Björn Ellensohn is a doctoral researcher at the Chair of Cyber-Physical Systems at Montanuniversität Leoben, working on perception and manipulation for autonomous robotic systems. His research focuses on building scalable, semantic 3D scene representations for real-world robotics, combining Gaussian Splatting, vision-language models, and multi-modal sensor fusion.

His work explores how robots can construct and continuously refine rich, queryable world models that enable robust interaction, reasoning, and generalization in complex environments. This includes research on active SLAM, semantic mapping, and language-conditioned perception for humanoid and mobile robots.

Björn is actively involved in developing large-scale robotic software systems based on ROS 2, as well as GPU-accelerated infrastructure for real-time perception and learning. His work has contributed to publications at leading robotics venues such as ICRA.

In addition to his research, he designs and maintains distributed GPU computing environments and software stacks for robotics and machine learning, enabling large-scale experimentation and deployment in both research and teaching settings.