Robots must become faster for use in hospitals

Robots can do a lot. But if they are to be used in a dynamic environment such as a hospital, for example to transport things from A to B, they are overwhelmed. "If robots work together with humans, they see too slowly," says Yulia Sandamirskaya. "If a person suddenly gets in the way, the robot's camera takes too long to reliably identify and locate the object. By the time it gets that far, the robot has already collided with the person," Sandamirskaya explains. "That's dangerous." The physicist took over as head of the ZHAW research focus Cognitive Computing in Life Sciences in March 2023 and is an expert in neuromorphic computing. The human nervous system serves as a model for Yulia Sandamirskaya to develop new algorithms and hardware systems.

Smart cameras

The key to robots that can move safely in a dynamic environment is smart cameras, according to the researcher. "Today's cameras are still based on the same principle as the first cameras: they deliver several images per second, which we perceive as film." Sandamirskaya, on the other hand, is developing a camera that works like the human eye, namely via impulses instead of images. A sensor to that effect already exists on the market, but Sandamirskaya wants to make such cameras smarter by making them see in 3D and ensuring that the robot understands its environment and can react to the unexpected in real time.

"I want new technologies to make their way from academic research into practice and the marketplace to improve people's quality of life."

- Yulia Sandamirskaya, Cognitive Computing in Life Sciences.

In the field of health, Sandamirskaya sees many potential applications for such brain-like robots. "An anesthesiologist often works alone. A robot could help her by grabbing a desired instrument and putting it in her hand," Sandamirskaya says. A real-time camera could also be used in rehabilitation by checking whether a patient is doing exercises correctly. In the laboratory, flexible robotic arms could be used to relieve humans of repetitive tasks. "These are just ideas," Sandamirskaya emphasizes. She plans to engage with potential users in the healthcare industry to determine the need for intelligent robots in everyday work and then conduct research on that application that is most pressing.

Speed up the process and expand the spectrum

Yulia Sandamirskaya is fascinated by studying and technically recreating complex human motion control. "There are many unanswered questions in this area, which challenges me," explains the Belarus native. "I also want to ensure more diversity in the algorithms - because today's artificial intelligences are all based on the same algorithm." This algorithm has to process a lot of data during the learning process and can therefore only be adapted slowly. With the help of new neurological networks, the process can be accelerated and the spectrum expanded. This serves the people, says Yulia Sandamirskaya. "I want new technologies to make their way from academic research into practice and the marketplace to improve people's quality of life."