Selected from techxplore
By Ingrid Fadelli
Machine Heart Compilation
Edit: Egg sauce
"Look, that big-eyed robot looks like Iron Man!"
All along, Iron Man Tony Iron Man is loved by Marvel fans around the world, wearing a high-tech "iron suit" and flying in the air, which is really enviable.
In the real world, many people want to have Iron Man's ability to fly, and there are many attempts. In May this year, the British Navy developed a manned jet backpack that attracted heated discussions, which flew at a speed of up to 136 km / h, although the flight altitude was not high, but more or less had the shadow of Iron Man.
https:// www .youtube.com/watch?v=NSeDy74pDNE
A research team from the Italian Institute of Technology is working to create humanoid robots capable of flying. They found that in order to efficiently control the movement of flying robots, flying objects, or vehicles, a system needed to be built that could reliably estimate the thrust intensity generated by the thrusters so that they could fly in the air. Because thrust is difficult to measure directly, estimates are often made based on data collected by airborne sensors.
IIT's team recently introduced a new framework that can estimate the thrust strength of flying multibody systems that are not equipped with thrust measurement sensors. The framework, published in a paper in IEEE Robotics and Automation Letters, could eventually help them realize the flying humanoid robots they envisioned.
"Our early ideas for making flying humanoid robots came around 2016," Daniele Pucci, head of the Artificial Intelligence and Mechanical Intelligence Lab who conducted the study, told TechXplore. "The main purpose is to envision robots that could operate in a disaster-like situation, in which case survivors could rescue themselves inside partially destroyed buildings that would be difficult to reach due to possible flooding and fires around them."
Address: https://ieeexplore.ieee.org/abstract/document/9622189
The main goal of Pucci and his colleagues was to design a robot that could manipulate objects, walk and fly on the ground. Given that many humanoid robots can both manipulate objects and move on the ground, the team decided to expand the capabilities of the humanoid robot to increase its ability to fly, rather than developing entirely new robot structures.
"Once equipped to fly, humanoid robots can fly from one building to another, dodging debris, fires and floods," Pucci said. "After landing, they can manipulate objects to open doors and close air valves, or enter buildings for indoor inspections, such as looking for survivors of fires or natural disasters."
Pucci and his colleagues tried to provide iIT's famous humanoid robot, the iCub, with the ability to maintain balance on the ground, such as standing on one foot. Having done this, they began working to broaden the robot's motor skills so that it could fly and move through the air. The team refers to the area of research they have been focusing on as "aerial humanoid robots."
"As far as we know, we completed the first study on a flying humanoid robot," Pucci says, "this paper is obviously just testing flight controllers in a simulated environment, but with expectations, we began our journey to design the iRonCub, which is the first jet humanoid robot introduced in our latest paper." The thrust estimation framework proposed by the researchers greatly simplifies the design of the flying robot and reduces the manufacturing cost. Instead of using force sensor data to estimate thrust, the framework combines two different sources of information into a single estimation process. Therefore, the researchers do not need to install force sensors on each jet engine that pushes the robot.
The first source of information used by the framework is a model that links commands sent to jet engines to the resulting thrust. This is a data-driven model that was trained on data collected by researchers. "We first built a special experimental device that looks like a fireproof, bulletproof room for jet engines and safe experiments," Pucci says.
"Then, by using this setup, we collect input/output data from jet engines and select models that describe engine behavior. These build on our previous work."
The second source of information the framework uses to estimate thrust is what the robot calls "mass heartbeat." This is a well-known theorem developed by humanoid systems of robots to control and estimate their motion.
"If used properly, for example, this theorem could describe the action of a diver jumping off a cliff," Pucci said. In other words, it can be used to relate the cause (i.e., thrust) and effects of robot movement before and after takeoff (such as vertical acceleration of takeoff).
If used separately, both sources of information for the framework will have significant limitations. For example, with a data-driven model, thrust can only be accurately estimated if jet engines always work in exactly the same way. However, jet engines behave differently depending on environmental factors.
"The second method doesn't use internal jet information," Pucci says. "So we used Kalman filtering to combine these two methods to overcome their respective shortcomings. It is worth noting that our estimation method is independent of the special properties of flying humanoid robots and can be used for flight controllers designed for any flying multibody robot."
To assess the effectiveness of the framework, Pucci and his colleagues tested a new type of robot called the iRonCub, an evolved version of the iCub robot with an integrated jet engine. While the team has been working on this robot for some time, it has only recently been able to demonstrate its full capabilities.
Pucci explains: "Operating a jet-powered robot is not an easy task, as the jet temperature can reach 700 degrees Celsius and the air velocity may have supersonic characteristics, with a flow rate of about 1800 km/h."
"For this reason, we have developed rigorous experimental procedures and protocols that allow us to work safely with iRonCub. The research team had to overcome a number of problems that were far removed from traditional robotics research and from avionics."
While the researchers have only tested the thrust estimation framework of their humanoid robot iRonCub, it could also be applied to other flying robots with different body structures. This includes reconfigurable flying robots that can be reshaped or configured to perform specific actions. "In any case, the issue of estimating thrust is the key to a successful flight," Pucci says.
"In addition to applications in disaster-like scenarios, we believe this work can be applied to simpler designs than flying humanoid robots, including jet-powered flight boxes."
If applied to jet boxes, the thrust estimation framework developed by the researchers could open up new opportunities for transporting a variety of products, including food and medicine, to remote areas. With funding from the European Union or the Science Foundation, Pucci and his colleagues hope to explore this possible application in more depth. In the meantime, the team plans to continue developing the iRonCub, focusing on its flight capabilities. They hope to eventually be able to create the first reliable and high-performance humanoid robot that can perform both land and air movements.
Original English: https://techxplore.com/news/2021-12-humanoid-robot.html
Quickly build an enterprise-grade TTS speech synthesis assistant with NVIDIA Riva
NVIDIA Riva is an SDK that uses GPU acceleration to rapidly deploy high-performance conversational AI services for rapid development of speech AI applications. Riva is designed to help you easily and quickly access sessionAL AI capabilities, out of the box, and quickly build high-level TTS speech synthesis services with a few simple commands and API operations.
January 12, 2022, 19:30-21:00, the main introduction of this online sharing:
Introduction to speech synthesis
Introduction to NVIDIA Riva features
Launch the NVIDIA Riva client to quickly implement text-to-speech functionality
Use Python to quickly build Riva-based TTS speech synthesis service applications