Imagine a self-driving car that can tell if a pedestrian is about to cross the street just by looking at where they are glancing. It's like giving a car a 'gut feeling' and a moral compass so it makes safe, predictable decisions. This system uses smart simulations to test every possible scary scenario before the car ever hits the real road.
Autonomous vehicles struggle to predict the erratic movements of pedestrians and cyclists, leading to safety risks and low public trust. Current AI models often lack transparency and fail to account for ethical dilemmas in urban traffic.
An interoperable digital framework for testing AI models and an online participatory space for citizen feedback. It includes 12 modeled traffic scenarios and a Trustworthy AI methodology.
If you are an AV developer dealing with unpredictable urban traffic—this project developed AI models for anticipating vulnerable road user behavior that improve road safety. By using visual gaze estimation, your vehicles can better predict if a cyclist will turn or stop.
If you are a testing firm dealing with the difficulty of simulating rare road accidents—this project developed metamorphic testing and generative adversarial networks to multiply simulation scenarios. This allows you to stress-test AI safety without needing millions of real-world miles.
If you are a city provider dealing with low public trust in automated shuttles—this project developed an online participatory space to identify AI risks and biases. This helps you build a roadmap for user acceptance among pedestrians and cyclists.
Based on available project data, no specific commercial pricing or licensing costs are mentioned; the project was funded by an EU contribution of EUR 5,965,630.
The project uses an interoperable digital framework and three complementary use cases to test the sense-plan-act cycle, suggesting a path toward industrial scaling.
Based on available project data, the specific IP and licensing terms are not disclosed, though it aims to create an open environment for integrating AI models.
The project leverages Trustworthy AI guidelines for general systems and ethics recommendations for connected automated vehicles to ensure compliance with safety and ethical standards.
The project period runs from 2023-01-01 to 2026-03-31, with the final version of the digital framework expected by the end of the project.
The consortium is well-balanced for commercialization, featuring 15 partners across 9 countries. With a 40% industry ratio (6 industrial partners, including 5 SMEs), there is a strong link between the research and market application, supported by 6 research organizations and 1 university.
Contact SIMULA RESEARCH LABORATORY AS in Norway
Contact us to explore the AI4CCAM digital framework for your AV safety pipeline.
