Robot That Walks and Flies Learns to Choose

Imagine a robot that can walk like a dog and fly like a drone. That's the Husky-Beta, a machine built by researchers at Northeastern University. Arjun, now an engineer at Boston Dynamics, will explain how he taught this robot to cross rough terrain by switching between walking and flying. The key is a type of AI called hierarchical reinforcement learning, which lets the robot learn not just how to move, but which mode of movement to use. It's like teaching a child to both walk and ride a bike, and then letting them decide which is better for the path ahead.

Why should you care? Robots that can both walk and fly could one day deliver packages to your door, inspect bridges, or search for survivors in disaster zones where vehicles can't go. They could also help with environmental monitoring or even space exploration. The ability to choose the right mode of movement makes them much more versatile than robots that only walk or only fly.

The talk is part of the Boston Robot Hackers monthly meeting on October 15 from 7 to 9 PM. It's free and open to the public, but you need to register. Arjun will explain the technical details in an accessible way, so even if you're not a robotics expert, you'll learn something fascinating. He'll also discuss how this approach could apply to other robots that need to adapt to different situations.

This is a great chance to peek into the future of robotics and see how AI is making machines smarter and more capable. Whether you're a tech enthusiast or just curious about where robots are headed, this event is for you.