Counterfactual Safety Reasoning: Post-Mission What-If Proofs for Autonomous Drones
Year: 2026
Type: preprint
Abstract: Safety cases for autonomous drones are typically static documents, disconnectedfrom the runtime behaviour they are meant to assure. We present SCC (Safety CaseCompiler), a tool that takes safety requirements in a small DSL and compilesthem into: (i) executable Rust test stubs, (ii) Goal Structuring Notation (GSN)nodes, and (iii) cryptographic proofs in a hash chain. One source of truth (therequirements file) generates three verified artefacts. We demonstrate SCC on a3-requirement UAV envelope (speed, altitude, obstacle clearance), showing thata single-line change regenerates all artefacts and re-validates the chain. SCCis implemented in Rust (no_std compatible) and evaluated on 10 unit tests and 1end-to-end demo. We discuss how SCC integrates with the DO-326A security processand identify open problems.
Author: Andre Oliveira
Language: English
Subfield: Safety, Risk, Reliability and Quality
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7230846384
