Explainable Compliance in Autonomous Driving – Turning Traffic Rules into an Operational Asset

Automated Driving SAE Level 3 and Level 4 have been “almost ready” for years – yet large-scale market adoption remains limited. The core challenge is not only perception or planning, but something less visible and far less solved: provable legal compliance. Every autonomous maneuver implicitly claims legality. But traffic rules are fragmented across jurisdictions, context-dependent, and often ambiguous. Today’s systems struggle not only to follow these rules, but to demonstrate and explain compliance in a way that satisfies regulators, auditors, and ultimately users. This talk introduces runtime traffic law evaluation as a missing building block for scalable automated driving. We show how legislation can be transformed into structured, machine-readable rule sets and continuously evaluated within the vehicle’s decision-making stack. Key topics include:

• Formalizing obligations, permissions, and prohibitions into executable logic
• Handling ambiguity and exceptions in legal texts systematically
• Resolving rule conflicts in a transparent and deterministic way
• Integrating compliance checks with perception, maps, and behavior planning

Beyond enforcement, we focus on explainability as an enabler. By generating traceable compliance evidence – such as proof trees, decision logs, and requirement mappings – systems can justify their behavior to engineers, safety assessors, and authorities.

The central idea: Traffic law should not be treated as documentation – but as an operational asset. Embedding explainable compliance into the driving stack is a key step toward unlocking safe, certifiable, and scalable Level 3 and Level 4 deployment.