Haresh Karnan
Papers
17
Total Citations
374
H-Index
10
About
Haresh Karnan is a robotics researcher whose work sits at the intersection of autonomous navigation, social robot behavior, and machine learning. He is best known for his foundational contributions to social robot navigation, most notably through the creation of the Socially CompliAnt Navigation Dataset (SCAND), a large-scale demonstration dataset that has become a key resource for training robots to move safely and appropriately among humans, garnering over 100 citations since its 2022 release. Karnan has also shaped how the field measures progress, co-authoring widely adopted principles and guidelines for evaluating social navigation algorithms, work that has accumulated nearly 75 citations across two versions. Beyond social navigation, he has made significant contributions to high-speed off-road robot control through learned visual-inertial inverse kinodynamics models and explored vision-only imitation learning through his VOILA framework, enabling robots to learn from observations without requiring matched hardware demonstrations. His earlier work on sim-to-real reinforcement learning and tensegrity robot control reflects the breadth of his expertise. With over 340 total citations, Karnan's research meaningfully advances the goal of deploying reliable, socially aware autonomous robots in real-world human environments.
Research Focus
Key Achievements
Top Papers
- 1
- 2Principles and Guidelines for Evaluating Social Robot Navigation Algorithms59 citations · 2024
- 3VOILA: Visual-Observation-Only Imitation Learning for Autonomous Navigation42 citations · 2022
- 4
- 5Grounded action transformation for sim-to-real reinforcement learning26 citations · 2021
- 6
- 7Rethinking Social Robot Navigation: Leveraging the Best of Two Worlds17 citations · 2024
- 8
- 9Principles and Guidelines for Evaluating Social Robot Navigation Algorithms15 citations · 2023
- 10Visual feedback control of tensegrity robotic systems11 citations · 2017