Ahead of IS Auto Europe 2019 we spoke to Prof. Dr. Wolfgang Koch, Head of Department Sensor Data and Information Fusion, about his presentation on 'AI-enhanced perception and action: cross-fertilization between automotive and defence autonomy’, challenges in the marketplace and future vision system developments and challenges.
You are going to talking about AI-enhanced perception and action: cross-fertilization between automotive and defence autonomy–what are some of the points that you are going to explore?
Essentially, AI technology is the automation and enhancement of human perception to the end of acting more appropriately in a challenging situation. In an improper sense, we may thus speak of "intelligence" and "autonomy". In this context, we should consider a continuous spectrum of levels of automation for perception and action. From a computer science perspective, automotive applications can benefit greatly from latest developments in defence. Three points are covered: comprehensive fusion of multiple sensor data, efficient management of sensor and actuator resources, meaningful human control.
What are the biggest challenges facing the automotive sensor industry right now and how to do you think this conference can address them?
From the presenter’s perspective, the main challenges are essentially the same as in defence applications where the technological development is more advanced. In particular, a key issue is rule-compliance by design, i.e. technical support for responsible use according to legally binding rules-of-engagement that clarify the international law in a given mission. The approval and certification of defence systems is only possible if human responsibility is guaranteed. This will be even more the case for automotive applications. The automotive industry must therefore focus on technically supporting the principle of discrimination, i.e. reliable perception based on sensor data of appropriate quality, the principle of proportionality, i.e. appropriate measures based on appropriate situational awareness, the precautionary principle, i.e. the priority of a pessimistic over an optimistic forecast, and finally the principle of accountability, which presupposes meaningful human control.
What do you see as the most significant changes coming up in vision systems development for automotive in the next 12-24 months?
The presenter believes multiple radar systems to be the backbone in automotive sensor systems also in the foreseeable future. Vision systems will have supportive functions. The big challenge and hopefully also change in the next 12-24 month will a substantially increased quality of the data produced by low-cost sensors. Data fusion can only in a limited sense recover what is lost at the sensor level. Moreover, continuous integrity checks of sensor and geodata are inevitable since malicious intervention must be taken into account.
What are you most looking for to about attending IS Auto Europe 2019?
The presentations delivered on events like IS Auto Europe will stimulate fruitful technical discussions. I am really looking forward to them.