Scientists have summarized billions of years of evolution in a flash with an AI that created a miniature walking robot from scratch.
We’ve come a long way since the dawn of robotics in the early 1950s, and scientists have tried to build on the discipline in the decades since. While there have been many successes, including much of the automation we see in factories today, there are also plenty that researchers are trying to iron out.
Likewise, we’ve seen plenty of examples of robots that can walk and navigate their terrain, from the likes of Boston Dynamics, but it’s taken years to get to this point.
Tapping into AI-driven ‘instant evolution’
Now, an AI system devised by a team at Northwestern University has intelligently designed a walking robot in just 26 seconds. The findings will soon be published in Proceedings of the National Academy of Sciences.
The AI program can design new systems and structures from scratch and does not run in a data center, but on a normal PC.
This is different from many similar AI systems that rely on massive amounts of power. Researchers then tested the AI-powered algorithm they created by giving it the task of designing a robot that could walk on a flat surface.
“We have discovered a very fast AI-driven design algorithm that bypasses the traffic jams of evolution without falling back on the biases of human designers,” says project leader Sam Kriegman. TechXplore.
“We told the AI that we wanted a robot that could walk on land. Then we simply pressed a button and presto – it instantly generated a blueprint for a robot that looks nothing like any animal that has ever walked the Earth. I call this process ‘instantaneous evolution.’
The computer started with a block the size of a bar of soap, and the AI went through several iterations to review the design, identify flaws, and physically cut out the structure to perfect the design.
After nine attempts, the robot he designed was able to walk at a pace equal to the human stride, but of course reduced in size compared to its own size.