Visual servoing of a quadrotor with suspended slung load for object detection and tracking

Abstract

This paper presents an autonomous vision-based controller for a quadrotor with a suspended slung load capable of positioning the platform and the load above a ground target. A model of the coupled system (quadrotor and load) is first developed, before augmenting the model with the image kinematics for a four point coplanar target object. The resulting image-based control framework considers two different visual servoing methods: Image-based (IBVS) and Position-based visual servoing (PBVS). The control law of the IBVS is developed using the direct feedback from image plane measurements, whereas the PBVS requires additional estimation of the target object pose (position and orientation). Simulated and experimental results are presented using 3 different control configurations. Using the hybrid configuration that combines both IBVS & PBVS methods or a solely PBVS configuration demonstrated robustness properties, whereas using the IBVS configuration showed unsatisfactory results and higher sensitivity to disturbances and control parameters. The IBVS & PBVS control configuration results in better performance in the path following task, and lower RMS error between the vehicle and target position compared to the other two configurations.