A new robot from the Technical University of Munich (TUM) is doing something most others can’t, which is actually understanding what it’s looking at.

Developed at Prof. Angela Schoellig’s Learning Systems and Robotics Lab, the robot is designed to solve a very human problem—finding your lost glasses. To do this, it doesn’t just wander around blindly. It uses its camera to build an accurate 3D map of a room down to the centimeter. A laptop connected to the bot then helps it figure out what the objects in that map actually are and why they matter to us.

A Robot That Thinks Like a Human

What makes this system stand out is that it uses a language model to understand common sense. For example, it knows that if you lost your glasses, you might have set them on a table or a windowsill. It also knows you probably didn’t leave it in a ridiculous and improbable place.

“We have taught the robot to understand its surroundings,” said Prof. Angela Schoellig. This kind of logic helps the robot search a room about 30% more efficiently than if it were just guessing.

The robot also has a great memory. It compares what it sees now to what it saw earlier. For example, if a new object suddenly appears on the kitchen counter, the robot notices the change with 95 percent certainty and marks it as a likely place to find what’s missing.

From the Home to the Factory Floor

This tech is a building block for robots that need to work in places that are always changing, like a busy factory or a private home. As Schoellig explains, this basic understanding “is important for all robots that move in spaces that are constantly changing.”

In the future, the team wants the robot to start looking inside drawers and behind cupboard doors. This will require the robot to do more than just look. To do this, it will need to use mechanical arms to figure out if a door swings sideways or pulls upward.