The tandem dual-rotor aerial-aquatic vehicle (AAV) of Wu et al. (Drones, 2024) demonstrated cross-media locomotion - hovering flight, water entry and exit, tight surface turns, and traversal of narrow underwater gaps - but only under remote control, with a cascaded PID controller validated in simulation and a simplified depth-interpolation model for the air-water transition. The paper's conclusions pose an open agenda: optimize the structure (grounded in the paper's measured propeller-occlusion thrust loss), develop control beyond cascaded PID for wind robustness in air and for underwater attitude under hydrodynamic loads, and equip the vehicle for underwater obstacle avoidance and autonomous navigation after water entry, 'to achieve the autonomous navigation of the tandem dual-rotor AAV once it enters the water'. Follow-up work by the same group has delivered advanced controllers mostly in simulation (active disturbance rejection control, super-twisting sliding mode with an extended state observer, line-of-sight guidance under ocean currents) and CFD water-exit studies, while a 2026 field review still identifies insufficient autonomy in complex environments as a key open challenge. The surviving open core is demonstrated onboard autonomy for this vehicle class: pilot-free underwater obstacle avoidance and navigation after water entry, carried by control that is robust to wind in air, to hydrodynamic loads underwater, and to the transition hydrodynamics that the paper's Assumption 3 and its depth-interpolation model leave unmodeled.

Background

Aerial-aquatic vehicles (AAVs) are unmanned systems built to fly in air, operate underwater, and cross the air-water interface repeatedly, targeting marine missions such as structure, under-keel, and confined-water inspection that neither drones nor autonomous underwater vehicles (AUVs) can perform alone. Because water is roughly a thousand times denser than air, one platform must reconcile contradictory demands - high rotor speed and light structure for flight versus low drag, sealing, and low propeller speed in water - so most AAV prototypes (floodable fixed-wing bodies, folding rotor arms, multirotors with added underwater propulsors) compromise one medium or the other.

The source paper (Wu et al., 2024) proposed a tandem dual-rotor AAV: a slender, streamlined fuselage of the kind used on AUVs, carrying two aerial rotors in tandem (fore and aft, non-overlapping) for flight, two lateral roll thrusters for roll balance and underwater hovering, and a steerable tail-vector thruster for agile low-speed underwater maneuvering. Its dynamics are written in the standard underwater-vehicle rigid-body form, including added mass (the effective increase in inertia caused by fluid entrained with an accelerating body), hydrodynamic damping, and buoyancy. The aerial and underwater regimes are stitched together by a depth-dependent interpolation operator (⋅)Ξ(\cdot)^{\Xi} (Eq. (15) of the paper): each hydrodynamic coefficient takes its fully-submerged value below the body, ramps linearly to zero across a band of one body height hbh_b at the surface, and vanishes in air. Three simplifying assumptions are adopted, notably Assumption 3 - the hydrodynamics of complex-geometry appendages are neglected except for the main body - and the paper itself notes that the error this introduces 'poses requirements for the robustness of the controller'.

The 4 kg prototype was demonstrated outdoors: hovering flight; water entry and exit in a one-meter-deep pool and a full flight-diving-resurfacing sequence in a small lake; surface 'S' turns with a measured minimum turning radius of 0.5 m (tail-thruster tilt between -36 and +36 degrees, roll held between about -7 and +5 degrees); and underwater traversal of a 0.3 m gap and a 0.5 m-diameter ring with pitch and roll held within about 10 degrees. Crucially, all underwater maneuvers were flown under remote control. Attitude and depth regulation used a cascaded PID controller (a standard two-loop structure in which an outer position/velocity loop commands an inner attitude loop with proportional-integral-derivative gains), validated in simulation along with an underwater 'S'-trajectory tracking study that exposed noticeable lateral error, attributed to underactuation - underwater the vehicle has only five independently controllable states - and to fluid resistance. Two further measured limitations matter: in hover the fuselage partially occludes the rotor discs (static thrust about 95 N unobstructed versus about 58.8 N at 50% motor duty in flight), and on water exit the prototype showed pitch excursions up to 20 degrees from surface effects.

The surrounding field has matured unevenly. Underwater obstacle avoidance is well developed for dedicated AUVs (forward-looking sonar and learning-based planners), sampling-based trajectory planning with collision avoidance exists for hybrid aerial-underwater vehicles in simulation, and the morphable quadrotor TJ-FlyingFish has demonstrated SLAM (simultaneous localization and mapping - onboard estimation of position and a map without GPS) across the air-water boundary, supporting autonomous operations in real flight experiments. Yet a 2026 review of aerial-aquatic unmanned amphibious vehicles still names 'insufficient autonomy in complex environments', alongside the lack of unified cross-medium modeling frameworks, among the field's key open challenges.

Problem Statement

The problem, posed by the conclusions of Wu et al. (Drones, 2024), is to carry the tandem dual-rotor AAV from a remotely piloted proof of concept to an autonomous system - in the paper's words, 'to achieve the autonomous navigation of the tandem dual-rotor AAV once it enters the water'. The common objective uniting all requested results is pilot-free execution of a cross-media mission. The paper's conclusions name the following jointly supporting elements:

(1) Control beyond cascaded PID: a control architecture that maintains aerial stability under wind disturbances and underwater attitude control under hydrodynamic loads, including the water-exit phase where the prototype exhibited pitch excursions up to 20 degrees, and that is robust to the unmodeled transition hydrodynamics - the appendage hydrodynamics neglected under Assumption 3 and the (⋅)Ξ(\cdot)^{\Xi} depth-interpolation model of Eq. (15) - whose error, the paper notes, 'poses requirements for the robustness of the controller'.

(2) Underwater autonomy after water entry: an onboard perception, guidance, and obstacle-avoidance capability - a sensor suite and path planning suited to constrained underwater environments - enabling the vehicle, after entering the water, to navigate autonomously without pilot input; in the source, all gap and ring traversals and all underwater turns were remotely controlled.

(3) Structural refinement: airframe and propulsion-layout improvements that address the measured propeller-occlusion thrust loss (about 95 N unobstructed versus about 58.8 N at 50% duty in hover) and further reduce underwater drag - the hardware substrate on which (1) and (2) must run.

These elements are interdependent: (1) is the low-level control layer beneath (2), and (3) defines the vehicle on which both are deployed. Element (2) is the mandatory deliverable, and (1) is mandatory insofar as it must carry (2) through the wind, water-exit, and underwater hydrodynamic regimes; (3) is an enabling refinement rather than a separate deliverable, required only to the extent that the vehicle under study must be capable of the mission.

The problem is complete when a control and autonomy architecture for the tandem dual-rotor AAV, or a directly derived configuration, is validated to execute without pilot commands a mission spanning flight, water entry, autonomous underwater navigation with obstacle avoidance through a constrained underwater course of the kind demonstrated in the source (a sub-meter longitudinal gap of about 0.3 m and a ring of about 0.5 m diameter, traversed at controlled depth with pitch and roll held within about 10 degrees), and recovery. The required evidence is this validated architecture - analysis and high-fidelity simulation of the full mission, together with hardware-in-the-loop or water-tank validation where physical hardware is proposed - and such in-medium artifacts do resolve the task; no on-vehicle field demonstration is required.

References

  1. S. Wu, M. Shao, S. Wu, Z. He, H. Wang, J. Zhang, Y. You (2024). 'Design and Demonstration of a Tandem Dual-Rotor Aerial-Aquatic Vehicle.' Drones 8(3):100. DOI: 10.3390/drones8030100. https://doi.org/10.3390/drones8030100 (primary source)
  2. S. Wu, Z. He, L. Wang, M. Shao, S. Wu, J. Zhang, Y. Liu (2025, online first September 2025; issue dated 2026). 'Nonsingular terminal super-twisting sliding mode control with extended state observer for multi-domain motion control of tandem twin-rotor Aerial-aquatic vehicles.' Defence Technology. DOI: 10.1016/j.dt.2025.09.010. https://doi.org/10.1016/j.dt.2025.09.010
  3. M. Shao, S. Wu, L. Wang, S. Wu, H. Wang, Z. He, M. Lin, J. Zhang (2025). 'Adaptive wind direction strategy for neuro-network-based active disturbance rejection control of tandem twin-rotor aerial-aquatic vehicle.' Aerospace Science and Technology. DOI: 10.1016/j.ast.2025.110089. https://doi.org/10.1016/j.ast.2025.110089
  4. S. Wu, M. Shao, S. Wu, Z. He, H. Wang, J. Zhang, Y. Liu (2025). 'A high maneuvering motion strategy and stable control method for tandem twin-rotor aerial-aquatic vehicles near the water surface.' Defence Technology. DOI: 10.1016/j.dt.2024.09.009. https://doi.org/10.1016/j.dt.2024.09.009
  5. Z. He, S. Wu, M. Shao, S. Wu, H. Wang, J. Zhang (2026). 'Improved Line-of-Sight Trajectory Tracking for Vector-Propelled AAV Exposed to Underwater Dynamic Ocean Currents.' IEEE Journal of Oceanic Engineering. DOI: 10.1109/JOE.2025.3624976. https://doi.org/10.1109/JOE.2025.3624976
  6. L. Wang, Y. Zhong, P. Wang, Z. Gao, M. Lin, J. Zhang (2026). 'Osprey-fish gaming inspired planning and intelligent control of dual-rotor AAV swarm for cooperative hunting USV swarm.' Aerospace Science and Technology. DOI: 10.1016/j.ast.2026.112008. https://doi.org/10.1016/j.ast.2026.112008
  7. S. Zhao, S. Wu, S. Wu, M. Shao, J. Wang (2024). 'Numerical Analysis of Water-Exit Characteristics of Tandem Twin-Rotor Aerial-Aquatic Vehicle.' 2024 IEEE International Conference on Unmanned Systems (ICUS). DOI: 10.1109/ICUS61736.2024.10839903. https://doi.org/10.1109/ICUS61736.2024.10839903
  8. Y. Sun, Y. Xu, H. Shen, J. Liu, Y. Gao, Q. Zhao, A. Ji (2025). 'Analysis of water surface effects during the water-exit process of the vector bi-rotor aerial-aquatic vehicle.' Aerospace Science and Technology. DOI: 10.1016/j.ast.2025.110404. https://doi.org/10.1016/j.ast.2025.110404
  9. K. Han, Y. Bi, Y. Bai, Y. Wang, H. Zhou, Z. Zhang, Z. Zeng, L. Lian (2026). 'Robust Submergence and Wave-Adaptive Cross-Domain Strategy for Hybrid Aerial-Underwater Vehicles in Challenging Sea States.' IEEE Transactions on Field Robotics. DOI: 10.1109/TFR.2026.3691636. https://doi.org/10.1109/TFR.2026.3691636
  10. P. M. Pinheiro, A. Alves Neto, R. B. Grando, C. B. da Silva, V. M. Aoki, D. Cardoso, A. C. Horn, P. L. J. Drews-Jr (2021). 'Trajectory Planning for Hybrid Unmanned Aerial Underwater Vehicles with Smooth Media Transition.' arXiv:2112.13819. https://arxiv.org/abs/2112.13819
  11. A. Farinha, J. Di Tria, R. Zufferey, S. F. Armanini, M. Kovac (2021). 'Challenges in Control and Autonomy of Unmanned Aerial-Aquatic Vehicles.' 2021 29th Mediterranean Conference on Control and Automation (MED). DOI: 10.1109/MED51440.2021.9480342. https://doi.org/10.1109/MED51440.2021.9480342
  12. Y. Meng, V. Puig, H. Tang, D. Wu, H. Ye, X. Yang, Z. Xiang (2026). 'Aerial-aquatic unmanned amphibious vehicles: A review of prototypes, modeling, and control.' Ocean Engineering. DOI: 10.1016/j.oceaneng.2026.126801. https://doi.org/10.1016/j.oceaneng.2026.126801
  13. X. Liu, M. Dou, R. Yan, D. Huang, S. Gao, B. Wang, J. Cui, Q. Ren, L. Dou, Z. Gao, J. Chen, B. M. Chen (2024). 'TJ-FlyingFish: An Unmanned Morphable Aerial-Aquatic Vehicle System.' Unmanned Systems. DOI: 10.1142/S230138502441019X. https://doi.org/10.1142/S230138502441019X