Purely Image-Based Dynamic Impedance Control of a Fully Actuated Aerial Vehicle

Existing works on the impedance control of aerial vehicles are generally based on the assumption of known translational velocity. However, velocity estimation can be unreliable in many cases. To solve this problem, we propose a concurrent velocity observation and impedance control framework to perform contact-based interaction tasks of aerial vehicles. This framework is cost-optimized since it only uses monocular vision and inertial measurement unit (IMU) measurements, without using global navigation satellite system (GNSS) modules, optical flow, and force sensors. Specifically, we propose a novel observer-based impedance model (VOIM) and a dynamic image-based visual servoing (DIBVS) controller to achieve the desired interaction behavior. Rigorous input-to-state stability analysis of the closed-loop system is subsequently conducted, proving the effectiveness of the method under unknown environment position and stiffness, along with the robustness to any positive estimate of the scale factor of visual features. Experiments demonstrate the performance of the proposed method by implementing an aerial drawing task and a point-contact impedance-based force regulation task. Furthermore, the robustness under vision system uncertainties is tested, highlighting the potential of application in unstructured environments.

Automated summary

Existing works on impedance control of aerial vehicles are based on known translational velocity, which can be unreliable. To address this issue, a concurrent velocity observation and impedance control framework is proposed. The framework uses cost-optimized monocular vision and IMU measurements instead of GNSS modules, optical flow, and force sensors. Experimental results demonstrate the effectiveness and robustness of the proposed method in unknown environments and under vision system uncertainties.