Publications

2026

1. CANCDO

   Fault-tolerant Composite Adaptive Neural Control with Safety Guarantees for Multirotors in UAM Applications

   M. Shahrajabian, and F. Saghafi^*

   Aerospace Science and Technology, 2026

   Abs Bib PDF

   The demand for autonomous eVTOL aircrafts is driven by the need for innovative solutions to address transportation challenges in urban environments. Developing autonomous systems requires safety guarantees and robust control systems capable of handling uncertainties, disturbances, and faults. This paper introduces an intelligent control framework for multirotor eVTOLs in Urban Air Mobility applications. The inner-loop attitude/altitude tracking employs a Composite Adaptive Neural Control with Disturbance Observer (CANCDO); it combines Lyapunov-based neural network adaptation to compensate for uncertainties with a disturbance observer that estimates the upper bound of disturbances while compensating for estimation errors. The framework simultaneously trains the neural network and disturbance observer online by using a composite error, which includes both tracking error and state estimation error from an adaptive observer. The outer-loop horizontal position control uses a PD controller enhanced with an ESO for precise trajectory tracking. To provide fault tolerance, the framework includes a two-stage Fault Detection and Diagnosis (FDD) scheme, which estimates actuator effectiveness coefficients in real time and pairs with dynamic control allocation to optimally distribute virtual control efforts during failures. Additionally, adaptive Control Barrier Functions are developed to enforce safety constraints on position and velocity by incorporating real-time uncertainty estimates from CANCDO, which ensures safe operation even when there are model discrepancies. The framework is validated on a high-fidelity multirotor air taxi simulation model, achieving a final Relative RMSE of 0.34% for trajectory tracking under simultaneous faults and severe wind disturbances.

   @article{2026_CANCDO,
     title = {Fault-tolerant Composite Adaptive Neural Control with Safety Guarantees for Multirotors in UAM Applications},
     author = {Shahrajabian, M. and Saghafi, F.},
     journal = {Aerospace Science and Technology},
     year = {2026},
     publisher = elsevier,
     dimensions = {true},
   }