Instructing robots to navigate new environments is hard. You’ll be able to practice them on bodily, real-world information taken from recordings made by people, however that’s scarce, and costly to gather. Digital simulations are a speedy, scalable strategy to educate them to do new issues, however the robots typically fail once they’re pulled out of digital worlds and requested to do the identical duties in the actual one.
Now, there’s doubtlessly a greater choice: a brand new system that makes use of generative AI fashions at the side of a physics simulator to develop digital coaching grounds that extra precisely mirror the bodily world. Robots skilled utilizing this technique labored with the next success fee than these skilled utilizing extra conventional strategies throughout real-world exams.
Researchers used the system, known as LucidSim, to coach a robotic canine in parkour, getting it to scramble over a field and climb stairs, regardless of by no means seeing any actual world information. The strategy demonstrates how useful generative AI may very well be in relation to instructing robots to do difficult duties. It additionally raises the likelihood that we may finally practice them in solely digital worlds. Learn the complete story.
—Rhiannon Williams
Africa’s AI researchers are prepared for takeoff
After we speak concerning the world race for AI dominance, the dialog typically focuses on tensions between the US and China, and European efforts at regulating the know-how. Nevertheless it’s excessive time we discuss one other participant: Africa.
African AI researchers are forging their very own path, creating instruments that reply the wants of Africans, in their very own languages. Their story isn’t solely one in all persistence and innovation, however of preserving cultures and combating to form how AI applied sciences are used on their very own continent. Nevertheless, they face many obstacles. Learn the complete story.
—Melissa Heikkilä