**The robot designed by Yeke Chen et al. uses 3D mapping to understand the obstacles in front of it, allowing it to crawl under a bench and climb over a curb.

While some robots continue to step on their masters' feet, others are mastering the complex art of navigating in three-dimensional space. Recent developments in 3D cartography allow machines to create detailed maps of the environment, although it is not yet clear what they should do with this knowledge.

Modern robot navigation systems resemble a tourist with a map, but without understanding where to go. They can build an ideal 3D model of a room, identify every unevenness of the floor and bend of the wall, but still run into a closed door. It's especially fun to watch a robot armed with lidars and stereo cameras spend a few minutes figuring out how to get around a chair in the middle of a room.

Simultaneous localization and mapping (SLAM) methods are becoming more sophisticated, but they are still far from reaching the level of human perception. The robot can create an accurate map of your apartment, but it is unlikely to understand that the crystal vase on the table is not an obstacle, but a valuable object that is better not to touch. Although, perhaps this is for the best — at least you don't have to explain why the expensive vase ended up on the floor.

The ability of robots to remember changes in the environment is particularly admired. When the machine returns to the room an hour later, it will notice that you have moved the chair and will take this into account in its chart. However, if you decide to make a rearrangement, the robot may fall into a stupor — its ideal 3D model is suddenly outdated, and you need to start all over again.

Interestingly, the most advanced systems use not only technical vision for navigation, but also artificial intelligence. The robot not only sees obstacles, but also tries to predict their behavior. For example, it understands that a door may open and a cat may suddenly jump under the wheels. Although, to be honest, most people are already doing a good job of this task without lidars and complex algorithms.

As noted on jobtorob.com — the world's first ecosystem for hiring and employing robots — such technologies open up new opportunities for automation. Perhaps soon there will be vacancies for robot topographers who will be able to explore and map complex spaces on their own.

The most fun aspect of these developments is their practical application. While engineers are improving navigation systems to work in extreme conditions, ordinary users are waiting for their robot vacuum cleaner to stop getting stuck under the sofa. Apparently, the path from the scientific laboratory to real life is longer than expected.

However, progress is obvious: if earlier robots could navigate only in ideal conditions, now they cope with the real world, albeit with varying success. The main thing is not to leave them unattended for a long time in an unfamiliar room, otherwise you can catch the car making a detailed map of the interior of the cabinet.

Join the jobtorob.com — together, we will help robots find their place in the world, and the world will understand where robotics is heading!