Interview with Laura Ruotsalainen

Role: Professor of Spatiotemporal Data Analysis for Sustainability Science at the Department of Computer Science at the University of Helsinki; Vice Chair of ELLIS Institute Finland

Areas of research: Machine Learning Algorithm Development using Spatiotemporal Data

Personal website: https://researchportal.helsinki.fi/en/persons/laura-ruotsalainen/

Featured in ‘Women in ELLIS” campaign: https://ellis.eu/news/spotlight-on-laura-ruotsalainen

Background & Professional Journey

Can you briefly describe your professional background and career path?

I am a Professor of Computer Science at the University of Helsinki, where I lead a research group on spatiotemporal data analysis. My work has evolved from computer vision and Bayesian state estimation for perception in dynamic environments toward reinforcement learning, representation learning, and generative, geometry-aware machine learning methods for visual and spatiotemporal understanding in real-world systems. This has led to a focus on methods that support decision-making under uncertainty, particularly in safety- and sustainability-critical domains. Alongside, I have worked closely with public and industry partners developing autonomous systems, which has grounded my research in robotics and safety-critical perception and control, while maintaining a primary focus on machine learning methodology.

What led you to your current field, and what motivates you?

I have been interested in mathematics and scientific problem-solving from an early age. I remember discussions in the 1990s about resource limitations and climate change, where fusion energy was seen as something that could essentially save the world by replacing fossil fuels. That shaped my view that science can tackle major global challenges. I was drawn to computer science because of how exciting and powerful it is as a field, and over time I realised it provides concrete tools to work on these challenges. When I later saw the opportunity to focus on machine learning for spatiotemporal, sustainability-critical real-world systems, it strongly resonated with this motivation and shaped my research direction.

What do you consider your most significant professional accomplishment to date, and why?

I would highlight the development of a multi-objective hierarchical reinforcement learning framework that integrates contextual information into decision-making. The significance is not only methodological, but conceptual: it addresses how AI systems can operate in complex environments where trade-offs, such as efficiency, sustainability, and livability, must be explicitly managed rather than implicitly ignored. This work builds on long-term efforts within my research group, where multiple researchers have contributed both to the methodological development and to creating more realistic simulation environments. It has also relied on close interdisciplinary collaboration to bring in domain knowledge and data from sustainability-related fields.

Research & Expertise

What are your primary research interests or areas of expertise, and what impact have they had on the wider field of Machine Learning?

My work focuses on three main areas: multi-objective and hierarchical reinforcement learning over spatiotemporal data, computer vision for dynamic environments with connections to robotics and autonomous systems, and representation learning for multidimensional spatiotemporal systems. The broader impact lies in pushing machine learning beyond single-objective optimization toward frameworks that can handle competing objectives, uncertainty, and long-term dynamics. This is particularly relevant for real-world deployment, where decisions are often safety-critical and have structural and societal consequences.

Are there any recent projects or publications you're particularly proud of? What was their significance?

A recent line of work develops contextual, multi-objective hierarchical reinforcement learning for urban systems, integrating traffic, air quality, and livability objectives. Its significance is in demonstrating how AI can be used not just to optimize isolated metrics, but to reason about trade-offs in complex, coupled systems, an area where current methods remain limited. In parallel, my research on computer vision for autonomous systems has addressed perception and uncertainty in dynamic environments. Both of these research directions have led to multiple publications and contribute to advancing machine learning methods for real-world, sustainable and safety-critical applications.

Where do you see your research or professional focus heading over the next few years?

I see the focus moving toward decision-making under deep uncertainty in dynamic, non-stationary environments. This includes better integration of learning, simulation, and real-world data, as well as methods that explicitly account for discontinuities, such as sudden societal or environmental changes. I also expect stronger connections between AI methods and regulatory or societal frameworks.

Engagement with the Community, Networks and Broader AI Ecosystem

Can you share an example of successfully aligning industry, research, and policy stakeholders, and what you learned from it?

In my work, we collaborate closely with industry partners, domain scientists, and public-sector stakeholders across multiple application areas, including autonomous systems, industrial processes, and sustainable urban environments. This includes long-term industry collaboration through research support and real-world, high-dimensional data, as well as my role in Finland’s scientific expert group on AI and data supporting the government’s work on the EU AI Act. The key learning is that alignment does not come from technical solutions alone, it requires shared problem formulations. Establishing a common language between disciplines is often the most critical step, but collaboration and hearing different viewpoints is rewarding.

Have you held leadership roles in other organisations? What did you learn from those experiences?

I lead a research group and contribute to national and European AI initiatives, including activities linked to ELLIS and FCAI. Before joining the university, I also held a leadership role at a governmental research organisation, which shaped how I work across research, industry, and policy. One thing I have seen repeatedly is that strong research alone is not enough, you also need the right infrastructure, people, and long-term collaboration to make it work.

Which communities, organisations, or ecosystems would you actively bring into Adra, and how would you mobilise them?

I would actively connect Adra with the ELLIS Society, Finnish AI ecosystem (including ELLIS Institute Finland and FCAI), and interdisciplinary partners in sustainability, and urban research. Mobilisation comes through joint research agendas, shared infrastructures, and coordinated funding initiatives.

What motivated you to apply to become a Board member of Adra?

Adra sits at a critical intersection of research, industry, and policy in Europe. For me, this is an opportunity to contribute to shaping AI that is both scientifically strong and aligned with societal needs, particularly in sustainability and responsible deployment, while strengthening core machine learning research in Europe and creating stronger links with the ELLIS network.