AI ASSISTED SYSML V2 SPACE MODEL INTERACTION AND KNOWLEDGE REUSE - EXPRO+

Objective:To develop and validate an AI-based assistant for interactive access, querying and assisted manipulation of SysML v2 space system models to reduce modelling effort and increase design quality.Description:Model-Based Systems Engineering (MBSE) is an approach to manage the complexity of future space systems.Working effectively with large digital models is challenging; understanding structure and dependencies,reusing assets across projects, and extracting decision-relevant information is time-consuming andexpert-dependent.Limited and non-intuitive access to very large models slows MBSE adoption and reduces the return oninvestments, especially in early phases where it drives duplication, and late inconsistency discovery causeshigher design risk. SysML v2 introduces combined textual and graphical modelling with a standardised API for programmatic access, which lowers the access barrier. AI - and particularly Large Language Models (LLMs) - can help translate user intent into controlled, explainable interactions with these formal SysML v2 models, overcoming the barriers identified.This activity will develop an AI assistant that acts as an intermediate layer between users and formal modelsand operates exclusively via the SysML v2 textual notation and standard API. It will support (i) natural-language assistance for incremental model creation and modification, (ii) model querying and automated consistency checking, (iii) guided exploration of dependencies, (iv) identification and extraction of reusable model elements, and (v) structured summarization of model content for expert and non-expert stakeholders. The focus is on human assistance, incremental model generation and knowledge exploitation that helps system engineerswork with large system models - not on autonomous end-to-end model generation or mission-specifictoolchains.This activity encompasses the following tasks:- Review of SysML v2 capabilities, APIs, and relevant AI techniques for model creation and interaction- Identification of representative use cases for querying, exploration, and reuse of SysML v2 models- Definition of a reference architecture ensuring controlled AI interaction via the SysML v2 API- Development of proof-of-concept AI components for model generation, querying and summarization- Validation of the approach on representative SysML v2 models, using the ESA MBSE methodology- Assessment of benefits, limitations, and roadmap towards higher TRL.The activity will result in validated concepts, software prototypes, and documented lessons learned, enablingfollow-on developments under GSTP or other programmes focusing on operational deployment.Software shall be delivered under an ESA Software Community License, so that any individuals or entitieswithin ESA Member States can access it and can provide update to the community of users.