SMILE IV – safety analysis and verification/validation of machine learning based systems

Autonomous vehicles heavily rely on deep learning (DL) algorithms to function effectively. Nevertheless, it is still an open research question on how to implement DL methods in safety critical applications. The SMILE research programme has studied and developed methods and processes that allow DL-based functions to be included into safety critical applications. The last project, SMILE III, has developed an experimental pedestrian emergency braking ADAS (SMIRK) with the first full application of AMLAS safety assurance process.

SMILE IV is the next research exploration in the roadmap of the SMILE programme, leveraging the results of the previous projects SMILE I, II and III to further develop enabling technologies and implement safety assurance framework on the new types of transport services enabled by small autonomous vehicles: transport robots in factory settings and delivery robots in public spaces.

SMILE IV focuses on optimising and implementing the safety architecture and safety components for the two distinct use cases of small autonomous vehicles operating in different environments and conditions. SMILE IV research facilitates the adoption of near future transport and mobility services enabled by connected small autonomous vehicles. By challenging the approach with diverse use cases and requirements from different domains, the results from SMILE IV will be also of high interest for automotive and other industrial actors that plan to use DL to support autonomous functions in safety critical applications.