The exponential growth of commercially available unmanned aerial systems (UAVs) provides a reliable, low-cost platform for mobile sensor deployment. Extensive work exists leveraging large numbers of these platforms for intelligence, surveillance, and reconnaissance (ISR) applications ranging from search and rescue to plume detection and tracking. The proposed Robotic Catch And Release Manipulation Architecture (CARMA) is designed to address flight time limitations of small UAVs, leveraging a robotic manipulation and computer vision to actively compensate for perturbations of the UAV in flight (caused by environmental conditions such as wind) and movements of the recharging station (cause by movement over rough terrain, sea state in maritime applications, etc.). CARMA leverages an industrial robotic manipulator to create a robust system capable of capturing multi-rotor UAVs in an agitated hover. Using a custom-designed end-effector incorporating a monocular camera system, CARMA employs a closed-loop control strategy informed by a relative position and orientation estimate of the UAV using an active marker approach. Results demonstrate UAV tracking accuracy sufficient for capture of a small, CrazyFlie 2.0 UAV. The architecture is deployed on the Universal Robots UR10 manipulator which is able to successfully track the Crazyflie for capture.

This content is only available via PDF.
You do not currently have access to this content.