iCub Is Growing Up

iCub3 may be only a little bit taller than its predecessor, but it’s much more capable, especially for telepresence

3 min read
A child-like robotic face

It’s a little weird to think that iCub, the first version of which was released in 2008, is now a full-on teenager—after all, the robot was originally designed after a three year old child, and it hasn’t grown all that much. But as it turns out, iCub has grown, at least a little bit, with the brand new iCub3, a bigger, heavier, stronger upgrade that can also be used as a sophisticated immersive telepresence platform.

Two technical images comparing two similar robots of slightly different sizes The larger iCub3, on left, compared to the original iCub, on right.

You might not notice the difference between the old and new iCub unless they’re standing side-by-side, but iCub3 is 25 cm taller and weighs 19 kg more, featuring more powerful legs and rearranged actuators in the legs, torso, and shoulders. The head is the same, but the neck is a bit longer to make the robot more proportional overall. In total, iCub3 is 1.25 m tall and weighs 52 kg, with 54 degrees of freedom.

Part of the reason for these upgrades is to turn iCub3 into an avatar platform—that is, a humanoid robot that can embody a remote humanoid of the more biological type. This is something that researchers at the Italian Institute of Technology (IIT) led by Daniele Pucci have been working on for quite a while, but this is by far the most complete (and well filmed) demonstration that we’ve seen:

Now, a couple of those shots right at the beginning look more like drone videos than an actual representation of the user experience, or at least, it won’t be accurate until iCub can fly, which they’re working on, of course. It’s also, honestly, kind of hard to tell what the user experience actually is just based on this very very fancy video, and very very fancy videos rarely inspire confidence about the capabilities of a robotic system. Fortunately, the researchers have published a paper on using iCub3 as an avatar which includes plenty of detail.

This particular demo took place over a distance of almost 300km (Genova to Venice), which matters in the sense that it’s not over infrastructure that is completely under the researcher’s control. Getting it to work requires (as you can see from the video) a heck of a lot of equipment, including:

  • HTC Vive PRO eye4 headset
  • VIVE Facial Tracker5
  • iFeel sensorized and haptic suit6
  • SenseGlove DK17 haptic glove
  • Cyberith Virtualizer Elite 28 omnidirectional treadmill

Put together, all of this stuff allows for manipulation, locomotion, voice, and even facial expressions to be retargeted from the human to the robot. The facial expressions retargeting is a new one for me, and it applies to eye gaze and eyelid state as well as to the user’s mouth.

Coming back the other direction, from robot to human, is audio, an immersive first-person view, and a moderate amount of haptic feedback. iCub3 has sensorized skin with a fairly high amount of resolution, but that gets translated to just a couple of specific vibration nodes worn by the user. The haptic feedback at the hands is not bad—there’s force feedback on each finger and fingertip vibration motors. Not quite at the level we’ve experienced with HaptX, but certainly functional.

A humanoid robot stands next to a human wearing a VR headset at haptic gloves in a laboratory

The application here is a remote visit to the Italian Pavilion at the 17th International Architecture Exhibition, which certainly seems to be a very remote-visit-worthy space. But a more practical test will be taking place soon, as iCub3 takes place in the final round of the ANA Avatar XPRIZE competition which “aims to create an avatar system that can transport human presence to a remote location in real time.”

The Conversation (1)
Richard Rankin 29 Mar, 2022

What is the point in creating robots that resemble humans? They don’t make people more comfortable with them, in fact the opposite. Deviating from pure functionality in design requires both cost and opportunity cost. Perhaps such devices have a “cuteness” appeal to those not interacting with them directly that works in publicity. I fail to see the point in this type of design.

How the U.S. Army Is Turning Robots Into Team Players

Engineers battle the limits of deep learning for battlefield bots

11 min read
Robot with threads near a fallen branch

RoMan, the Army Research Laboratory's robotic manipulator, considers the best way to grasp and move a tree branch at the Adelphi Laboratory Center, in Maryland.

Evan Ackerman

“I should probably not be standing this close," I think to myself, as the robot slowly approaches a large tree branch on the floor in front of me. It's not the size of the branch that makes me nervous—it's that the robot is operating autonomously, and that while I know what it's supposed to do, I'm not entirely sure what it will do. If everything works the way the roboticists at the U.S. Army Research Laboratory (ARL) in Adelphi, Md., expect, the robot will identify the branch, grasp it, and drag it out of the way. These folks know what they're doing, but I've spent enough time around robots that I take a small step backwards anyway.

This article is part of our special report on AI, “The Great AI Reckoning.”

The robot, named RoMan, for Robotic Manipulator, is about the size of a large lawn mower, with a tracked base that helps it handle most kinds of terrain. At the front, it has a squat torso equipped with cameras and depth sensors, as well as a pair of arms that were harvested from a prototype disaster-response robot originally developed at NASA's Jet Propulsion Laboratory for a DARPA robotics competition. RoMan's job today is roadway clearing, a multistep task that ARL wants the robot to complete as autonomously as possible. Instead of instructing the robot to grasp specific objects in specific ways and move them to specific places, the operators tell RoMan to "go clear a path." It's then up to the robot to make all the decisions necessary to achieve that objective.

Keep Reading ↓ Show less