About Astral
Astral builds the autonomy stack for uncrewed systems. We treat autonomous robots as a systems problem—perception, geometry, planning, safety, simulation fidelity, and honest evaluation—and ship it as an open SDK, high-fidelity simulation, operator apps, and public datasets. Bring your own platform and models, or work with us on custom quadcopters, fixed-wing aircraft, rovers, and beyond.
How we work
We separate semantic understanding from metric geometry so perception, planning, and control can each be evaluated and improved independently—the foundation of a deployable stack that generalizes across aircraft and ground robots.
Every capability is measured in high-fidelity simulation before it reaches hardware. Honest, reproducible benchmarks—not demos—decide what ships.
Our SDK, documentation, and large-scale datasets like Yonder are public, so teams can build on our work, reproduce our results, and integrate their own platforms and models.
Built in the open
Our software, research, and datasets are public. The fastest way to understand Astral is to use them.