Nikhil Keetha
Papers
7
Total Citations
534
H-Index
5
About
Nikhil Keetha is a robotics and computer vision researcher whose work sits at the intersection of 3D scene understanding, simultaneous localization and mapping (SLAM), and robot perception. He is best known for SplaTAM, a groundbreaking 2024 system that pioneered the use of 3D Gaussian Splatting for dense RGB-D SLAM, enabling richer, more explicit scene representations for robotics and augmented reality — a contribution that has already garnered over 320 citations, signaling its rapid influence on the field. His work on AnyLoc (163 citations) tackled one of visual place recognition's most persistent challenges: building a universal system that performs reliably across wildly diverse environments, from structured urban roads to unstructured natural terrain, leveraging foundation models to transcend task-specific limitations. Keetha has also contributed to open-set 3D semantic mapping through ConceptFusion and the more recent RayFronts, pushing robots toward open-world scene understanding. His meta-analysis on foundation models for general-purpose robotics further reflects his broad, forward-looking vision for the field. Across his career, Keetha's research consistently advances robot autonomy by making perception more generalizable, expressive, and applicable to real-world deployment.
Research Focus
Key Achievements
Top Papers
- 1SplaTAM: Splat, Track & Map 3D Gaussians for Dense RGB-D SLAM323 citations · 2024
- 2<i>AnyLoc</i>: Towards Universal Visual Place Recognition163 citations · 2023
- 3
- 4AnyLoc: Towards Universal Visual Place Recognition10 citations · 2023
- 5SplaTAM: Splat, Track & Map 3D Gaussians for Dense RGB-D SLAM6 citations · 2023
- 6ConceptFusion: Open-set Multimodal 3D Mapping4 citations · 2023
- 7