Ganesh Iyer

Papers

Total Citations

H-Index

About

Ganesh Iyer is a researcher at the forefront of combining dense 3D mapping with differentiable programming and open-set scene understanding. His work centers on the critical question of representation in simultaneous localization and mapping (SLAM)—specifically, how to move beyond handcrafted features toward learned representations that can be optimized end-to-end. Iyer’s most influential contribution, **∇SLAM** (2020, 24 citations), introduces a framework that makes dense SLAM fully differentiable, enabling gradient-based optimization through the entire SLAM pipeline. This breakthrough allows researchers to integrate SLAM directly into larger learning systems, paving the way for task-driven representation learning. Building on this, his **gradSLAM** (2019, 3 citations) laid the foundational idea of differentiable SLAM, while his more recent **ConceptFusion** (2023, 4 citations) tackles the next frontier: open-set multimodal 3D mapping. ConceptFusion enables robots to build 3D maps that reason about arbitrary semantic concepts—not just a predefined set—by fusing language, vision, and other modalities. Together, Iyer’s work bridges the gap between classical geometric mapping and modern representation learning, offering a powerful toolkit for robots that must understand and navigate complex, open-world environments.