Figure, OpenAI demo new robot that makes Musk’s Optimus look like a joke

A new startup called Figure partnered with OpenAI to develop a humanoid robot and new demo footage puts Elon Musk’s robot to shame.

The Figure video shows the startup’s ‘Figure 01’ robot doing tasks while communicating in natural-sounding language with a human. You can check out the video yourself below or read on for a description of what happens.

At the start, the human asks the robot to identify what it sees, and the robot says it sees a “red apple on a plate in the centre of the table, a drying rack with cups and a plate, and you standing nearby with your hand on the table.”

The human then asks Figure 01 for something to eat and the robot picks up the apple and hands it to him. After that, the person puts some trash on the table and asks the robot to explain why it gave him the apple while it picks up the trash.

“On it. So I gave you the apple because it’s the only, uh, edible item I could provide you with from the table,” the robot says. Note the concerningly human-sounding “uh” the robot sneaks into the sentence.

To wrap things up, the person asks Figure 01 where the plate and cup on the table should go, and the robot responds that they should likely go to the drying rack. Then, the person confirms and asks Figure 01 to put them in the rack. The robot places the items with impressive dexterity, picking up the cup, rotating it and passing it from one hand to the other. It also places the plate correctly in the drying rack next to the other plates.

Figure 01 vs. Optimus

At this point, you might be wondering what’s so special about the Figure 01. Other than it’s weirdly natural language, the tasks it performs in the video are relatively simple (at least by human standards). And besides, we’ve seen Musk and Tesla’s Optimus bot do plenty of tasks like folding a shirt, pick up an egg and even walk around. Well… have we?

As Gizmodo points out, Tesla’s Optimus bot appears to be teleoperated. As in, a human performs the action, and Optimus mimics it like a big puppet. In the folding shirt video Musk shared in January, for example, you can briefly see what appears to be a human beside the robot (and likely supposed to be off camera) doing the same folding shirt actions as the robot. Musk also admitted the video was staged, vaguely saying, “Optimus cannot yet do this autonomously,” and promising it would be able to in the future.

Teleoperation is, unfortunately, much less impressive, especially since we’ve been doing it with humanoid robots since at least the 1960s. Meanwhile, Figure claims there’s no teleoperation involved with its Figure 01 robot.

The video is showing end-to-end neural networks

There is no teleop

Also, this was filmed at 1.0x speed and shot continuously

As you can see from the video, there’s been a dramatic speed-up of the robot, we are starting to approach human speed

— Brett Adcock (@adcock_brett) March 13, 2024

“The video is showing end-to-end neural networks. There is no teleop,” wrote Figure co-founder Brett Adcock on Twitter/X. He also highlighted that the video was filmed at 1.0x speed and shot continuously. And Adcock isn’t the only one who said Figure wasn’t using teleoperation, with Corey Lynch, who works on AI at Figure, saying, “All behaviors are learned (not teleoperated).”

Let's break down what we see in the video:

All behaviors are learned (not teleoperated) and run at normal speed (1.0x).

We feed images from the robot's cameras and transcribed text from speech captured by onboard microphones to a large multimodal model trained by OpenAI that… pic.twitter.com/DUkRlVw5Q0

— Corey Lynch (@coreylynch) March 13, 2024

Coupled with the fact that Figure built a team of people from companies like Boston Dynamics and Google’s DeepMind, it seems like they know what they’re doing. And Figure’s repeated claims that there’s no teleoperation could be seen as a dig at Musk.

Source: Figure Via: Gizmodo