Pulsar flight computer detail

Technology

The nervous system and the intelligence.

Pulsar flies the vehicle. Arete makes it smarter every flight. Together they let a small team launch, operate, and learn from every mission with the discipline of a much larger program.

Program 02

Pulsar

Flight Computer

A radiation-tolerant flight computer and real-time software stack that runs every Astroncia vehicle. Pulsar is our chance to eliminate the single largest source of aerospace risk — the software the vehicle actually runs on.

Traditional avionics ship as bespoke boxes with vendor-specific tooling. Pulsar is designed as a platform: one board family, one OS, one deterministic bus. Every Astroncia vehicle — Starlight and everything after — runs the same core stack, tested against the same simulator, debugged with the same tools, and continuously improved by models developed on Arete.

Class

Flight computer, avionics stack

Processor

Multi-core, redundant lockstep

Radiation

SEE-hardened, TID-tolerant

Bus

TSN Ethernet + CAN-FD

Real-time OS

Preemptive, deterministic, in-house

Certification target

DO-178C spirit, ECSS aligned

Abstract visualization of Arete AI research platform

Program 03

Arete

Open-Source AI Research Platform

Arete is Astroncia's open-source AI research platform, free for researchers everywhere. It gives labs, students, and independent scientists the same training, evaluation, and deployment pipeline we use in-house — so the next breakthrough in autonomous flight can come from anywhere, not just from us.

Platform capabilities

Arete, at a glance.

Class

Open-source AI research platform

License

Permissive open source — free for researchers, labs, and students

Focus areas

Guidance, vision, autonomy, anomaly detection

Stack

Training, evaluation, and deployment pipeline — reproducible end to end

Compute

Runs on commodity GPUs, HPC clusters, and Pulsar-class edge inference

Data

Public telemetry, imagery, simulation, and hardware-in-the-loop datasets

Community

Open contributions, shared benchmarks, published research