Researchers at Örebro University have built a navigation system that lets a robot move through dense crowds by following the way people walk, instead of treating every person as an obstacle. The work was presented at ICRA 2026, the IEEE robotics and automation conference.
Robots that only avoid people can freeze when a crowd gets too dense. The system, called HiCrowd, uses Maps of Dynamics that record how people usually move through a place at different times of day. The robot uses those patterns to predict where people are heading and moves along with them.
The robot, about the size of a small suitcase, was tested in dense crowds at the Miraikan science museum in Tokyo and at Expo 2025 in Osaka. In one run it followed a group that turned right and so avoided a large crowd coming the other way without stopping. People nearby did not keep away from it.
Yufei Zhu, who developed the system for his doctoral thesis in computer science, said a robot guiding a person with a visual impairment cannot stop every time someone walks past.
Background
HiCrowd
Robot navigation system
A navigation system from Örebro University that lets a robot move through dense crowds by following how people walk.
Yufei Zhu
Computer scientist
Computer scientist who developed HiCrowd for his doctoral thesis at Örebro University.



