iRonCub3 is equipped with a jet engine, weighs 70 kg, produces a thrust of more than 1,000 N and an exhaust temperature of over 600 degrees Celsius.
Robot developers at the Italian Institute of Technology (IIT) have introduced a testing area and preliminary verification of the world’s first jet-powered humanoid robot, Interesting Engineering The robot, called iRonCub, has four compact jet engines that give it the ability to fly and perform advanced tasks. The team believes that flight capabilities will benefit robots in a variety of applications, including disaster relief.
The IIT team, which began testing in 2021, faced several challenges, including preventing the robot from catching fire or even exploding due to engine exhaust. They developed prototypes of the iRonCub, built on the iCub v2.5 and v3.0 platforms. The iCub is a humanoid research robot created by IIT, designed to help develop and test AI algorithms.
The iCub can rotate 53 degrees of freedom, has a full-body skin, force/torque sensors, cameras, microphones, gyroscopes, accelerometers and encoders at each joint. With the iRonCub, both versions have 4 jet engines, 2 in the arms and 2 in the jetpack attached to the robot’s back. Current testing focuses on the iRonCub version 3. The research team has made several important improvements to the iCub’s hardware design to accommodate the external engines, such as creating a titanium spine and adding a heat-resistant protective layer. The iRonCub3 weighs about 70 kg. The turbine produces a maximum thrust of more than 1,000 N and an exhaust temperature of over 600 degrees Celsius.
The researchers are now testing the iRonCub3 in the flight and control area, marking a significant step forward from the iRonCub2. The iRonCub3 has several upgrades over its predecessor. Based on the iCub3 platform, this version removes the tether and integrates torque sensors into the jetpack. New electronics have also been designed, including a next-generation control and planning system that operates at higher frequencies. According to the team, the improvements enhance the robot’s capabilities and performance.
A key challenge for aerial humanoid robots is flight and travel path planning, including transitions between the two. A momentum-based path planning algorithm was developed in Python, using a multiple-shot approach to solve the problem. The algorithm has been validated in simulation and will soon be tested on a real robot.
For flight control, the algorithm is designed to manage the robot’s orientation and position, applying conditional optimization. The team says the mechanism can be adjusted to accommodate a wide range of jet turbines. They say the project is far more complex than a traditional humanoid robot. Thermodynamics are essential, with the exhaust gases from the turbines reaching hundreds of degrees Celsius and approaching the speed of sound. The system’s aerodynamics require neural networks to evaluate in real time.