The intersection of robotics and machine learning

Machine Learning


Ten years ago, robots could barely move. Today, the autonomous walking robots developed by Marco Hutter and his team are revolutionizing every field. In the future, these highly maneuverable robots may even be able to assist search and rescue operations and explore other planets. Hutter, a professor in the Department of Mechanical and Process Engineering at ETH Zurich, received this year's Rössler Prize for his research. “How can we make robots that move over terrain like humans and animals?” was the initial question that sparked his research journey.

It soon became clear that the robot needed legs, not wheels or a caterpillar drive. But this answer raised other questions about what a walking robot should look like and how it could function. “We started working on developing a new drive and control concept for a robot that initially couldn't do anything and was far from being able to be used effectively,” explains Hutter.

Robots that can overcome obstacles

Over the years, technology has continually evolved, and today's robots are capable of navigating the most challenging terrains and being used in real-world situations. Hutter gives some examples: “Currently, our walking robots are used commercially to carry out industrial inspections, and we are also investigating new capabilities that would allow us to use the robot for search and rescue operations in completely unknown environments.”

Hatter's team has been using artificial intelligence for years. Thanks to machine learning, robots learn to understand and interact with their environment. If the walking robot stands in front of an obstacle, it recognizes it with a camera and an artificial neural network helps it understand what kind of obstacle it is. Then, like an eager dog, it executes the behavior it successfully learned in previous training sessions. So it's only fitting that these four-legged robots are called “ANYmal,” a combination of “anywhere” and “animal.”



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *