Using a wide range of sensors and an ultra-broadband network created by team members, emergency services can orientate themselves and coordinate effectively even without GNSS, light or external communication.
In the NIKE MATE project, the focus is on navigating tunnels in difficult circumstances — collapse of infrastructure, GNSS outages, presence of smoke and debris, all of which make orientation challenging.
NIKE MATE is funded by the Austrian Research Promotion Agency (FFG), a research team consisting of Graz University of Technology (TU Graz), the University of Leoben, the Federal Ministry of Defence, OHB Austria and the Laabmayr Engineering Office.
The NIKE MATE team has developed a system for tough tunnel missions that combines sensor data from robots and rescue teams with a self-built UWB (ultra-wideband) network. The result is a dynamic map of the environment in which the team can locate and coordinate itself.
A human/robot team
The central innovation of the project is “teaming”. A robot with highly developed sensor technology first explores the surroundings and creates the dynamic map. The position information obtained is exchanged via a UWB transmitter with emergency personnel following behind or working in parallel, who are themselves equipped with UWB tags and who place UWB anchors along their route.
In addition to stable data transmission, the anchors also enable distance measurements between all participants even without a direct line of sight. This creates a network of distance measurements in which the positions of robots and people can be determined with an accuracy of closer than one meter.
“This precise localization is a decisive safety factor, for example if there is an open lift door or a precipice in front of a person,” said project manager Philipp Berglez, Institute of Geodesy, TU Graz.
Sensor technology plays an important role in localization. The robot uses a laser scanner, a camera, and wheel sensors to create a map of its surroundings. This means emergency services do not have to rely on plans that may be outdated or no longer correct due to damage.
The rescue workers who follow have inertial sensors (accelerometers and angular rate sensors) on their shoes. Using AI-based analysis, the system recognizes various movement patterns such as walking, crawling on all fours, or belly crawling.
Drone data to be included
To ensure that the position calculations are not only accurate but also reliable, the project team uses factor graph optimization methods. These originate from robotics and make it possible to take past measurements into account again, and thus better determine the current position. If robots or people pass the same place at different times, their data can be linked and the map continuously improved.
“The prototype we developed proved its suitability for use during our tests at Zentrum am Berg at the University of Leoben,” Berglez said. “For real-life use, we now need to make the individual components even more robust so that they can withstand real-life conditions and function reliably.
“We would also like to expand the system to include mini-drones in order to obtain additional data from a slightly higher position in the event of an emergency, which could significantly help emergency services in their work.”
