The race towards autonomous driving has lasted for years. With demonstrated success and audacious claims from leaders, the pressure on followers has drastically increased. I argue that the chance for the chasing contestants to close the gap lies in scientifically rooted engineering methods that can efficiently improve the quality of AI, in particular on safety. We need improvements in data collection, training, testing, and finally, techniques for detecting black swans during operation and fleet testing. Some of my conducted prior work will be concisely mentioned with an honest statement on their limitations. I will also cover recent efforts in standardizing autonomous driving safety. I conclude this talk with my personal view on opportunities in Taiwan to be included in the ecosystem of safe autonomous driving. 

Currently, Chih-Hong Cheng is a technical manager at DENSO, researching safety technologies for autonomous driving. He is also an adjunct faculty (unpaid/voluntary) at Academia Sinica, the most prestigious academic research institute in Taiwan.

Previously, Chih-Hong was a permanent researcher at fortiss - Research Institute of the Free State of Bavaria, where he developed the research topic of dependable AI for autonomous systems. Before his tenure at fortiss, he was a scientist in ABB Corporate Research Germany, where he worked on projects related to intelligent production systems (Industry 4.0), cloud-related technologies for industrial automation (PaaS, IaaS), and the analysis of complex industrial software systems. Chih-Hong received his doctoral degree with distinction in CS from the Technical University of Munich.