NMPC-based Unified Posture Manipulation and Thrust Vectoring for Fault Recovery

2025-03-24

AI Summary

To address the safety risk of catastrophic failure during high-altitude flight of multi-rotor UAVs under single-actuator failure, this paper proposes a rapid fault-tolerant recovery method integrating attitude regulation and thrust vectoring control. We innovatively develop a high-fidelity–reduced-order dual-dynamics model that jointly balances control accuracy and computational efficiency, unified within a nonlinear model predictive control (NMPC) framework. Unlike conventional approaches relying solely on attitude adjustment, our method enables concurrent optimization of angular velocity, attitude, and thrust direction—achieving instantaneous attitude reorientation and stable hover reconstruction at the moment of actuator failure. Validation on the Simscape-based high-fidelity M4 robot simulation platform demonstrates millisecond-level response to single-actuator failure, real-time recovery of stable hover, and significantly enhanced fault tolerance and safety in high-altitude operational scenarios.

Technology Category

• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Intelligent Robots: State Estimation
• Multiagent Systems: Multiagent Systems under Uncertainty

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Multi-rotors face significant risks, as actuator failures at high altitudes can easily result in a crash and the robot's destruction. Therefore, rapid fault recovery in the event of an actuator failure is necessary for the fault-tolerant and safe operation of unmanned aerial robots. In this work, we present a fault recovery approach based on the unification of posture manipulation and thrust vectoring. The key contributions of this work are: 1) Derivation of two flight dynamics models (high-fidelity and reduced-order) that capture posture control and thrust vectoring. 2) Design of a controller based on Nonlinear Model Predictive Control (NMPC) and demonstration of fault recovery in simulation using a high-fidelity model of the Multi-Modal Mobility Morphobot (M4) in Simscape.

Problem

• Recovering multi-rotors from actuator failures mid-flight
• Unifying posture control and thrust vectoring for fault tolerance
• Designing NMPC-based controller for high-fidelity fault recovery

Innovation

• NMPC-based posture and thrust control
• High-fidelity and reduced-order flight models
• Fault recovery via unified manipulation