I-PAR-5: DIGIHALL platform for trustable-by-design AI-based system
"In this experiment, DIGIHALL supports Sherpa Engineering (https://www.sherpa-eng.com/) in the development of an intelligent control system of a simulated autonomous vehicle that must ensure correct driving along a closed circuit including activity zones and living areas open to pedestrian traffic, leading to important safety constraints. The goal of the experiment is to illustrate the offer of service on technical advising and feasibility assessment offered by DIGIHALL and to explore the functionality of a proof-of-concept technology stack for the development of software modules for safe, reliable and quality-assured autonomous transportation services. Built on top of open-source technologies, the DIGIHALL Platform for trustworthy AI-based systems enables (i) a low-code approach to the specification of vehicles' software architecture and behavior; (ii) the safety analysis of designed behavior against road scenarios and traffic rules; (iii) testing of the composition of perception, supervision and control componenents to ensure safety, reliability and quality of service of the design. The backbone of the experiment platform depends on open frameworks and technologies, such as the Eclipse-based Papyrus for Robotics low-code platform (https://eclipse.dev/papyrus/components/robotics/), the CARLA simulator (https://carla.org/) and the ROS 2 middleware (https://www.ros.org/)."
For SMEs, the experiment has significantly accelerated the design and development process, enhancing the quality and consistency of designs through an integrated environment. This was achieved by successfully linking each requirement specification, including safety, quality, and performance, with the simulation model. The experiment also effectively connected system definitions, use-case and scenario definitions, as well as linking scenario definitions with scenario execution and visualization.
One of the key outcomes was the ability to visualize operational scenarios directly. This visualization allowed for the verification of these scenarios and validation of the coherence of the objectives, facilitating a smooth transition between system definition and simulation models. This integration streamlined design-refinement cycles, ultimately leading to the creation of solutions that are trustworthy by design.
For DIGIHALL, the impact has been the consolidation and further development of the Papyrus for Robotics platform. This achievement has not only benefitted DIGIHALL but also supported other SMEs and companies in the domain of intelligent transportation systems. The platform has been particularly instrumental in advancing technologies like collaborating drones and ground vehicles for autonomous goods transport, marking a significant step forward in the field of intelligent transportation systems.