ETH Zurich Teaches Four-Legged Robot to Climb Ladders Using AI
Researchers at ETH Zurich successfully train a quadruped robot to climb standard ladders using reinforcement learning and specialized grippers. The new system achieves a 90% success rate and climbs 232 times faster than previous state-of-the-art methods.
Researchers at ETH Zurich tackle one of the biggest remaining challenges for four-legged robots by teaching the ANYmal quadruped how to climb standard ladders. While robots like Boston Dynamics' Spot easily handle stairs and uneven terrain, ladders remain a major obstacle in industrial environments where these machines operate. Previous attempts to solve this problem rely mostly on slow, bipedal humanoid robots and specially modified ladders.
The ETH Zurich team equips the ANYmal robot with specialty hook-like end effectors to grip the rungs and uses reinforcement learning as the system's secret weapon. This artificial intelligence approach allows the robot to adapt to the unique variations and peculiarities of different ladders in real-time. By combining the robot's physical morphology with this advanced control policy, the system learns to perform this highly complex skill autonomously.
The results show significant improvements over existing technology, with the robot achieving a 90% success rate on ladders angled between 70 and 90 degrees. It also climbs 232 times faster than current state-of-the-art systems and corrects itself mid-climb if it misjudges a rung or mistimes a step. This breakthrough expands the potential for industrial quadruped robots beyond basic floor inspections to navigating complex infrastructural features.