Papers
15
Total Citations
92
H-Index
6
About
Madhava Krishna is a leading researcher at the intersection of robotics, computer vision, and embodied AI, with key contributions spanning motion planning, visual place recognition, and semantic 3D scene understanding. His work on EDMP introduced ensemble-of-costs-guided diffusion for motion planning, achieving 22 citations since 2024 by enabling adaptive, off-the-shelf manipulation in novel scenes. Krishna has also pioneered advances in sensor robustness, developing methods to correct rolling shutter and motion blur distortions in depth cameras like the Kinect, a critical contribution for real-world robotic perception. His impact is further demonstrated through AnyLoc (10 citations), a universal visual place recognition system that overcomes the degradation of traditional methods in unstructured environments, and ConceptFusion, which enables open-set multimodal 3D mapping for richer scene understanding. Krishna’s work on object goal navigation, including data-regularized Q-learning and spatial relation graphs, has set benchmarks for assistive robotics in homes and factories. With over 80 total citations across his top papers, his research consistently bridges theoretical innovation and practical deployment, making him a pivotal figure in advancing autonomous navigation and perception.
Research Focus
Key Achievements
Top Papers
- 1EDMP: Ensemble-of-costs-guided Diffusion for Motion Planning22 citations · 2024
- 2Rolling shutter and motion blur removal for depth cameras10 citations · 2016
- 3AnyLoc: Towards Universal Visual Place Recognition10 citations · 2023
- 4Sequence-Agnostic Multi-Object Navigation9 citations · 2023
- 5Object Goal Navigation using Data Regularized Q-Learning8 citations · 2022
- 6
- 7
- 8
- 9ConceptFusion: Open-set Multimodal 3D Mapping4 citations · 2023
- 10