AiRite: Towards Accurate Infrastructure-Free 3-D Tracking of Smart Devices

摘要

Precise tracking of smart wearables in 3-D space is foundational in domains ranging from Augmented Reality (AR) and gesture recognition to robotics and indoor localization. Usually, a sensor-dependent approximation model, and a complementary infrastructure like acoustic, vision etc. is implemented alongside an onboard IMU (Inertial Measurement Unit) for an effective tracking. In this work, we present a novel 3-D tracking solution, ‘AiRite’ for commercial-grade smart wearables/mobiles using only their onboard IMU. Our tracking method mitigates the manifested inertial errors using a novel progressive zero correction, yielding superior results both for 2-D and 3-D trajectories. AiRite aims for a pervasive usage and is independent of device's form factor, or make of sensor enabling it to be used right out-of-the-box. The solution is shown to be more accurate than prior arts in tracking trajectories of basic shapes by moving the hand both in 2-D and 3-D. We depict trajectories of medium to long duration words written in air in a cursive handwriting using both smartwatch and smartphone. We further demonstrate robustness of AiRite by depicting panoramic 3-D texts written in air in a concave shape and visualizing the same from multiple angles, which are first results of their kind in this domain.