High-frequency MAV state estimation using low-cost inertial and optical flow measurement units

Abstract

This paper develops a new method for 3D, high rate vehicle state estimation, specially designed for free-flying Micro Aerial Vehicles (MAVs). We fuse observations from inertial and optical flow low-cost measurement units, and extend the current use of this optical sensors from hovering purposes to odometry estimation. Two Kalman filters, with its extended and error-state versions, are developed, and benchmarked alongside a large number of algorithm variations, using both simulations and real experiments with precise ground-truth. In contrast to state-of-the-art visual-inertial odometry methods, the proposed solution does not require image processing in the main CPU. Instead, the data correction is done taking advantage of the recently appeared optical flow sensors, which directly provide metric information about the MAV motion. We hence reduce the computational load of the main processor unit, and obtain an accurate estimation of the vehicle state at a high update rate.