LLM-Empowered Event-Chain Driven Code Generation for ADAS in SDV systems

TL;DR

This paper introduces an event-chain-driven workflow enhanced by large language models (LLMs) for generating validated automotive code from natural language, utilizing retrieval-augmented prompts to improve accuracy and consistency.

Contribution

It presents a novel integration of retrieval-augmented LLMs with event chains for safe, real-time automotive code generation from natural language requirements.

Findings

Achieved valid signal usage in emergency braking case study.

Generated consistent code without retraining LLMs.

Ensured behavioral correctness and real-time feasibility.

Abstract

This paper presents an event-chain-driven, LLM-empowered workflow for generating validated, automotive code from natural-language requirements. A Retrieval-Augmented Generation (RAG) layer retrieves relevant signals from large and evolving Vehicle Signal Specification (VSS) catalogs as code generation prompt context, reducing hallucinations and ensuring architectural correctness. Retrieved signals are mapped and validated before being transformed into event chains that encode causal and timing constraints. These event chains guide and constrain LLM-based code synthesis, ensuring behavioral consistency and real-time feasibility. Based on our initial findings from the emergency braking case study, with the proposed approach, we managed to achieve valid signal usage and consistent code generation without LLM retraining.