14th European Dependable Computing Conference
10-14 September 2018
Iasi, Romania

The 21st century society heavily relies on computers and networks for a multitude of tasks, from life critical applications, like autonomous vehicles, to the internet of things and social media. Complex integrated circuits (IC) are the foundation of the infrastructure on which all these applications run. The impact of soft errors - the errors induced by energetic particles like neutrons and protons - is a serious threat that both researchers and developers must deal with.
This talk begins with the fundamentals of soft errors. Some of the main factors influencing the soft errors rates (SER) are the semiconductor manufacturing process, technology scaling, and circuit design. Measurement and modeling techniques currently employed in computing industry for deriving SER are presented. Error injection is frequently used to find the most sensitive logic in a complex IC. Simulated error injection, employed to study the impact of soft errors on an AMD processor module is presented. Computer Aided Design (CAD) tools, are specifically developed to estimate SER of modern processors, in the early stages of the design process. The functionality of a SER CAD tool is briefly discussed. The importance of data-based decisions, in the context of competing design priorities, is underscored. The presentation concludes with high level guidelines for development of IC resilient to soft errors.

Vehicle and passenger safety is a critical issue for public authorities as well as for car manufacturers and suppliers. Besides drivers' training and information many technological advances related with passive safety and now active safety have been performed since a couple of years and have drastically improved the road safety.
Despite the development of more and more sophisticated Advanced Driver Assistance Systems (ADAS) and partially automated vehicles that substitute the driver in specific or critical situations, the driver remains a key factor in the driving or monitoring tasks. The deployment of ADAS and partially automated vehicles will really be efficient and accepted since it will fit with driver's needs, aptitudes and capacities to deal with complex environmental contexts, but also when providing an adapted assistance through a human centered design.
The most recent development related with ADAS and vehicle automation demonstrated the need for integrating at the early stages of the system design the interactions with the driver dealing with task sharing, and the degree of freedom, authority, level of automation, prioritisation of information and management of the various systems. They should provide the drivers with adapted assistance, to their current situation or state, to the vehicle or environmental situation. They should use multimodal interfaces appearing as much transparent as possible to the users, depending on the context and the task to be performed. Last and not least they should create an emphatic relation between the system and the driver in order to encourage the usage of the system in a permanent way while giving to the driver the possibility to configure the assistance systems in relation with his understanding and capacity.

The rules of the road today are all focused around one key element: drivers. Licensing, insurance, traffic laws - everything assumes vehicles are operated under the control of a human. For driverless vehicles, this presents a dilemma: How can you tell which car is at fault in an accident? More importantly: How can self-driving and human-driven cars co-exist safely?
There is little argument that machines will be better drivers than humans. Yet there is very real risk that self-driving vehicles will never realize their life-saving potential if we can't agree on standards for safety. Intel has proposed a model we call Responsibility Sensitive Safety (RSS), offering a safe and scalable approach to consider. We believe that it is important for the automotive industry to collaboratively establish a methodology and standard for safety validation in partnership with global standards-bodies and regulators. In this session we will dive into RSS, a formal, mathematical model for ensuring that a self-driving vehicle operates in a responsible manner under different conditions. We will explain how RSS provides specific and measurable parameters for the human concepts of responsibility and caution and defines a safe state, where the autonomous vehicle cannot cause an accident, no matter what action is taken by other vehicles. We will also talk about how the industry can collaborate to help put these types of safety standards in place to ensure the greater acceptance and deployment of autonomous vehicles.