| Location | Zurich AI Festival, Vitus-Areal in Winterthur |
| Guests | Yulia Sandamirskaya – ZHAW, Kateryna Portmann – ANYbotics, Marisa Kappeler-Schudel – Stadt Winterthur |
| Interviewer | Max Rheiner, Riva Pinto |
On 28 September 2026, LINA hosted “AI Meets Reality: From Algorithms to Autonomous Systems” at the Vitus-Areal in Winterthur. Part of the Zurich AI Festival, the event combined keynotes, live demonstrations and an innovation fair with a discussion of the societal implications of autonomous systems.
At the panel “Among Us: Living Alongside Robots and Autonomous Systems”, Yulia Sandamirskaya from ZHAW, Kateryna Portmann from ANYbotics and Marisa Kappeler-Schudel from the City of Winterthur brought research, industry and local government perspectives to the discussion. Max Rheiner and Riva Pinto from Zurich University of the Arts (ZHdK) moderated the exchange about what today's robots can do and the experiences, expectations and concerns of the people who will live and work alongside them.
More about the event: https://www.lina.swiss/news/ai-meets-reality
There is often a considerable gap between an impressive demonstration and dependable everyday use. For Sandamirskaya, professor at ZHAW, that gap has been a concern since her first assistive robotics project in 2005. Many capabilities attracting attention today were already being researched then. What frustrated her was how rarely that work found its way into everyday life.
Robots can navigate, avoid obstacles and move objects accurately in a fixed, structured environment. Combining these abilities safely around people remains difficult: a mobile robot with an arm must cope when someone unexpectedly enters its workspace. Sandamirskaya is developing applications of this kind for hospitals and care facilities.
She is sceptical that language models will resolve these difficulties. Physical movement requires rapid responses and the control of forces. Explaining to a child how to ride a bicycle does not teach them to balance. Sandamirskaya therefore argued for different models suited to different tasks.
Portmann described how robots are already being integrated into industrial operations. ANYbotics' four-legged robots collect inspection data in industrial facilities and critical infrastructure, including in heat, darkness or areas with hazardous gases. On top, they take on repetitive tasks and can feed their results directly into existing systems. For Portmann, acceptance develops through working with the people who operate the robots and use their data. It matters which tasks they want to delegate and how the technology fits their routines.
She explicitly distinguished these applications from robots in homes. Industrial staff are trained, and operating conditions are understood. A household presents different demands. Safety and regulation, she explained, belong in development from the outset, with testing and verification integrated into the process. She also sees regulation as a potential competitive advantage for Switzerland and Europe. As deployments expand to larger robot fleets, new organisational responsibilities emerge. In her view, final decisions and accountability should remain with people.
Kappeler-Schudel extended the discussion to everyday settings through planned trials of service and cleaning robots in Winterthur's care facilities. Her focus is on how the systems interact with people on site. The city also acts as a mediator between stakeholders and authorities, whose responsibilities vary with the setting.
Her account of employees reluctant to use exoskeletons showed how an anticipated benefit can fail to translate into willingness to use a technology. They feared it would malfunction or make their work harder. Sandamirskaya explained possible safety and ergonomic difficulties with devices worn on the body. Portmann drew on her experience of established applications in rehabilitation. Their exchange illustrated how different the requirements can be even for related technologies.
Working closely with people also raises questions about data collection. In her care facility projects, Sandamirskaya aims to process information on the robot or within the facility. An audience contribution sharpened the point: to avoid a person, a robot does not need to know their identity. Sandamirskaya confirmed that spatial information can be sufficient for obstacle avoidance.
The discussion also addressed people's willingness to share personal data when an application offers a clear benefit. Portmann questioned the different standards applied to robots and familiar services such as Outlook or WhatsApp. The discussion exposed the tension between protecting personal information and accepting its use in exchange for a tangible benefit.
The audience also helped shape the exchange through a live poll, submitting responses on their phones. The resulting answers paired expectations around industry, research and healthcare with concerns about safety, loss of control and privacy. This built the starting point for discussion with the panelists.
The conversation connected technological development with the experiences and expectations of the people affected by it. For LINA, these perspectives suggest concrete priorities for testing: how systems respond to unexpected situations, how they fit existing routines and what information they actually need. Moving into everyday use requires examining these demands in each setting, together with the people who will live and work with the technology.
Continue the conversation: share your perspective on physical AI and autonomous systems here.