ECSEL-prosjekt ArchitectECA2030, Trustable architectures with acceptable residual risk for the electric, connected and automated cars

Visjonen til prosjektet ArchitectECA2030 er å etablere et harmonisert pan-europeisk rammeverk som muliggjør validering av elektroniske komponenter og systemer (EKS) for elektriske, tilkoblede og automatiserte (ECA-Electric, Connected and Automated) kjøretøy for å forbedre pålitelighet, robusthet, sikkerhet og sporbarhet. ArchitectECA2030 rammeverket dekker både metodene, verktøyene, prosessene og dataene som er nødvendige for å standardisere og periodisk re-sertifisere elektroniske komponenter og systemer for elektriske, tilkoblede og automatiserte kjøretøy. Prosjektet bidrar til å etablere en åpen og transparent metodikk som reduserer utviklings-, verifiserings- og valideringstider betydelig, samtidig som høyeste nivå med hensyn til sikkerhet garanteres, og dermed tilliten fra brukere/kunder så vel som produsenter gjennom hele kjøretøyets levetid. ECA kjøretøy bidrar til å redusere energiforbruk, øke trafikksikkerheten, samt bedre trafikkflyten, og dermed også redusere CO2-utslipp. For å implementere pålitelige arkitekturer med akseptabel restrisiko for elektriske, tilkoblede og automatiserte kjøretøy, har ArchitectECA2030 definert fem overordnede mål: > Implementere en kontinuerlig robust designoptimalisering for hver del i EKS-verdikjeden. > Etablere et rammeverk for sikkerhetsvalidering av EKS-verdikjeden. > Identifisering og styring av gjenværende risiko (restrisiko) i hele EKS-verdikjeden > Sikre sluttbrukers aksept ved pålitelighet i EKS-verdikjeden. > Støtte visjonen om null utslipp, null trafikkulykker og null trafikkaos gjennom ECA2030 kjøretøyer. SINTEF, NxTech og TracSense har samarbeidet i prosjektet for å implementere en sensorfusjonsplattform for overvåkning av veitilstanden som kombinerer sensorer for friksjonsdeteksjon (radar, optisk, fuktighet og temperatur), 3-akser vibrasjon, et kamera, en KI-basert prosesseringsenhet og en kjøretøy-til-alt (V2X) innebygd enhet for kommunikasjon med andre kjøretøy og infrastruktur.

Autonomous vehicles (AVs) are designed to operate without human intervention by navigating and sensing their environment through technologies like radar, LiDAR, GPS, odometry, and machine vision. Autonomous vehicles can potentially reduce road accidents and improve overall road safety significantly. However, they are not without risks; residual risk refers to the risks that remain even after extensive testing and safety measures have been implemented. As the vehicles rely on various sensors, hardware, software, and artificial intelligence (AI) based algorithms to navigate and make decisions, the technologies may have limitations, such as poor performance in adverse weather conditions (e.g., heavy/light rain, snow, ice, etc.) or difficulty in recognizing uncommon road hazards. Hardware or software failures can occur even with redundant systems and safety checks, leading to unexpected behaviour or accidents. Residual risk is associated with the possibility of such failures. As autonomous vehicle technology continues to evolve, the residual risk is expected to decrease over time as safety measures are enhanced and the technology becomes more robust. Achieving full autonomy without any residual risk is a challenging task, and it may only partially be wholly eliminated, given the complex and dynamic nature of the real-world driving environment. As a result, the vehicles must adapt their Operational Design Domain (ODD) to the different driving scenarios. ODD refers to the specific conditions and environments under which a self-driving vehicle is designed and intended to operate safely. It defines the boundaries and limitations of an AV's operational capabilities. It considers various factors, including environmental conditions, road types, traffic conditions, speed range, operational constraints, operation time, and emergency scenarios. Monitoring road conditions is an ongoing challenge for autonomous vehicle developers. The goal is to ensure that the vehicles can safely operate in various environments, conditions, and road types. Continuous advancements in sensor technology, data processing, AI, and machine learning (ML) are helping autonomous vehicles become more capable and adaptable to varying road conditions. The project's benefits are a successful demonstration of integrating the road condition monitoring system with the V2X communication modules and conceptualising a residual risk assessment and reduction framework to provide reliable and robust connectivity systems.

Independent validation is critical to define the capability and safety of any solution in the electric, connected and automated (ECA) vehicles space. Appropriate and audited testing needs to be performed in a controlled environment before any deployment takes place. As the software and hardware components come from multiple vendors and integrate in numerous ways, the various levels of validation required must be fully understood and the integration with primary and secondary parts must be considered. The key targets of ArchitectECA2030 are: ? Robust mission-validated traceable design of electronic components and systems (ECS) ? Quantification of an accepted residual risk of ECS for ECA vehicles to enable type approval, and ? Increased end-user acceptance due to more reliable and robust ECS. The methods include automatic built-in safety measures in the electronic circuit design, accelerated testing, residual risk quantification, virtual validation, and multi-physical and stochastic simulations. A validation framework comprised of harmonized methods and tools able to handle quantification of residual risks using data different sources (e.g. monitoring devices, sensor/actuators, fleet observations) is provided to ultimately design safe, secure, and reliable ECA vehicles with a well-defined, quantified, and acceptable residual risk across all ECS levels. The Norwegian consortium is addressing the architecture and design of components/systems (e.g. perception camera and connectivity) for reliable and safe automated driving. The work focuses on contributing to the development of a framework for the V&V of automated vehicles with Level 3/4/5 capabilities covering extensive risk analysis of pilot data, quantification and reduction of residual risks w.r.t. the degree of automation and the driving environment. Standardized criteria for the evaluation of the completeness of test programs for automated vehicle operation (e.g. coverage metrics) will be evaluated and optimized.