magnetometer, gyroscope, and acceleration sensor data. The problem, however, is that the acceleration signals cannot be separated into vertical and horizontal directions during the movement. Therefore, the absolute tilt of the smartphone is often unknown. Nevertheless, this separation is required for many step-based distance determination methods to estimate the length of the detected steps and the acceleration signals measured between the step events for each directional component. Therefore, a near-ultrasound-based distance determination was realized, which applies the Frequency-Modulated Continuous-Wave Radar method (FMCW Radar) to determine the difference between the onboard microphone’s position of the smartphone and the ground (or the nearest reflective surface). These distances allow for a precise inclination determination of the smartphone relative to the ground. The tilt estimation was compared with a camera-based reference position determination. The smartphone’s angle relative to the ground could be determined with a median accuracy of 2.73°, mean accuracy of 3.37°, and 4.56° for the third quartile error.