Full-Pose Tracking via Robust Control for Over-Actuated Multirotors (2506.16427v1)

Abstract: This paper presents a robust cascaded control architecture for over-actuated multirotors. It extends the Incremental Nonlinear Dynamic Inversion (INDI) control combined with structured H_inf control, initially proposed for under-actuated multirotors, to a broader range of multirotor configurations. To achieve precise and robust attitude and position tracking, we employ a weighted least-squares geometric guidance control allocation method, formulated as a quadratic optimization problem, enabling full-pose tracking. The proposed approach effectively addresses key challenges, such as preventing infeasible pose references and enhancing robustness against disturbances, as well as considering multirotor's actual physical limitations. Numerical simulations with an over-actuated hexacopter validate the method's effectiveness, demonstrating its adaptability to diverse mission scenarios and its potential for real-world aerial applications.

• The paper presents a cascaded control architecture that combines INDI and H∞ control for robust full-pose tracking in over-actuated UAV systems.
• It employs a weighted least-squares quadratic programming approach to allocate actuator commands while managing physical constraints and disturbances.
• High-fidelity simulations demonstrate tracking errors below 2 cm and control cycles under 0.25 ms at 500 Hz, confirming computational efficiency and resilience.

Robust Full-Pose Tracking Control for Over-Actuated Multirotors

This paper addresses the challenge of robust, precise full-pose tracking for over-actuated (OA) multirotor UAVs by extending the Incremental Nonlinear Dynamic Inversion (INDI) and structured $H_\infty$ control architecture beyond its traditional application in under-actuated systems. The work contributes both theoretically and practically to the control of aerial vehicles with redundant actuation, providing a high-frequency, robust solution that explicitly accounts for physical and operational constraints.

Architecture and Formulation

The central technical contribution is a cascaded control architecture that integrates INDI for model-agnostic disturbance rejection and a structured $H_\infty$ approach for enhanced robustness. The system is partitioned into two control loops:

• Outer (Guidance) Loop: Addresses translational dynamics—implementing INDI/$H_\infty$ control followed by a weighted least-squares (WLS) geometric control allocation formulated as a quadratic programming (QP) problem.
• Inner (Stabilization) Loop: Manages rotational dynamics—using INDI and $H_\infty$ control for attitude tracking, ensuring rapid and robust stabilization.

The allocation of actuator commands from the virtual control inputs considers the full geometric distribution and physical limits of the actuators, addressing the inherent redundancy and constraints of OA configurations. The control allocation layer ensures that the tracking commands not only aim for the desired pose but are guaranteed to be feasible given actuator saturation, force/torque limits, and the attainable force set (AFS).

Modeling and Feasibility Analysis

A detailed modeling framework is established for general OA multirotor systems, encompassing $n$ motors with fully three-dimensional orientation capabilities. The allocation matrix is formulated to map force/torque demands to individual actuator commands, in both nominal and linearized forms, facilitating efficient high-rate implementation.

Crucially, the concept of the attainable and feasible force sets (AFS and S) is incorporated not as simplifications but as active constraints in the control allocation process. This ensures:

• Only physically feasible commands are issued.
• Priority can be dynamically assigned to either position or attitude tracking in situations where both cannot be simultaneously achieved, e.g., during actuator saturation or external disturbances.
• Optimization-based allocation (WLS QP) efficiently navigates the null space arising from actuation redundancy to enforce hard input and dynamic constraints.

Numerical Validation and Results

The proposed architecture is validated via high-fidelity simulations using a 6-DOF OA hexacopter model with accurately parameterized inertial and actuator limits. Three comprehensive scenarios are evaluated:

1. Hover at Varying Orientation: Demonstrates precise position holding with attitude tracking to the limits of physical feasibility. When reference attitudes exceed actuator limits, the system automatically computes and tracks the nearest feasible attitude, prioritizing position as per the specified weights.
2. Translation with Minimal Orientation: The system maintains minimal orientation deviation (<0.6° attitude error) while accurately tracking aggressive position references, contingent on correct tuning of the WLS cost function weights.
3. Full-Pose Tracking Under Disturbances: Robustness to external acceleration disturbances is empirically validated. The system compensates for disturbances mainly through attitude adjustments, with position tracking remaining highly accurate throughout. Computational efficiency is evidenced by guidance loop optimization consistently requiring less than 0.25 ms per control cycle at 500 Hz.

Quantitative Performance Highlights

• Position tracking errors: Maintained below 2 cm in all tested scenarios.
• Attitude tracking errors: Confined within actuator/saturation limits, with automatic feasible pose computation when references are unattainable.
• Control allocation solve time: Maximum 0.25 ms, supporting >500 Hz control rates even on standard PC hardware.
• The architecture displays significant resilience to disturbances and actuator saturation, confirmed by consistent tracking performance at control boundaries.

Implications and Future Directions

This work provides a practical, computationally efficient framework for robust full-pose tracking in OA multirotors, directly addressing critical challenges in UAV reliability, actuator fault tolerance, disturbance rejection, and operational safety. By integrating INDI and $H_\infty$ in both position and attitude loops, and solving the guidance allocation with WLS-QP informed by real-time attainable force set estimation, the proposed method is well-suited to embedded real-time applications such as industrial inspection, aerial manipulation, and flight in confined or dynamic environments.

Key implications and anticipated developments include:

• Onboard Implementation: The architecture is amenable to deployment on UAV flight controllers (e.g., Paparazzi), with ongoing work to experimentally validate performance.
• Actuator Failure Management: The explicit modeling of redundancy and AFS enables real-time recomputation of feasible commands under actuator failure scenarios, supporting fault-tolerant operation.
• Parameter Tuning: The ability to weight position versus attitude tracking in the cost function allows adaptation to mission-specific requirements but necessitates careful gain and weight tuning to balance tracking priorities, especially during aggressive or constrained maneuvers.
• Extensions: Future research may focus on improving inner-loop actuator allocation (to further avoid saturation), adaptive or learning-based weighting strategies, and broader experimental benchmarking across diverse OA/FA multirotor platforms.

Concluding Perspective

By integrating contemporary robust control concepts and explicit optimization-based allocation in a manner fully cognizant of the unique structure and redundancy of OA multirotors, this work both advances the theoretical foundation for UAV control and delivers concrete tools for real-world aerial robotics applications. The approach's demonstrated computational efficiency and disturbance resilience position it as a promising candidate for next-generation safe and reliable multirotor UAV systems.