Is the embedded industry driven today more by regulatory requirements or do the people responsible now recognize the need to act sustainably?
Prof. Sikora: I believe it is actually both. There is now a broad awareness of the need to do what is ecologically feasible within the bounds of economic possibilities - even if this is not yet the case everywhere in the world. One must also accept that not all lip service automatically leads to more sustainable use of resources. But the regulators – especially those from the EU - are doing their best in this area to increase the pressure. So it's coming from both sides.
Are there any on-going projects that already take into account the sustainability of the embedded system during development?
Prof. Sikora: From my point of view, the most striking examples are currently energy-autonomous systems. In other words, systems that draw their energy from the environment via energy converters (e.g. from pressure, light, temperature) and thus supply themselves. For me, this is a perfect example of a sustainable embedded system that can actually be operated indefinitely with no waste of resources.
Embedded systems support sustainability and efficiency
Where do embedded systems already help to save resources in everyday life?
Prof. Sikora: This is the second dimension in which embedded systems can be used to support sustainability and the efficiency of the systems above them. The importance of embedded systems for sustainability can hardly be overestimated. With intelligent, controlled, adaptive systems, one can significantly improve the efficiency of an overall system. An old example is the energy efficiency of the internal combustion engine as applied in today’s vehicles, which could not be achieved at all without all the microelectronics. And of course, and even more appealing: all the control electronics and communications technology in an electric car.
But of course all of these are also control mechanisms relating to environmental technology, power plant technology, smart city, smart home, smart grid, smart energy, smart farming, etc., etc. These are areas in which embedded systems are an indispensable element in making overall systems more efficient and sustainable.
Another example comes to mind where an embedded system played a key role in a field demonstration. There is a company that is developing smart rat traps. Until now, whenever rat infestation is detected, they simply put large amounts of rat poison into the sewer system to drive the rats away. That too, can be intelligently controlled to reduce environmental impact: By setting up sensor-based rat traps and then only putting in as much rat poison as is actually needed. If think about it, you'll find thousands of super-exciting examples.
Which slots in the conference program focus specifically on the topic of sustainability?
Prof. Sikora: We have included a session titled 'Sustainability' in Track 1 'IoT', which will deal with sustainability challenges by referring to use cases based on the topics of Repairability and Renewability. To complement this, we are planning a lively panel discussion on 'Sustainability and IoT'. Sustainability aspects are also to be found in other topics, such as optimizing the energy efficiency of motor control systems, or in update topics in the program that contribute to extending the service life of devices.
Responsible AI and the accountability of the embedded industry
In addition to 'sustainable', the word 'responsible' is also part of the claim of the embedded world Exhibition&Conference 2023. How does the embedded community deal with its responsibility?
Prof. Sikora: I think the biggest issue with regard to responsibility in embedded systems is the inclusion of artificial intelligence. With AI we are entering a completely new level of complexity and adaptability. Currently, there are many dependable systems and subsystems that have become dependent on each other through networking, but where individual stakeholders no longer fully grasp the complexity of the overall system. We are consequently already at a stage today where we have to accept the fact that these systems work without us completely understanding their overall complexity. This is based on the assumption that the combination of stable systems as an overall system is also stable. Which is not necessarily the case ...
This level of complexity will increase even further when we include AI - meaning adaptive systems which learn by themselves and adapt themselves. They adapt in ways which were unforeseen at the time of their development regarding how a system might behave in the future. Accountability plays a central role when it comes to ensuring the reliability of such systems.
Responsible AI has at least one more dimension: with intelligent systems, we very quickly approach areas that can be considered ethically questionable. Prime examples of this would be surveillance systems of the kind often used in dictatorships. However, if we are honest, we can already do many ethically reprehensible things with embedded systems. AI is thus 'only' an extension of this, but its effectiveness makes it potentially even more critical.
Is the embedded community aware of this responsibility and is it addressing it?
Prof. Sikora: I have to answer that in a differentiated way ... There are many serious activities connected to responsible AI that are making great progress. At the embedded world Conference 2023 we will have the chairman of the IEEE 7000 working group that is looking at this. Here, a clear trend towards certification is emerging. There are also working groups at the national and European level that are preparing regulations. This was also an issue at embedded world 2022.And at the embedded world Conference 2023, Prof. Ali Hessami, the chair of the P7000 working group at IEEE, will deliver one of the keynotes.
But of course there are certain differences worldwide. In regulated democracies and autocracies, embedded systems are sometimes deliberately used in combination with AI to explore and transcend ethical boundaries.
Do companies consider responsibility from the very beginning of a development project, or is it just a side issue?
Prof. Sikora: There are many organizations that consider the issue of accountability very early in their development process. For example, in the case of image processing, where personal data is deleted directly at a very early stage and only the data actually required for analysis is used. But as mentioned above, there are certainly differences here worldwide.
What discussion initiatives will there be on this topic at the embedded world Conference 2023?
Prof. Sikora: We held our first panel discussion on the topic of 'Responsible AI' along with VDE at embedded world 2022. We are now planning a follow-up discussion at the embedded world Conference 2023 at which we want to move forward on more concrete terms.
