Visual-inertial Odometry on Resource-constrained Systems
Author | : Mingyang Li |
Publisher | : |
Total Pages | : 207 |
Release | : 2014 |
ISBN-10 | : 1321551126 |
ISBN-13 | : 9781321551129 |
Rating | : 4/5 (26 Downloads) |
Book excerpt: In this work, we focus on the problem of pose estimation in unknown environments, using the measurements from an inertial measurement unit (IMU) and a single camera. We term this estimation task visual-inertial odometry (VIO), in analogy to the well-known visual-odometry (VO) problem. Our focus is on developing VIO algorithms for high-precision, consistent motion estimation in real time. The majority of VIO algorithms proposed to date have been developed for systems which are equipped with high-end processors and high-quality sensors. By contrast, we are interested in tracking the motion of systems that are small and inexpensive, and are equipped with limited processing and sensing resources. Such resource-constrained systems are common in application areas such as micro aerial vehicles, mobile phones, and augmented reality (AR) devices. Endowing such systems with the capability to accurately track their poses will create a host of new opportunities for commercial applications, and lower the barrier to entry in robotics research and development.