ASTR-1 is a controlled evidence workflow, not just a candidate generator.
ASTR-1 is a battery-first, decision-first materials discovery architecture. It routes candidate materials through generation, screening, prediction, validation, repair, synthesis planning, optional physical testing, and failure learning — so each decision is anchored in evidence appropriate to its risk.
An interactive view of the ASTR-1 pipeline.
Hover or tap any node to see what happens at that stage. The Master Route Controller orchestrates every step.
- 01Request
- MRCMaster Route Controller
- 02Generate
- 03Uncertainty Gate
- 04QCVS (Validation)
- 05Dopant Prediction
- 06Property Prediction
- 07Result Capturing
- 08Reporting
- 09Optional Robotic Synthesis
A research goal enters the system — target chemistry, application envelope, and constraints define the run.
Seven specialised pipelines, one controller
Each pipeline produces evidence the Master Routing Controller uses to decide what happens next to a candidate.
QIGA candidate generation
Quantum-Inspired Generative Algorithms propose candidate compositions and structures. QIGA is quantum-inspired but implemented with classical algorithms; no quantum hardware is required.
Direct property prediction
Learned models estimate target properties from structure and composition to triage candidates before more expensive evaluation.
Feasibility screening
Filters for stability, element availability, toxicity, cost, and known failure patterns remove unviable candidates early.
Validation escalation
An uncertainty gate routes high-impact or low-confidence candidates to higher-fidelity validation rather than accepting predictions at face value.
Dopant and composition repair
When a candidate is promising but flawed, ASTR-1 proposes targeted dopants or composition adjustments to repair specific weaknesses.
Synthesis planning
Surviving candidates are paired with candidate synthesis routes and precursors, considering practical lab constraints.
Physical testing / robotic lab
Optional, permission-gated handoff to physical or robotic experimentation, with results fed back into the failure database.