Parkour on the rooftops of Paris may be a pastime exclusively for daredevils, but the quadruped ANYmal robot is definitely no slouch either. Researchers at ETH Zurich have trained ANYmal to smoothly navigate obstacles in an urban environment, including the tricky terrain commonly found on building sites or in disaster areas.
To achieve this, the team combined machine learning with model-based control, the latter of which provides an easier way of teaching the robot accurate maneuvers, like how to recognize and get past gaps / recesses in piles of rubble. Machine learning aids the robot in mastering movement patterns that it can then flexibly apply in unexpected situations.
- Pre-assembled realistic robot dog STEM kit with lots of fun plays: Bittle is the world's first palm-sized, servo-activated bionic robotic dog with 1...
- Coding / STEAM / STEM / robotics learning: program your legged robot Bittle to perform new skills and tricks. Begin with Scratch-like block-based...
- Program AI tasks: program quadruped Bittle to see, listen, sense the surrounding environment and navigate the world with the optional intelligent...
Before the project started, several of my researcher colleagues thought that legged robots had already reached the limits of their development potential…but I had a different opinion. In fact, I was sure that a lot more could be done with the mechanics of legged robots,” said Nikita Rudin, ETH doctoral student.
