Autonomous Quadrotor Landing Using Vision and Pursuit Guidance

Abstract

Autonomous landing is the most difficult maneuver in a quadrotor mission because of challenges like (i) external disturbances causing deviation from the desired trajectory resulting in a crash and (ii) inaccuracy of GPS based landing due to localization errors. In this paper, we use onboard vision to detect and estimate the landing pad coordinates accurately. We develop a novel closing velocity controller and integrate it with pure pursuit guidance law to achieve faster and accurate landing in 3D as compared to traditional vertical landing approaches. We test the efficacy of the proposed controller for stationary and moving landing pad through simulations and validate the controller through outdoor experiments for a stationary landing pad.