Connected automated vehicles (CAV) are expected to provide more efficient, accessible and safer transport solutions. But the complexity of the systems is a challenge. The project's goals were new methods for safety assurance of CAVs that enable faster development with continuous deployment of safety-critical systems.
The automotive industry is currently undergoing rapid changes where increasingly advanced functions for driver support or unsupervised operation ("self-driving vehicles") are being developed in order to provide safer, more efficient and more accessible transport solutions. But since such functions must be able to interpret their surroundings and make the right decision in every situation that arises, it is a challenge to ensure that the product is both safe and provides good performance, for example that it can handle different traffic or environmental conditions. In order to provide the best customer value throughout the product's lifetime, it must be possible to update it to gradually improve performance as we learn more, for example how to operate safely under different environmental conditions, or if for example traffic regulations are changed.
Agile development is used in many domains to enable regular updates, but for safety-critical systems, a problem is that the safety argumentation must be complete and consistent for each update made. The project's goal was therefore to develop methods for safety assurance that fit into an iterative development process with continuous deployment. The methods should suit development of advanced features for different types of vehicles. The project involved partners who develop functions for both road vehicles and mining vehicles. Iterative development enables new automated functions to be more easily introduced to the market with initially more limited use cases, followed by gradual development of performance and number of use cases.
The project targeted methods in areas such as continuous safety assurance, human/machine interaction, safety concepts that are suitable for continuous deployment and that also include machine learning components, and handling of product variability with formal methods. Expected positive effects were contributions to state-of-the-art for CAV safety assurance, and enhanced development processes that strengthen the competitiveness of OEMs, subcontractors, and service providers who develop CAV features.
Coordinator, Project manager
Agreat, Comentor, Epiroc, KTH, Qamcom, Semcon, Veoneer, Zenseact