Case Study: Agility Robotics achieves faster, more reliable humanoid robot deployment with NVIDIA Isaac Sim and Isaac Lab

Agility Robotics trains Digit in hours with NVIDIA, cutting controller iteration from weeks to days

Agility Robotics, a developer of general-purpose humanoid robots, faced the inherent complexity of achieving reliable whole-body control for its Digit robot to operate safely alongside humans in various environments. To overcome this challenge and accelerate development, Agility Robotics leveraged NVIDIA Isaac Sim and Isaac Lab, robotics simulation and learning frameworks built on NVIDIA Omniverse.

Using a simulation-first approach powered by NVIDIA GPUs, Agility Robotics trained Digit across billions of simulation steps. This NVIDIA solution enabled a breakthrough training methodology that reduced controller iteration cycles from weeks to days. The result was improved step-recovery performance and enhanced real-world deployment success, with Digit robots now reliably performing material-handling tasks in live operations for customers like GXO Logistics and Schaeffler.

Agility Robotics