Digit's compute needs continue to grow to match increasing real-world capabilities
At Agility Robotics, we’re constantly pushing the boundaries of what robots can perceive, decide, and do in the real world. That means our hardware isn’t the only thing evolving – our compute needs are growing just as fast. That’s why we’re excited to begin working on the new NVIDIA Jetson Thor.
Robots like Digit rely on deep learning for a wide range of capabilities:
Each of these requires massive amounts of onboard computation. As robots take on more challenging, safety-critical tasks,the demand for real-time AI that runs directly on the robot only grows, and the size and complexity of models increases as well.
We use a simulation-first workflow with NVIDIA Isaac Lab for reinforcement learning, and sim-to-sim validation. We’re also exploring NVIDIA Nvblox to advance navigation and obstacle avoidance—pipelines that ultimately depend on stronger onboard compute to run at scale.
This is where Jetson Thor comes in. Thor will allow us to run larger, more powerful, and more intelligent policies and reasoning models on our robots locally, in customer facilities and deployments.We see two particularly exciting areas for exploration:
End-to-End Manipulation
Digit’s ability to interact with objects in human environments depends on reasoning about context, contact, and control all at once. Running larger, end-to-end manipulation policies onboard will allow Digit to react faster and handle more complex tasks, from stocking shelves to assisting in logistics, and will improve generalization to new environments.
Advanced Scene Understanding
For robots working around people, awareness is everything. With Thor, we’ll explore richer onboard semantic perception. Again, with deep learning, scale at the edge is everything – so we’ll be able to test larger and more powerful models that can better handle the variable environments and problems our robots have to act in.
Exploring Jetson Thor isn’t just about raw performance benchmarks, it’s about informing the design of future robots, where compute is as fundamental as actuators and sensors. The ability to scale intelligence directly into the robot opens doors to building more flexible, more useful, and more general-purpose robots.
Learn more about NVIDIA Jetson Thor.