ANYmal Robot

The ANYmal Robot is a highly versatile quadrupedal robot designed for navigating and performing tasks in challenging environments. Developed by the Robotic Systems Lab at ETH Zurich, ANYmal combines advanced locomotion capabilities with a modular design that allows it to be equipped with various sensors and tools. It can dynamically run, climb, and even perform parkour-like movements, making it ideal for operations in rough terrains such as industrial sites, mines, and disaster zones. ANYmal’s body is lightweight yet robust, being water- and dust-proof (IP67 rated), and it can withstand falls from up to 0.5 meters. Equipped with LIDAR, stereo, and wide-angle cameras, it has comprehensive environmental perception capabilities, allowing it to navigate autonomously. The robot operates for 2-4 hours on a single charge and can autonomously dock for recharging. The hardware architecture is highly extendable, accommodating additional payloads such as robotic arms or specific sensors ANYbotics ScienceDaily Research Features).

. Also see the website.

• 4 Intel Realsense D435 RGB-D Cameras: Primarily used for local navigation and terrain assessment.
• 16-Beam Velodyne LiDAR: Provides 3D mapping and obstacle detection.
• Force-Torque Sensors: Installed in the robotic arm for precise manipulation tasks.
• Multi-Sensor Fusion: Integrates data from IMUs, cameras, and LiDAR to ensure stable and accurate localization even in GPS-denied environments.

ANYmal’s modular design allows it to be equipped with a robotic arm for manipulation tasks, such as opening doors or moving debris, with a payload capacity of up to 7 kg. In search and rescue (SAR) scenarios, ANYmal can autonomously explore, map, and interact with the environment to support human responders, offering real-time data through a wireless network. Additionally, it can be teleoperated using a 6DoF joystick, with data visualized on a 3D screen, making it a versatile tool in dynamic SAR operations.

Interview insights

Field teams prefer robots first, then humans where practicable; the robot can verify ambiguous detections, check voids, and pre-mark hazards before entry. Operators stressed a split between control and analysis: one person supervises locomotion/mode management, another verifies imagery and places markers with confidence and rationale. To fit work rhythms, setup happens at the BoO, with just-in-time activation at the worksite; handovers (e.g., after a battery swap or link drop) must be explicit, with a clear “resume from last waypoint” option. Avoid blocking access routes and ensure the robot’s thermal/noise signature does not interfere with communication or victim detection.