ANYmal C

ANYmal C is the third generation of the ANYmal quadruped robot, developed by ANYbotics AG (ETH Zurich spin-off) and introduced in 2019. The design evolved directly from the research of Prof. Marco Hutter's Robotic Systems Lab (RSL) and was the first ANYmal generation intended simultaneously for commercial sale and broad academic access through the ANYmal Research Program.

The robot weighs less than 30 kg and features 12 active joints (3 DOF per leg × 4 legs) powered by proprietary ANYdrive series elastic actuators (SEA). The IP67-rated enclosure enables operation in industrial environments exposed to rain, dust and moisture. A lithium-ion battery provides more than 2 hours of autonomous operation at normal walking speed.

ANYmal C's perception stack is based on a 3D LiDAR scanner (Velodyne VLP-16) mounted on the robot's back and four Intel RealSense D435i depth cameras distributed around the body. An onboard Intel i7 computer runs the full autonomous navigation, localization and motion-planning stack based on ROS 1.

ANYmal C was used extensively in landmark research — including "Learning to Walk in Minutes Using Massively Parallel Deep Reinforcement Learning" (Rudin et al., CoRL 2021), where the legged_gym framework enabled training locomotion policies in under 20 minutes on a single GPU. The robot also featured in DARPA Subterranean Challenge research and industrial facility inspection studies.

Since 2022, ANYmal C has been superseded by ANYmal D, which offers greater payload capacity (50 kg), advanced inspection sensors (thermal camera, ultrasonic microphone) and a more powerful onboard computer (2× Intel i7 8th gen). ANYmal C remains available as a research platform through the ANYmal Research Program at dozens of universities worldwide.