Ankur Handa

About

Ankur Handa is a pioneering researcher at the intersection of robotics, computer vision, and machine learning, with particular expertise in semantic 3D mapping, dexterous manipulation, and GPU-accelerated simulation for robot learning. His landmark work, **SemanticFusion** (2017, 657 citations), demonstrated how convolutional neural networks could be fused with dense 3D mapping systems to give robots richer, semantically meaningful representations of their environments — a foundational contribution to intelligent mobile robotics. Handa has been instrumental in advancing simulation-to-reality transfer, co-developing **Isaac Gym** (2021, 322 citations), a high-performance GPU-based physics platform that dramatically accelerated reinforcement learning for robotics by keeping simulation and training entirely on the GPU. His work on dexterous robotic hands — including **DexYCB**, **DexPilot**, and **DeXtreme** — has pushed the boundaries of hand-object interaction, teleoperation, and agile in-hand manipulation. More recently, **CuRobo** introduced parallelized, GPU-accelerated collision-free motion planning for manipulators. Across his career, Handa's research has accumulated over 1,900 citations, reflecting his broad and sustained influence on how robots perceive, simulate, and physically interact with the world.

Research Focus

Key Achievements

Top Papers

1. 1
   SemanticFusion: Dense 3D semantic mapping with convolutional neural networks

   657 citations · 2017
2. 2
   Isaac Gym: High Performance GPU-Based Physics Simulation For Robot\n Learning

   322 citations · 2021
3. 3
   DexYCB: A Benchmark for Capturing Hand Grasping of Objects

   250 citations · 2021
4. 4
   DexPilot: Vision-Based Teleoperation of Dexterous Robotic Hand-Arm System

   197 citations · 2020
5. 5
   Sim2Real in Robotics and Automation: Applications and Challenges

   151 citations · 2021
6. 6
   Self-Supervised Siamese Learning on Stereo Image Pairs for Depth Estimation in Robotic Surgery

   104 citations · 2017
7. 7
   DeXtreme: Transfer of Agile In-hand Manipulation from Simulation to Reality

   88 citations · 2023
8. 8
   CuRobo: Parallelized Collision-Free Robot Motion Generation

   82 citations · 2023
9. 9
   GPU-Accelerated Robotic Simulation for Distributed Reinforcement\n Learning

   74 citations · 2018
10. 10
    Factory: Fast Contact for Robotic Assembly

    54 citations · 2022

Key Collaborators

Contact & Links