A publication from Cornell University focuses on machine learning’s ability to overcome one of the nagging issues for robotics: robots fall over and can’t get up.
Solving this problem could make a big difference for cobots, for robots working in unpredictable environments, and for humanoid robots everywhere. Though quadrupedal robots have recently been found to be a bit more stable than bipedal ones, all robots with any significant level of mobility have the problem of falling over. The Cornell paper brings up the real possibility of a robot that can right itself and get back to work.
What’s getting attention?
Ars Technica points out that this truly important step forward for robotics is not what’s getting attention. Social media reactions are all about the casual cruelty of the researcher who keeps pushing the poor little robot down while it’s trying to play.
In fairness, comments in the Twitter thread also celebrated the smooth operation of the little dudes.