On October 21, 2026, at the Institute for Robotics and Intelligent Machines This, a lecture from the IRIM Fall 2026 seminar series will be held, focusing on a key challenge in robotics: achieving seamless physical interaction between humans and machines. The speaker will be Nadia Figueroa from the University of Pennsylvania, who will present a new approach based on dynamic systems, enabling robots to behave safely, adaptively, and consistently in real time - even under uncertainty.
Seamlessness as a new standard in human-robot interaction
The event, part of the IRIM Fall 2026 series, which will take place on October 21, 2026, in room Klaus 1116 E&W at the Institute for Robotics and Intelligent Machines This, focuses on one of the most ambitious goals of modern robotics: achieving seamless physical interaction. The goal is to enable robots not only to perform tasks but also to react to human movements, forces, and intentions in a natural way - just like a human. The speaker will be Nadia Figueroa, an assistant professor at the University of Pennsylvania, who has been researching the integration of machine learning, control, and estimation in the context of safe human-robot collaboration for years.
Instead of treating robots as static data processing systems, the new approach is based on modeling their behavior as dynamic systems - where movement, force, and intention are constantly co-regulated. This makes it possible to create motion policies that are not only safe but also flexible and predictable even in uncertain conditions. Research shows that this approach enables robots to generalize complex behaviors from minimal demonstrations - which means that they do not have to be taught each task separately.
Dynamic systems as the foundation for safe collaboration
The key to achieving seamless interaction is the use of a dynamic systems perspective. Nadia Figueroa presented how shaping robot behavior through stable vector fields, confidence manifolds, and adaptive dynamics models enables the generation of motion policies that are reactive, safe, and consistent. This is not just theory - research shows that this approach allows robots to avoid obstacles in real-time, adapt to changes in the environment, and maintain safety over a long horizon during physical collaboration.
It is also important that this approach allows for formal safety guarantees - which is crucial in human environments. A robot cannot simply be "pleasant to work with"; it must be predictable and safe even in non-standard situations. The integration of control theory, machine learning, and perception enables the construction of robotic intelligence that not only reacts but also "understands" the physical context - just like a human. This represents a breakthrough towards true collaboration, rather than just automation.
From simulation to real-world action: the future of robotics aligned with human needs.
Enlarged imageClose zoomPrevious imageAs part of Nadia Figueroa's research, work is also being done on combining dynamic systems principles with vision-based learning, based on data from simulations. The goal is to achieve stable, compliant, and force-aware robot behavior - even without direct interaction with humans during training. This means that a robot can "learn" the appropriate approach to a task in simulation and then transfer it to reality, while maintaining safety and smoothness. Research shows that this approach allows for faster and more efficient learning of complex tasks.
An important aspect is also that the robot does not have to be "perfect" - it may be somewhat uncertain or imprecise, but its behavior will still be safe and compliant. This changes the way we think about robots: not as ideal machines, but as partners that can learn, adapt, and collaborate in a natural way. This approach has the potential to transform sectors where physical interaction with humans is key - from rehabilitation to logistics.
Significance and limitations: what does this really mean for the future?
The IRIM Fall 2026 event is not just a presentation of research - it is a signal that robotics is approaching a new stage. Instead of focusing solely on precision and speed, more and more attention is being paid to smoothness, safety, and adaptation. This has implications for many areas: from assisting the elderly at home, through assistive robots in factories, to therapeutic interactions. However, it is worth remembering that this is not yet mass implementation - the research is at the experimental and demonstration stage.
The dynamic systems-based approach has the potential to change the way robots are designed, but its effectiveness depends on the quality of the models, data, and environmental conditions. A lack of full compliance with human expectations or unpredictable behavior in unusual situations may still pose a challenge. Therefore, it is crucial to continue researching, testing, and developing formal safety guarantees - which is what Nadia Figueroa promotes.