Machine learning challenge focuses on developing realistic digital hands

Many tasks in the digital world require precise control by means of a controller. In collaboration with the Artificial Intelligence Lab at Meta (Facebook), researchers at the University of Twente have set up a machine learning challenge. This involves developing controllers that make it possible to perform tasks requiring deft manipulation with a realistic digital hand, the University of Twente says in a press release.

The researchers are challenging anyone involved in AI or machine learning to use MyoSuite, a new open-source software. The MyoChallenge consists of two parts. The first challenge is to make the digital hand flip a die to achieve the desired accuracy. “The manipulation of a single object seems simple, but this task requires precise control of several muscles to be able to flip the die correctly without dropping it,” says Guillaume Durandau, one of the initiators of the challenge.

In the second track, participants will make the hand rotate two balls around each other. “Simultaneous rotation of two balls is even more complex. This not only requires dexterity like in the first track but also a high level of coordination,” explains Durandau.

Megaverse trains cybersecurity-awareness via adaptive and interactive 3D games

Founder and Technical Director, Greg Roy, swapped his job as creative art director at Ubisoft to start the Swiss company Megaverse.

Everyone can participate

The entire competition is split into two stages: an open stage and a play-off stage. In the open stage, anyone can join. Each participant can submit a maximum of five solutions per day. This stage is open right now until the 20^th of October. All participants are automatically ranked on a leaderboard and the top scorers can participate in the play-off stage.

In the play-off stage, the participants submit their top-3 policies which will be tested in a new environment. This means that the organizers will test the policies with new physics settings, for example, balls with a higher bounciness. Durandau: “We wanted to really challenge the submitted solutions. Using this setup, we can tune the difficulty of the final stage based on the results of the first stage.”