My vision as full professor

Tackling societal challenges with algorithmics

I feel honoured, appreciated, proud and above all happy with my appointment as full professor on Algorithmics for Planning and Scheduling at TU Delft. A big thank you to all who I have worked with and who have supported me! I’d like to use this occasion to share some insight into my vision and plans for the future.

On the one hand not much will change: I’ve been chairing the section on Algorithmics within the Department of Software Technology already for four years. But on the other hand, the process towards this promotion and the people involved have challenged me to formulate my vision more clearly, and better think through my longer-term plans.

Vision

Developments in computer science and artificial intelligence research have been stormy these last years: break-through results such as a computer winning from the leading Go-champion in 2015 provide evidence that these recent inventions have the potential to significantly improve strategic decision making and reduce costs in complex operations, also beyond games. I’m convinced that algorithmic insights can contribute to solving planning and scheduling problems across all disciplines, as well as contribute to addressing societal challenges. Moreover, this has important consequences for fundamental research in algorithms. I’ll illustrate this with two domains that are close to my heart: rail transport and the energy transition.

Rail transport

For example, the creation of schedules for train servicing is become more and more challenging with an increasing number of trains without an increase in rail capacity. Moving trains around a shunting yard is becoming like a sliding puzzle with just a single free space. Reasoning about how to use this space efficiently can be supported with computation, allowing more trains on a shunting yard, and thus more effective use of the available capacity. 

Energy transition

Similarly, the energy transition involves a change from fossil fuels to electricity, and therefore a much larger demand on the electricity infrastructure. Also here, this demand can be serviced in a cost-effective manner by making use of the already available capacity in a smarter and more dynamic way: by a combination of prediction and scheduling algorithms (in a suitable market), deciding when which device is allowed to use the available capacity.

Better understanding effect of modelling decisions on computation

In the other direction, bringing recent algorithmic developments into practice brings new, exciting research challenges. We need efficient methods to reason about the involved uncertainty in predictions and their consequences, and communicate the involved risks to the human decision makers. We need methods that can cope with situations where multiple stakeholders interact. But most importantly, we need a better understanding of how to model these practical planning and scheduling problems such that reasonable answers and support can be computed within the time available.

Multidisciplinary collaborations

I plan to work on these challenges through research projects together with experts from the domain and researchers in the respective disciplines. On the other hand, I work with my colleagues in AI and of course also within the section Algorithmics at TU Delft on understanding the fundamental limitations. 

Beyond my direct collaborations, I aim to engage other AI and algorithmics researchers in these challenges, e.g. by providing benchmarks and prepared data sets. And by supporting others in these three-way matches of experts (often from industry), researchers in the respective field, and AI researchers, e.g. through the NLAIC.

The importance of educating AI 

Last but not least, education plays an important role in bringing algorithmics insights into practice. We need more people with understanding of algorithms and AI. So we should offer algorithms & AI courses to all students who want, including PhD students. Also we should provide Computer Science students with some training in modelling and working with other disciplines, and of course with a solid, scientific basis in AI and algorithms.

Join me!

If you’re working on a planning or scheduling problem (especially of course if this is in a context I’ve started to understand: rail or energy), recognize the discussed challenges, and are interested in the potential of algorithms for planning and scheduling, please reach out! 

I’ll update you on my progress towards my vision and my plans in my inaugural address on June 7, 2023, 15:00 at TU Delft. If you’re interested, you’re welcome to join. Please register here.

I now feel more connected to TU Delft than ever, and I look forward to working with my colleagues on the challenges the world is facing in the coming decades.