Dr Martin Tomitsch is a Professor in Interaction Design at the University of Sydney School of Architecture, Design and Planning, where he leads the Urban Interfaces Lab and is a member of the Smart Urbanism Lab. His research investigates new ways for designing the interaction between people and technologies with a particular focus on cities. He is the lead investigator for the research project “Trust and Safety in Autonomous Mobility Systems: A Human-centred Approach” funded by the Australian Research Council.
Dr Martin Tomitsch will give a keynote speech on 'Trust and Safety in Autonomous Mobility Systems: a Human-centered Approach' at the FIA Smart Cities eForum Asia-Pacific on 23 November.
It’s incredibly difficult to test future applications for mobility, such as autonomous vehicules (AVs). They are costly and complex to build, trials require approvals that can take a long time to get, and there is also a huge risk factor in terms of people’s safety. At the same time, it is critical to get the interface between people and new systems like automated vehicles right. One of the things we have been exploring is the use of virtual reality to test interactions between people and AVs. In addition to using computer-generated environments, we have recorded fully autonomous vehicles operating in a controlled environment and then imported the 360-degree recordings into virtual reality. This allows us to provide a highly realistic experience to participants. We’ve been using this approach to test aspects such as user experience and trust.
There is a clear indication from research studies that communicating awareness (what the AV sees) and intent (what the AV is planning to do next) positively affects people’s trust. In artificial intelligence (AI), there is a large body of work studying this phenomenon under the umbrella term “explainable AI”. The problem is that if we remove the manual driver, in the case of AVs, there is no way for pedestrians to communicate with the vehicle, or for them to predict an AV’s behaviour – which is the explainability element. So one of the innovations is to use human-machine interfaces to facilitate this kind of communication. This is also necessary for the people inside the AV – who equally need to build trust into the machine’s decision making – which the industry is already exploring. But it’s much more difficult to get this right for people around the car, that’s where a lot of innovation is still necessary: to enable better communication between humans and AVs, whether those take the form of cars, small pods, or drones. Policy makers need to make sure that these human concerns are considered in AV regulations. This is still mostly unexplored terrain, and we hope that our work in this area can provide insights into what the humans concerns are and how they could be addressed.
We are using a multi-tiered approach. For example, I am working on speculative design solutions and developing new ways to prototype and test solutions, like human-machine interfaces for AVs, before they are rolled out. Others in the Smart Urbanism Lab use a social science approach to study the experiences people have with current transport services, linking this to policy development, or studying the factors that influence global smart city development more broadly. What is unique about the School of Architecture, Design and Planning is that it brings a diverse range of labs together and encourages researchers to transcend the boundaries of labs. This kind of environment is highly effective for enabling new ways of thinking and innovations. For example, in addition to being a member of the Smart Urbanism Lab, I also lead the Urban Interfaces Lab and I work closely with colleagues from the Designing with AI Lab. Our work on smart cities and AVs spans all those labs. At the same time, we have access to a fully autonomous vehicle platform through my collaboration with colleagues at the Australian Centre for Field Robotics, which is also part of the University of Sydney. Our approach is also multi-tiered in that we are working on discovery research as well as partnering with government bodies and industry.