Safety assurance of cooperative, connected and automated mobility (CCAM) systems is a crucial factor for their successful adoption and deployment in society. But the wide variety of individual initiatives in this field tends to silo solutions and hamper the large-scale and safe introduction of CCAM systems in our society. It is for these reasons that the SUNRISE project has been initiated. Halfway through this project, a draft version of the safety assurance framework has recently been completed, on which international vehicle safety bodies and individual CCAM experts are now invited to provide their feedback.

Now that we have removed the human from the vehicle has the AV become safer in traffic? No. We have removed the human and not the human element. But wait, is removing the human the solution? No. In fact the opposite is true, more the merrier. In Vivetha's multi-pillared approach to continuously improve and deploy safe and trustworthy autonomous vehicles, 'inclusion is key'. Data, verification and personas need to be weighed equally. Humans in our different roles (as engineers, as drivers, as traffic inspectors etc.) are to be modelled using AI in order to understand and prove safety of AV.

Ensuring automotive safety requires a shift from standard type-approval tests to comprehensive evaluations. This study compares traditional tests with real-world assessments by the Joint Research Centre (JRC), revealing ADAS/AD function limitations. It promotes defining Operating Design Domains (ODDs) within which the functions and systems are supposed to reach the minimum requirements. The type approval checks should consist of semi-random testing within this ODD. A proposed "robustness index" enhances reliability across diverse conditions. Market surveillance gains importance for independent testing, needing tailored protocols. This integrated approach aligns regulatory compliance with real-world safety, ensuring vehicles meet stringent standards beyond laboratory tests.

The concept of operational design domain (ODD) has been discussed in research and industry for a number of years but real-world applications beyond testing are not yet on the road. Regulations will demand a comprehensive ODD description for type approval as well as service area approval. Today’s ODD definitions are very generic and deliver only a few insights about the boundary conditions. This leads to additional efforts for explanation and showing (scenario-based) test results. Additionally, the argumentation of the completeness of the tests gets complicated. Having a technically precise but still human and machine-readable ODD definition could bridge this gap.

Safe behavior of autonomous vehicles needs to be assured before being deployed on public roads and maintained safely during the operation phase. This presentation will delve into the following key aspects: 1. Autonomous Vehicle Safety Assurance and SafetyOps approach 2. Challenges toward Rapid SafetyOps, exemplified by System Theoretic Process Analysis (STPA) 3. Generative AI and Large Language Models as a Potential Solution

The Centre for Connected and Autonomous Vehicles (CCAV) is a joint unit of the Department for Business and Trade (DBT) and Department for Transport (DfT) unit. This presentation will provide an update on UK government activities on its £100 million, commercialising connected and automated mobility (CAM) programme from 2022-25 and the UK’s future plans for CAM innovation, commercialisation, regulation and legislation. In November 2023, the UK government committed up to £150 million funding between 2025-30 for CAM R&D and early commercialisation of these technologies, whilst also ensuring these are safe and secure for all.

Deploying safe autonomous trucks requires a deep understanding of real-world driving. A key part of this process is the collection and analysis of traffic data to derive requirements, test cases and statistical distributions for safety validation. In this joint presentation, we give insights into DeepScenario’s collaboration with Torc Robotics to accelerate the safe deployment of autonomous trucks. At the core of the solution are stationary cameras combined with DeepScenario’s AI software, providing Torc with unparalleled data collection capabilities at critical locations. This allows Torc to significantly increase the efficiency of scenario mining and advance the safety of its autonomous trucks.

Technical regulations and recommendations do not yet fully account for the safety-related availability of power supply systems. Specifically, the application of ISO 26262 on the power supply of autonomous vehicles is troublesome. Concurrently, engineers have gathered to write a recommendation (VDA 450) that facilitates the application of safe electrical power supply to other systems with safety-related availability like autonomous vehicles. This presentation highlights the framework of the VDA 450 recommendation and its principles that comply with ISO 26262 and help to ensure safe power supply for autonomous vehicles.

UTAC is a worldwide reference in testing and type approval and an accredited lab for Euro NCAP tests and for type approval tests in France. The presentation will review today's regulation and testing capacities proposals and new challenges including Level 1-2-3-4-5 regulations status for M1 vehicles, shuttles, buses, delivery robots, SW/OTA, cybersecurity, AI and virtual type approval, and today's needs/solutions for testing and validation: physical, new protocols, new targets, rain and fog testing, digital and VIL testing, AI, V2X, cybersecurity, etc.

The presentation will focus on developing comprehensive safety metrics for real-time monitoring of connected and automated vehicles (CAVs). This monitoring ensures the operational safety of CAV systems and infrastructure in automated driving, remote operation and advanced driver assistance system (ADAS) use cases. These metrics align with international safety assurance frameworks and standards for global relevance. Emphasis will be on detailed assessment methodologies to evaluate vehicle system and infrastructure performance. This critical approach feeds into the validation and verification of CAV safety arguments, facilitating safe and secure deployment.

Germany has recently adopted one of the most innovative legal frameworks on self-driving vehicles in the world. Yet the new law also requires compliance with extensive requirements regarding vehicle design that manufacturers, suppliers and the engineering community must be aware of. The presentation will place particular emphasis on these technical challenges. For instance, autonomous vehicles must be equipped with systems for mastering dilemma situations that prioritize the interests at stake in the event of an unavoidable accident. Manufacturers also have to ensure safety against cyberattacks throughout the entire vehicle lifecycle. Non-compliance bears significant risks of civil or even criminal liability.

Modern vehicles are no less than 'smartphones on wheels'. The huge growth in in-vehicle functionalities demands increased and continuous connectivity while regulatory requirements mandate extensive testing and cybersecurity. The situation is intensified by ecosystem partners ranging from 'steel to software' and 'simulation to street'. As the cost ratio of software to whole vehicle rises, efficient 'software quality' becomes crucial to manage costs, win market share and achieve early compliance. The presentation proposes a process- and product-based software quality framework in an autonomous and V2X environment through qualitative measurement of customizable KPIs through assessments and tests.

The presentation will take you on a technical journey through some of the challenges of adapting a general-purpose Linux OS to an embedded automotive use case with strict regulatory and safety requirements. Extra seconds of boot time are hardly noticeable or impactful for servers with hundreds of days of uptime, but for a platform rebooted many times in a day and that must provide user interaction in seconds from power on for safety reasons, suddenly every CPU cycle is under the microscope.

The presentation will cover common software architecture for different autonomy levels, how the development and deployment of products at different autonomy levels benefit each other, and examples and lessons learned from Plus’s deployments in Europe, the US and Australia.

Developing AVs is a time-intensive and complex process that requires best-in-class data and AI training infrastructure. Companies developing software-defined vehicles need to accelerate time-to-market and minimize costs without sacrificing safety. Combining vehicle sensors, map data, telematics and navigation guidance using machine learning and data fusion techniques, data-driven development is not without its challenges. One of the biggest challenges is data collection and integrity, as data needs to be collected accurately and consistently to drive accurate decisions.

The presentation will discuss the roadmap toward software-defined vehicles. What does a software-defined connected and autonomous vehicle look like and what are the requirements to design and deploy those vehicles? It will also cover the requirements and design considerations that are taken into account for vehicle architecture and various vehicle components. The change in approaches to the development, testing and production of the vehicles will be discussed. The drastic migration toward utilizing new technologies for the service and maintenance phase planned for such vehicles will be discussed.

The presentation will show how vehicle motion management, a software system solution from Bosch for the complete vehicle motion domain, helps to drive the transformation toward software-defined mobility through multi-actuator control, a modular and scalable software architecture that simplifies the separation of hardware and software, centralized computing and standardized interfaces.

The session focuses on the application of Generative AI for the testing and validation of ADAS/AD use cases. We will look at how OEMs and Tier 1s are using Large language models and enriching computer vision models to build , test and develop perception models quicker. Finally we will delve in the new Microsoft Azure AI building blocks that enhance training and inference of these AI models.

The EU Parliament adopted the AI Act: the world’s first legal framework for AI. Stakeholders in the sector of autonomous driving technologies should familiarise themselves with the provisions of the new law and check if and to what extent the legal framework is relevant to their business model.

The concept of a software-defined vehicle (SDV) has become a hot topic across the industry as automakers look for ways to address the dramatically growing complexity of developing and deploying software while simultaneously building the foundation for self-driving vehicles. Automotive Grade Linux (AGL), an open source software platform for connected car technology, has been working on software-defined vehicles for the past eight years. Dan Cauchy, Executive Director of AGL, will discuss the current state of SDVs and the work being done by automakers and Tier 1s as part of the AGL SDV Expert Group. He’ll also provide insight into the driving trends behind SDVs and enabling technologies including virtualization, containers, and the cloud.

Most computational capabilities that will power vehicles starting production in 4-5 years have already been designed in. For a fast-moving software function such as automated driving, this is massive challenge. Thus, in the absence of a crystal ball, Arm builds on close collaborations with leading automotive OEMs and suppliers to define next-generation compute platforms. This presentation will discuss their challenges and how upcoming technologies will help enable more efficient ADAS/AD software innovation.

The commercialization of AVs in San Francisco defined a new phase, as the maturity of technology is now proven at small scale. But scaling AVs requires an overall understanding of the occurrence of errors and their impact on decision making or misbehavior, as faults and failures can lead to a decrease in safety and confidence in AVs. It underlines the relevance of adopting the methods of explainable AI and applying conformity to the upcoming AI regulation. Transparency in self-driving systems ML/DL black boxes is essential for the operation and therefore the scalability of AI-enabled self-driving systems.