Research · June 2026

Reinforcement Learning & Robotics Testbed

Training control policies on our GPU cluster and moving them onto physical robotic hardware — the ground floor of autonomous systems.

AI guidance and control visualization

The autonomy we write about in our mission areas has to start somewhere physical. This is it.

Our robotics program pairs deep reinforcement learning — policies trained in simulation on our on-premises GPU cluster — with real hardware, working toward controllers that transfer from sim to the physical world.

Why it matters

Every long-term SDI mission — debris capture, autonomous manufacturing, supply-chain drones — reduces to the same core competency: machines that perceive, decide, and act without a human in the loop. Reinforcement learning on our own compute is how we build that muscle now, on hardware we can afford to crash.

Status

Early-stage research, actively underway.

← All projects

Ready to Build the Future With Us?

Whether you need AI capability on the ground today or share our vision for sustainable space industry, we want to hear from you.

Get in Touch