AI4SDV/Project

About AI4SDV

AI4SDV increases the productivity of Software-Defined Vehicle (SDV) engineering and shortens time-to-market through automation and AI-enabled development support – fully integrated with the open-source initiatives of the European automotive industry.

Overview

Software complexity in vehicles has grown roughly four-fold over the past decade, while software-development productivity has barely kept pace, and the industry faces a growing shortage of skilled engineers. AI4SDV develops, tests and matures 90+ AI-assisted methods and tools that automate key steps of SDV engineering – from requirements to deployment and in-use monitoring – orchestrates them through an open and extensible platform, and demonstrates them at TRL 6–7 in seven industry-led use cases along the automotive value chain.

The project runs from 1 September 2026 to 31 August 2029.

Purpose

Objectives

  1. SO1

    Accelerate Engineering

    Develop AI-assisted methods, tools and toolchains that extend state-of-the-art CI/CT/CD pipelines, substantially increasing ECS engineering efficiency and shortening development cycles.

  2. SO2

    Master Complexity

    Integrate holistic ECS engineering principles into highly automated toolchains to manage and reduce the exploding complexity of modern automotive systems.

  3. SO3

    Open Platform

    Build a modular, open and extensible AI-assisted integrated platform that serves as a reference architecture for next-generation SDV engineering toolchains.

  4. SO4

    Showcase and Evaluate

    Demonstrate the benefits, performance improvements and industrial value of the AI4SDV tools and platform in realistic engineering settings.

  5. SO5

    Human-Centric Engineering

    Enable seamless and effective human–AI collaboration, supporting engineers in key SDV development use cases and enhancing innovation capacity.

  6. SO6

    Global Competitiveness

    Strengthen Europe’s position in SDV and ECS by providing cutting-edge engineering capabilities that boost productivity across the automotive value chain.

  7. SO7

    Maximise Exploitation and Alignment

    Align with European research initiatives, projects, associations and industrial players to accelerate adoption, avoid fragmentation and drive a common European SDV vision.

Research pillars

Where the work happens

  1. 01

    WP1 — Requirements & Architecture

    Lays the groundwork for the AI-assisted tools and services and for an open, extensible, standards-aligned platform architecture.

  2. 02

    WP2 — Methodologies

    Establishes the methodological and technological foundations for next-generation SDV engineering toolchains: AI-assisted engineering with generative, agentic and predictive AI, integration of formal and model-based methods, service-oriented toolchain architectures and workflow automation, including techniques against hallucination and for human feedback.

  3. 03

    WP3 — Implementation

    Implements the AI-assisted tools and services that support every SDV engineering phase (A–F).

  4. 04

    WP4 — Orchestration and Toolchain Integration

    Orchestrates heterogeneous development tools into a cohesive, traceable and repeatable workflow, validates their interoperability end-to-end in realistic industrial environments with real-world data, and provides the credibility evidence needed to use AI-assisted tools for safety-critical automotive software.

  5. 05

    WP5 — Verification and Validation in Use Cases

    Verifies and validates the AI4SDV tools against functional and non-functional requirements and KPIs by applying them in seven industry-led use cases, and turns these use cases into demonstrators that show reduced cycle times, lead times and defect rates.

  6. 06

    WP6 — Impact and Ecosystem Building

    Builds a sustainable ecosystem around the AI4SDV platform, its tool developers and users: collaboration and knowledge sharing with ECS engineering communities and open-source initiatives such as Eclipse, open standards and interfaces, and demonstrations released as open-source reference implementations.

  7. 07

    WP7 — Communication, Dissemination, Exploitation, Standardization

    Defines and implements the dissemination, communication and exploitation plan, establishes the AI4SDV brand, runs the project website as a knowledge hub for EU-funded SDV research, fosters clustering with related initiatives, supports standardisation and prepares a sustainable exploitation strategy beyond the project lifetime.

  8. 08

    WP8 — Project Management

    Ensures the successful execution of the project through professional coordination: the contractual link with the funding authority, risk and quality management, effective communication between partners and work packages, scientific coordination and data management guidelines.

Validation

7 demonstrators

Demonstrator 1

UC1 — Trustworthy OTA Testing and Validation Orchestration for SDV

AI-assisted architectural analysis and over-the-air deployment of software updates. New software versions are validated in silent mode without affecting vehicle behaviour, while AI agents analyse telemetry data to detect anomalies and optimise update strategies.

Led by Traton AB
Demonstrator 2

UC2 — AI-driven Continuous Evolution for Enhanced Perception in Autonomous SDV Operations

Continuous improvement of camera-based perception for on-road and off-highway vehicles. AI-assisted CI/CT/CD pipelines automate training, deployment and optimisation of perception models using continuously collected real-world data.

Led by AVL Araştırma ve Mühendislik
Demonstrator 3

UC3 — Secure Feedback-Driven Engineering

A closed-loop feedback system that derives requirements and validates features from real-world driver behaviour, combining contextual data, privacy-preserving edge AI and agentic AI to shorten iteration cycles.

Led by IAV GmbH Ingenieurgesellschaft Auto und Verkehr
Demonstrator 4

UC4 — SDV Augmented Vehicle-Human Interfaces

Integration of third-party software and AI-based components for multimodal user interaction (vision, speech, touch), with hardware abstraction, code optimisation and robust modular update mechanisms.

Led by Ampere Software Technology
Demonstrator 5

UC5 — Intelligent and Adaptive SDV Navigation and Positioning

AI-assisted tools for ADAS, automated parking and autonomous charging, improving requirement quality, formalising timing constraints and increasing test coverage using LLMs and neural-network compression.

Led by Valeo Comfort and Driving Assistance
Demonstrator 6

UC6 — Evaluation of SDV AI Tools for Lighting Lifecycle

Transfers the AI4SDV tools to a non-automotive domain – lighting electronics – to show that AI-assisted engineering accelerates innovation and toolchain efficiency across sectors.

Led by Signify Netherlands B.V.
Demonstrator 7

UC7 — Intelligent Configuration Assistant for SDV Middleware

AI-assisted middleware configuration with schema-aware validation, explainable AI feedback and historical data analysis, reducing complexity and preventing misconfigurations across OEMs and Tier-1 suppliers.

Led by Qorix GmbH