A Robot That Sees, Feels, and Responds to Humans
The award-winning doctoral thesis by Jakub Rozlivek, titled Perception and Safety in Close-Proximity Human–Robot Interaction, is motivated by one of the key challenges of modern robotics: how to enable robots to operate safely in environments shared with humans.
Rozlivek focused on how robots can use cameras and other sensors to better perceive their surroundings and respond to human movement in real time. The research introduced, among other contributions, a new approach to modeling the space around a robot—the so-called perirobotic space—as well as a reactive motion control system that allows a robot to quickly avoid people or adjust its movement trajectory.
The research also explores the use of soft protective elements that reduce the force of a potential impact during contact with a human. This allows collaborative robots to operate faster and more efficiently without the need to separate them from people with protective barriers.
“Robots are gradually moving from isolated industrial cells into environments where people are present—from factories to healthcare or elderly care. Safety is therefore essential,” explains Jakub Rozlivek.
The award-winning research was conducted in the Humanoid Robotics Laboratory under the supervision of Assoc. Prof. Matěj Hoffmann from the Department of Cybernetics at FEE CTU.
Artificial Intelligence Helping Robots Grasp the World
The master’s thesis by Ing. Elizaveta Isianova, titled Object Grasp Pose Estimation Using Multimodal Models, addresses another fundamental challenge in robotics: how to teach robots to correctly manipulate objects in the real world.
In her work, Isianova used modern vision–language models (VLMs)—a type of artificial intelligence capable of processing not only text but also visual information. This enables a robot to better understand what it actually sees.
The proposed method allows robots to generate optimal grasp positions for different types of grippers. For example, a robot can recognize that a drill should be grasped by its handle rather than by the drill bit or the trigger.
Such an approach has significant potential particularly in automated manufacturing, logistics, or e-commerce, where robots handle a wide variety of objects.
The thesis was supervised by Ing. Varun Burde, a doctoral student at FEE CTU who also works as a researcher in the Industry 4.0 Testbed at the Czech Institute of Informatics, Robotics and Cybernetics of CTU.
The Werner von Siemens Awards are among the most prestigious recognitions for students, young scientists, and educators in the Czech Republic. Each year, the awards highlight outstanding work in technical and natural sciences and promote collaboration between academic research and industrial practice. More information about the results of the 2025 edition can be found in the Siemens press release Nejlepší mladí vědci a pedagogové získali Ceny Wernera von Siemense - Tiskové centrum SIEMENS.
Photo credits: Siemens
