
The Italian Institute of Technology (IIT) has reached a milestone in humanoid robotics by demonstrating the primary flight of iRonCub3, the world’s first jet-powered flying humanoid robotic particularly designed to function in real-world environments.
The analysis crew studied the complicated aerodynamics of the unreal physique and developed a sophisticated {control} model for methods composed of a number of interconnected elements. The total work on iRonCub3, together with actual flight assessments, took about two years. In the most recent experiments, the robotic was capable of raise off the ground by roughly 50 cm whereas sustaining its stability. The achievement paves the way in which for a brand new technology of flying robots able to working in complicated environments whereas sustaining a human-like construction.
The aerodynamics and {control} research have been described in a paper published in Communications Engineering.
The analysis was carried out by roboticists of IIT in Genoa, Italy, in collaboration with the group of Alex Zanotti at DAER Aerodynamics Laboratory of Polytechnic of Milan—where a complete collection of wind tunnel assessments have been carried out—and the group of Gianluca Iaccarino at Stanford University—where deep {learning} algorithms have been used to establish aerodynamic models.
The robotic flight demonstration represents the most recent milestone of the Artificial and Mechanical Intelligence (AMI) Lab at IIT in Genoa, led by Daniele Pucci. Their analysis goals to push the boundaries of multi-modal humanoid robotics, combining terrestrial locomotion and aerial mobility to develop robots able to working in unstructured and excessive environments.
iRonCub3 is the technological evolution of earlier prototypes and is predicated on the most recent technology of the iCub humanoid robotic (iCub3), developed to be teleoperated. It integrates 4 jet engines, two mounted on the arms and two on a jetpack hooked up to the robotic’s again.
Modifications to the iCub {hardware} design have been required to help the exterior engines, equivalent to creating a brand new titanium backbone and including heat-resistant covers for defense. The robotic mixed with the jet engines weighs about 70 kg, whereas the generators can present a most thrust drive of greater than 1000 N. This configuration allows the robotic to hover and carry out managed flight maneuvers even within the presence of wind disturbances or environmental uncertainties. The exhaust temperature can attain 800 levels.
“This analysis is radically completely different from conventional humanoid robotics and compelled us to make a considerable leap ahead with respect to the cutting-edge,” explains Daniele Pucci. “Here, thermodynamics performs a pivotal function—the emission gases from the generators attain 700°C temperature and stream at practically the velocity of sound. Aerodynamics should be evaluated in real-time, whereas {control} methods should deal with each sluggish joint actuators and quick jet generators. Testing these robots is as fascinating as it’s harmful, and there’s no room for improvisation.”
The AMI analysis crew centered on the platform’s dynamic stability, which is made in particular complicated by the robotic’s humanoid morphology. Unlike standard drones, which have symmetric and compact constructions, iRonCub3 has an elongated form, with plenty distributed throughout movable limbs and a variable middle of mass. This required the event of superior flight stability models that think about the robotic’s multibody dynamics and the interplay between jet propulsion and limb actions.
Moreover, the movable limbs considerably complicate the aerodynamics, which change with each movement of any of the robotic’s limbs.

The researchers have carried out intensive wind tunnel experiments, superior Computational Fluid Dynamics (CFD) simulations and developed AI-based models able to estimating aerodynamic forces in actual time.
“Our models embrace neural networks skilled on simulated and experimental information and are built-in into the robotic’s {control} structure to ensure secure flight,” explains Antonello Paolino, first writer of the paper and Ph.D. scholar in a joint program between the IIT and Naples University, who spent a semester as a visiting researcher at Stanford University.
As a consequence, iRonCub3 is provided with AI-powered {control} methods that enable it to fly whereas dealing with high-speed turbulent airflows, excessive temperatures, and the complicated dynamics of multi-body methods.
The superior aerodynamic modeling developed by IIT demonstrates that it’s doable to keep up posture and stability even throughout non-stationary maneuvers, equivalent to sequential engine ignition or modifications in physique geometry.
These research will be transferred to different robots with unconventional morphologies, representing a novel case in comparison with classical drones, whose stability depends on symmetry and simplified {control} methods that usually neglect the robotic’s personal aerodynamics and thermodynamics.
The last design of iRonCub3 is the results of a sophisticated co-design course of, particularly developed to combine synthetic intelligence and multi-physics into the design of flying robots. These methods, that are progressive within the subject of robotics, enable for the simultaneous optimization of each physique form and {control} methods, contemplating the complicated interactions between aerodynamics, thermodynamics, and multibody dynamics.
Co-design was used to find out the optimum placement of the jet generators to maximise {control} and stability throughout flight. Advanced design methods have been additionally employed to handle the warmth dissipation generated by the engines, thus guaranteeing the structural integrity of the robotic even below excessive working circumstances.
The robotic has been utterly re-engineered to resist the tough circumstances related to aerial locomotion, introducing main enhancements centered on precision actuation, enhanced thrust {control} by way of built-in sensors, and superior planners for coordinated takeoff and touchdown.
Throughout the design course of, quite a few iterative changes have been made based mostly on the outcomes of superior simulations and experimental testing, resulting in the robotic’s present configuration. This strategy has allowed the crew to beat the restrictions of conventional methodologies and represents a step ahead within the automated and built-in design of complicated robotic methods.
The first flight assessments of iRonCub3 have been performed in IIT’s small flight-testing space, where the robotic was capable of raise off the ground by roughly 50 cm. In the approaching months, prototype testing will proceed and will probably be additional enhanced due to a collaboration with Genoa Airport (Aeroporto di Genova), which can present a devoted space that will probably be arrange and outfitted by the Italian Institute of Technology in compliance with all required security laws. The space will host future experimental campaigns.
Applications of flying humanoid robots like iRonCub3 are envisioned in quite a lot of future situations, equivalent to search-and-rescue operations in disaster-struck areas, inspection of hazardous or inaccessible environments, and exploration missions where each manipulation capabilities and aerial mobility are important.
More data:
Antonello Paolino et al, Learning aerodynamics for the {control} of flying humanoid robots, Communications Engineering (2025). DOI: 10.1038/s44172-025-00447-w
Citation:
Humanoid robotic achieves managed flight utilizing jet engines and AI-powered methods ( 18)
24
-humanoid-robot-flight-jet-ai.html
.
. The content material is supplied for data functions solely.
