NDOB-Based Control of a UAV with Delta-Arm Considering Manipulator Dynamics

TL;DR

The paper tackles the challenge of achieving precise, high-speed end-effector motions in aerial manipulators by integrating a nonlinear disturbance observer (NDOB) for low-frequency disturbances with a high-frequency compensation path based on end-effector dynamics processed through a high-pass filter. This composite control framework is embedded in a fully autonomous delta-arm AM system and validated via real-world experiments, achieving millimeter-level end-effector accuracy and robust quadrotor stability even with payloads up to 400 g. Key findings include substantial disturbance rejection gains over PX4 baselines, improved trajectory tracking under wind, and rapid stabilization of the end-effector while the UAV executes independent maneuvers, demonstrating practical viability for high-speed aerial manipulation. The approach advances practical AM performance in windy environments and under fast manipulator motions, with potential for broader deployment in 3D printing, inspection, and grasping tasks.

Abstract

Aerial Manipulators (AMs) provide a versatile platform for various applications, including 3D printing, architecture, and aerial grasping missions. However, their operational speed is often sacrificed to uphold precision. Existing control strategies for AMs often regard the manipulator as a disturbance and employ robust control methods to mitigate its influence. This research focuses on elevating the precision of the end-effector and enhancing the agility of aerial manipulator movements. We present a composite control scheme to address these challenges. Initially, a Nonlinear Disturbance Observer (NDOB) is utilized to compensate for internal coupling effects and external disturbances. Subsequently, manipulator dynamics are processed through a high pass filter to facilitate agile movements. By integrating the proposed control method into a fully autonomous delta-arm-based AM system, we substantiate the controller's efficacy through extensive real-world experiments. The outcomes illustrate that the end-effector can achieve accuracy at the millimeter level.

Figures (6)

• Figure 1: (a) The UAV performs a circular maneuver while keeping the end-effector stabilized at the origin. The dotted line represents the UAV trajectory, and the position error of the end-effector is only 6 mm. (b) The Delta-arm weighs 103.4 g. (c) The aerial manipulator platform, designed for various tasks, weighs 1382.2 g.
• Figure 2: Schematics of the delta parallel robot and the coordinate frames of the aerial manipulator (AM) system.
• Figure 3: Flowchart of our control framework: The desired quadrotor and delta-arm trajectory passes through three blocks in sequence to obtain the servo motor desired agile velocity $\omega_d^e$, the AM desired thrust $T_d$, and the AM desired angular velocity $\omega_d$. (a) delta-arm control: process the delta-arm desired trajectory $p^e(t)$ to obtain the servo desired angular velocity. (b) High-Frequency Disturbance Estimator: The end-effector dynamics pass through a high pass filter to obtain $f^{high}$. (c) Low-Frequency Disturbance Estimator: The actual state $\partial$ of the quadrotor passes through an NDOB to estimate $f^{low}$.
• Figure 4: Snapshot of disturbance rejection. The end-effector’s trajectory is depicted as a dotted line. In this test, the end-effector performed agile movements, carrying a 400 g payload, while the quadrotor stablize to the origin to prove its robustness. With our control framework, it will have 80% performance improvement compared to the PX4.
• Figure 5: (a) shows the end-effector and quadrotor position error. (b) shows the end-effector and quadrotor tracking trajectory in the X-Y plane.