Vaishakh Patil

ETH Zurich, Robotic Research (United States)

Papers

Total Citations

H-Index

About

Vaishakh Patil is a robotics researcher whose work sits at the intersection of 3D perception, scene understanding, and manipulation. His primary research areas include LiDAR point cloud processing, novel view synthesis for teleoperation, and instance-centric robotic grasping. Patil’s most influential contribution is **TULIP**, a transformer-based architecture for upsampling sparse LiDAR point clouds, which has already garnered 18 citations since its 2024 publication. This work directly addresses a critical bottleneck in autonomous vehicle perception by converting irregular 3D data into a format amenable to image super-resolution techniques. In parallel, his work on **radiance fields** (NeRFs and 3D Gaussian Splatting) for robotic teleoperation has demonstrated how photo-realistic novel view synthesis can enable more intuitive remote control of robots, earning 10 citations. Patil also introduced **ICGNet**, a unified approach for instance-centric grasping that integrates geometric analysis with grasp feasibility in cluttered environments (9 citations). Earlier, his 2022 work on improving depth estimation using map-based priors showed how prior map data can substitute for expensive depth sensors, achieving 5 citations. Across these contributions, Patil consistently pushes toward making robots perceive and interact with the world more accurately and efficiently, with a clear trajectory from foundational perception to applied manipulation.