Konstantine Tsotsos

Papers

Total Citations

H-Index

About

Konstantine Tsotsos is a researcher whose work sits at the intersection of computer vision and robotics, with a primary focus on visual-inertial odometry (VIO) and ego-motion estimation. His major contributions lie in developing robust state estimation systems that fuse data from cameras and inertial measurement units, enabling machines to understand their own movement in complex environments. A key achievement is his work on "Learned Monocular Depth Priors in Visual-Inertial Initialization" (2022, 14 citations), which advanced the crucial initialization phase of VIO by incorporating deep learning to improve accuracy and reliability. Earlier, his foundational paper "Visual-inertial ego-motion estimation for humanoid platforms" (2012, 6 citations) addressed a specific challenge for bipedal robots: handling significant scale changes during forward motion with a limited field of view. This work introduced a novel system that leveraged sparse multi-scale feature tracking to achieve robust performance. Tsotsos’s research is notable for its direct application to humanoid robotics, where precise, real-time motion tracking is essential for balance and navigation, demonstrating a clear impact on the practical deployment of autonomous systems.