RoboTalks
RoboTalks @ TUM
Tracking Reality: Visual-Inertial Systems for Spatial Agents
How do we achieve precise real-time tracking in robotics and XR, without relying on expensive or closed hardware?
In this week’s RoboTalk, Mateo de Mayo, Ph.D. student at the Computer Vision Group at TUM, dives into egocentric visual-inertial tracking using classical methods — a low-cost, energy-efficient solution capable of sub-centimeter precision. Mateo will explore why open solutions are critical for the future of spatial agents, how Linux limitations sparked innovation in open-source VI tracking, and what’s next for tracking in XR, robotics, and HCI.
Learn how motion blur, noisy data, and human perception challenges (like kinetosis) are being tackled through robust algorithm design.
A 15–20 minute talk will be followed by a 30–40 minute open discussion and Q&A.
RoboTalks is a weekly speaker series by RoboTUM – Innovation & Education, giving students, researchers, and professors the stage to share ideas and spark interdisciplinary exchange.