The use of autonomous robotic systems in hospitals and long-term care facilities offers a wide range of opportunities to reduce the workload on medical and nursing staff, streamline workflows, and support patient care. Intelligent assistance and service robots are becoming increasingly important, particularly in light of the growing need for care and the rising shortage of skilled workers. Potential applications range from the autonomous transport of medical supplies, medications, and care materials to logistical support tasks and direct interaction with people in need of care. Robotic systems are not intended to replace skilled staff, but rather to assist with time-consuming and physically demanding tasks and to free up additional capacity for personal care.

However, hospitals and long-term care facilities present particularly challenging operating environments for autonomous robotic systems. Hallways, patient rooms, and common areas are characterized by high levels of activity and close proximity to people. Patients, nursing staff, medical equipment, and transport carts often move simultaneously within confined spaces. In addition, spatial conditions can change at any time due to objects left lying around, open doors, or sudden changes in workflows. Autonomous robots must therefore continuously map their environment, reliably detect moving and stationary obstacles, and flexibly adapt their navigation to the situation at hand.

A key challenge is that the safety and Wellbeing of people must always be the top priority. Particularly when interacting with older adults, those with limited mobility, or individuals requiring care, high standards must be set for the reliability and predictability of robotic behavior. At the same time, hidden obstacles, limited visibility, and incomplete sensor data can lead to uncertainties in environmental perception. It is therefore crucial not only that autonomous systems perceive their environment, but also that they can quantify the uncertainty in their perception, evaluate the reliability of available information, and adapt their behavior accordingly.

At the Chair of Adaptive Autonomy and Off-Road Robotics, we research methods for reliable perception,uncertainty estimation, and adaptive decision-making in human-centered environments. In doing so, we investigate how robots can account for uncertainties from various sensor sources, identify potential risks, and adapt their navigation and behavior based on the situation. Additionally, we explore the cooperative exchange of environmental knowledge among multiple robots to enable reliable decision-making even under conditions of limited perception. Our approaches are developed and evaluated using mobile robotics platforms and realistic hospital simulations. The goal is to improve the safety, robustness, and adaptability of autonomous care and service robots and to enable their reliable deployment in close proximity to humans.