Michael Zollhoefer

Papers

Total Citations

125

H-Index

About

Michael Zollhoefer is a leading researcher in computer vision and graphics, whose work bridges the gap between neural representations and real-time perception for robotics. His key research areas include neural implicit surfaces, 3D reconstruction, and differentiable rendering, with a focus on enabling machines to understand and interact with dynamic environments. Zollhoefer’s major contribution is the development of iSDF (2022), a real-time neural signed distance field framework that allows robots to continuously learn and update scene geometry from streaming sensor data—without relying on RGB input. This work, which has garnered over 125 citations, introduces a batch-based self-supervision method that enables online optimisation, providing collision costs and gradients for downstream planning tasks in navigation and manipulation. Beyond iSDF, Zollhoefer has made foundational advances in real-time facial performance capture and neural rendering, earning recognition for making complex 3D representations practical for interactive applications. His research exemplifies how neural fields can be deployed in real-world, resource-constrained systems, influencing both academic inquiry and industrial robotics. For students and researchers, Zollhoefer’s work offers a compelling model of how to fuse theoretical depth with engineering pragmatism to solve pressing problems in embodied AI.