Automated Driving

Automated Driving and Beyond

The Future of Mobility is Automated

Automated driving is the decisive trend for the automotive industry. Technological progress is making it possible to create a new driving experience that holds manifold advantages for society and economy. A reduced number of accidents, enabling mobility for elderly and disabled people as well as reducing traffic jams and thus environmental pollution. This development of course also bears economic potential. On the one hand, the time previously spent with actively driving can be turned into additional time for productivity or leisure to recharge the passenger’s batteries for upcoming meetings or other activities. The overall global economic benefit of autonomous vehicles is estimated to amount to 3.5 billion $ until 2025.[1]

This potential is changing the face of an entire industry. Realizing safe automated driving systems is more than developing applications and algorithms – it is a venture that requires know-how in many areas. Strong technological partnerships and a rich ecosystem will be key for car manufacturers and OEMs when developing safe systems and staying at pace with the rapid technological innovation. These safe platforms orchestrate automated driving and act as the heart of a new generation of in-vehicle architectures. Creating these safe environments requires networking and safety know-how derived from experience in safety-critical applications.

[1] Source: Cars 2025: Vol.3 „Monetizing the rise of Autonomous Vehicles“ by Goldman Sachs Global Investment Research

Safety for complex automated systems

Automated driving leads to the modularization of automotive electronic systems and the implementation of new customer functions controlled by software only, leading to the emergence of central domain ECUs on the basis of platform architectures. Generic platforms are needed to foster function software reuse across different car models, which in turn is necessary to cope with the immense costs and efforts for function validation especially in the automated driving domain. Dramatically increased technical complexity and a new cooperation models creates new roles such as the software integrator and new requirements for the domain ECU architecture. Real-time capabilities, adherence to highest safety standards as well as dealing with the overall high complexity of a system containing numerous hardware and software components are such challenges in this area. A deterministic hardware and software architecture, including a Deterministic Ethernet switch, exploits the time-triggered paradigm for ECU-internal communication and task scheduling, thus enabling predictable timing behavior of application software components and fully integrated systems. A common, generic middleware layer abstracts the diversity of the underlying microcontrollers and operating systems for the application’s software components. This architectural basis is implemented with an extensive set of tools and services. The integration process enabled by the deterministic paradigm guarantees full predictability and composability and thus ensures smooth software integration without the time-consuming and costly integration hassle of traditional ad hoc approaches.

Introducing in-car compute

The next step in the evolution of automated driving will be level 4 and level 5 automated driving, meaning a semi- and fully autonomous operation of a vehicle. This will lead to further tremendous growth of software functionality. The associated compute requirements push the well-established domain architecture to its limits in terms of cost-efficiency. Having different ECUs with different compute platforms, different software stacks and tools, is a huge cost driver and severely limits software synergies as the boundaries between the domains are blurring at a rapid pace. Current technology already provides virtually unlimited scaling, highest availability and lowering costs. However, these commercial architectures are missing two key properties that are essential for automotive: safety levels up to ASIL D and real-time performance.

When realizing level 4 and 5 automated driving systems, it is time to shift compute architecture paradigms from a domain architecture which isolates compute resources and software functionalities within each to a coherent cross-domain architecture that builds on an “In-Car Compute Platform (ICCP)”. Centralized cross-domain architectures based on ICCPs contribute to cost-efficiency of the E/E system, are flexible in the deployment of new software and hardware solutions and are an effective way to realize fail-operational architectures for level 4 and 5 autonomous driving.