Towards dexterous continuum manipulator for aerial robotics

Abstract

This research presents a comprehensive exploration of advanced aerial manipulators, focusing on enhancing their performance through innovative designs and control strategies. One of the aerial systems can transition between motion-coupled and motion-decoupled modes, enhancing performance in inspection tasks. Its self-locking gimbal system allows the manipulator to operate independently of the quadrotor's motion, the dynamic gravity compensation mechanism optimizes battery placement, and the number of teeth to counteract weight imbalance.

The Aerial Continuum Manipulator (AeCoM) is introduced as a novel system combining a quadrotor with a tendon-driven continuum robotic arm. The AeCoM's design enables precise end-effector pose estimation and robust closed-loop kinematic control, preventing tendon slackening. Extensive experimental testing validates the system's capabilities. The continuum manipulator plays a crucial role in developing an IMU-based kinematic model and a robust closed-loop controller, which excels in rapid shape-deformation control and tendon tension management.

The Aerial Elephant Trunk (AET), inspired by the elephant trunk, is another innovative aerial manipulator. Comprising a small-scale quadrotor and a triple-section tendon-driven continuum arm, AET employs EKF-based state estimation, piecewise constant curvature-based shape estimation, and minimum jerk-based motion planning. Experimental results demonstrate AET's adaptability to various environments and objects, making it suitable for challenging tasks like pipeline maintenance and high-altitude inspections. The system's design allows it to navigate through narrow spaces and manipulate a wide range of objects, from slender and deformable to irregular and heavy.

A tendon-driven continuum robotic manipulator with three distinct sections is designed, fabricated, and evaluated. A robust shape estimation technique, leveraging multiple IMUs and the Piecewise Constant Curvature (PCC) assumption, is developed. A visualization environment aids in understanding the robot's behavior, and experiments validate the system's bending range, tip velocity, workspace, and durability. The system's compact design outperforms existing continuum robots. Multiple IMUs provide detailed attitude information for each section's end disk, enabling precise configuration state identification through a comprehensive coordinate transformation scheme.

Finally, a comprehensive coordinate transformation scheme and numerical optimization strategy are employed for forward and inverse kinematic modeling of a tendon-driven continuum manipulator. An IMU-based closed-loop controller ensures high robustness and agility. Experimental validation confirms the accuracy and efficiency of the kinematic models and controller. This research pioneers a fully integrated kinematic control architecture for physical continuum robotic systems, addressing the limitations of previous approaches.

Overall, this work advances the field of aerial manipulators, addressing challenges in attitude estimation, mechanical design, and control, and paving the way for more versatile and capable aerial robotic systems. The integration of advanced filtering techniques, innovative mechanical designs, and robust control strategies results in systems that are not only more accurate and stable but also capable of performing a wider range of tasks in challenging environments. The research contributes to the development of aerial manipulators that can operate autonomously and efficiently, opening up new possibilities for applications in inspection, maintenance, and other critical tasks.