Hybrid systems for autonomous driving
Background
Our secondary technology is based on architectural designs to guarantee safety in autonomous driving (AD) within the ethical and legal context. Our independent decision and validation architectures constitute critical hybrid systems with AI components forming a part of the overall vehicle system.
We focus on the development of hardware and software architectures that efficiently carry out rule-based and other actions based on the existing sensors and computing resources. For this reason, we also implement these ideas on physical camera, radar, lidar, simulation and EMU systems.
The implementation of the required features is only possible through a redundancy of independent and differently prioritized system components that mutually monitor each other. They are complemented by legally and technically required safety components like hybrid Machine Learning procedures (bottom up/top down), preset a priori databases, scenario catalogs, efficient pipelining technologies for semantic segmentation and data fusion including V2X. All these components are considered in the architecture.
In autonomous driving, we closely collaborate with OEM partners, suppliers, expert commissions, trade associations and politics. Within the framework of our long-term partnership with the SOCAI of the Würzburg University, we do interdisciplinary work on machine normativity in autonomous driving.
Motivation
Nine of ten fatalities in motorized traffic can be attributed to human error. Furthermore, many vehicles are just sitting idle for 23 hours per day, and we encounter increasing problems with the number of cars in urban areas and their sustainability.
At the same time, our demand for mobility is increasing due to globalization. Aren’t these valid reasons to bring our kind of mobility, which has grown evolutionarily for a century, into a more sustainably shape by the prudent use of disruptive technologies?
Outlook
Thanks to our advanced architectural concepts, we were able to position ourselves on the autonomous driving market early on. Together with our partners, we want to actively help shape progressive AD standards in line with our social values. In future, we do not only strive to license architecture IPs with local, volatile software but also to deliver specific system components.
The discussion of the ethical and legal issues of autonomous driving is still at an early stage. The special field of ethics by design gets more and more into focus, not least due to the establishment of the AD ethics committee in Germany a few years ago and the increasing initiatives of OEMs, industry associations and the research community. First standardization efforts like the 2019 policy paper of large German automotive companies underline the high relevance but particularly point out the broad scope of possible designs.