Traceable Reasoning for Autonomous Systems Engineering (TRASE)

Purpose and goal

The purpose of TRASE is to develop and validate Forge, an agentic AI platform for product development of mission-critical systems, in real R&D environments. Forge shall navigate, analyse and process product data such as requirements, tests and technical documentation, and deliver output that is more reliable and verifiable than existing tools. The goal is to achieve AI decision support and analysis that engineers can trust in safety-critical decisions, enabling significant productivity gains.

Expected effects and result

The project delivers Forge as an agentic software platform for product development, a benchmarking framework from Chalmers Industriteknik that measures system trustworthiness, and validated use cases at Einride and DREV. The project expects to contribute to the applied AI research community. Long term, results contribute to faster and safer Swedish product development, strengthened competitiveness and increased adoption of traceable, verifiable AI in mission-critical engineering.

Planned approach and implementation

Forge (Kiselsmedjan AB) coordinates the project and develops the agentic platform. Validation runs in two parallel industrial use cases: ML validation for autonomous vehicles at Einride and system safety for mechatronic systems at DREV. Chalmers Industriteknik develops a benchmarking framework that measures agent performance across a multitude of use cases and performance metrics within the safety-critical engineering domain.