The world’s first jet-powered humanoid robot has early validations and a sneak peek into the experimental area, according to roboticists at the Italian Institute of Technology (IIT).
With four little jet engines under its belt, the humanoid known as iRonCub can fly and perform complex missions.
Since flying is presently an unexplored field of study for humanoids, the researchers believe that these robots’ ability to fly may be advantageous in certain applications, such as disaster relief.
Also Read: Anushka: The Budget-Friendly Humanoid Robot from KIET with Advanced AI Capabilities
The IIT team started their trials in 2021, but since then, they have had difficulties keeping their robot from catching fire or even blowing up from motor exhaust.
Future-generation robotics
The iCub v2.5 and v3.0 platforms are the foundation for the iRonCub prototypes that the team has produced. IIT developed iCub, a research-grade humanoid robot, to assist in developing and testing embodied AI algorithms.
The majority of the iCub’s 53 degrees of freedom are found in the upper torso, with nine in each hand. It has a full-body skin, encoders in every joint, gyros, accelerometers, cameras, microphones, force and torque sensors, and cameras.
Now let’s talk about iRonCub. This robot has two jet engines on its arms and two more on a jetpack that is fastened to its back. iRonCub version 3 is the subject of current investigations.
Also Read: Figure AI Releases Advanced Humanoid Robot, Figure 02, with Cutting-Edge Features
The iCub’s hardware design underwent significant changes to make room for the external engines. A new titanium spine and heat-resistant coverings were included as safety measures.
The weight of the jet-powered iRonCub3 is about 154 pounds (70 kilograms). The exhaust temperature can rise above 600 degrees Celsius, and the turbines can produce a maximum thrust of more than 1000N.
Compared to the work on iRonCub2, the team has made tremendous progress with iRonCub3 testing, which is now in a newly constructed flight and control facility. iRonCub3 improves upon its predecessor in some ways.
This iteration, which is based on the iCub3 platform, has integrated force-torque sensors into the jetpacks and eliminated tendons. In addition, a new generation of control systems and planners that operate at higher frequencies have been devised, along with new electronics.
Researchers claim that these upgrades, as a whole, improve the robot’s functionality and performance.
Also Read: New Algorithm Enables Humanoid Robots to Mimic Human Facial Expressions
Humanoid flight control
A major obstacle in the field of airborne humanoid robotics is determining flight and walking trajectories, as well as the transitions between them.
To address this, a Python trajectory planning algorithm based on momentum was created by employing a direct multiple-shooting method. This planner will soon undergo testing on the real robot after being verified via simulation.
Algorithms for flight control were created using constrained quadratic programming optimization to control the robot’s attitude and position. Researchers claim that the framework guarantees adherence to physical limitations and can be modified to accommodate different numbers of jet turbines.
Also Read: Tesla to Deploy Humanoid Robots by 2025, Says Elon Musk
According to the team, the finding poses issues in humanoid robotics that go much beyond conventional ones.
Since turbine exhaust gases can reach temperatures of up to 800 degrees Celsius and speeds close to the speed of sound, thermodynamics is an important factor. Neural networks with physics-informed components are needed for real-time evaluation of multi-body aerodynamics.
Joints and turbines are examples of high- and low-bandwidth actuators that must be integrated into the controller settings. The duty of creating trajectories for turbines and motor dynamics falls to planners.
Researchers say experimental validation is dangerous and crucial, providing little opportunity for creativity.
Also Read: Qinglong, China’s Advanced Humanoid Robot, Showcased at WAIC 2024
