Papers
8
Total Citations
216
H-Index
5
About
Igor Kalevatykh is a robotics researcher whose work sits at the intersection of robot manipulation, visual perception, and machine learning. His research primarily focuses on enabling robots to plan and execute complex manipulation tasks using visual inputs, simulation, and learned skills — challenges that lie at the heart of making robots practically useful in unstructured, real-world environments. Kalevatykh's most influential contribution, "Monte-Carlo Tree Search for Efficient Visually Guided Rearrangement Planning" (2019, 73 citations), introduced a sophisticated pipeline allowing robots to reorganize objects based solely on RGB camera input — a significant step toward vision-driven autonomy. His work on sim-to-real transfer, "Learning to Augment Synthetic Images for Sim2Real Policy Transfer" (45 citations), tackled the critical challenge of bridging the gap between simulated training environments and real-world deployment. His 2021 paper on differentiable simulation for physical system identification (44 citations) advanced how robots model frictional contact — a notoriously difficult problem in physics-based learning. Further contributions exploring the combination of primitive skills with reinforcement learning (37 citations) demonstrate his consistent effort to build versatile, adaptive robotic systems. With over 200 cumulative citations, Kalevatykh has established himself as a meaningful contributor to modern robot learning and task-and-motion planning research.
Research Focus
Key Achievements
Top Papers
- 1
- 2Learning to Augment Synthetic Images for Sim2Real Policy Transfer45 citations · 2019
- 3Differentiable Simulation for Physical System Identification44 citations · 2021
- 4
- 5
- 6
- 7Distance robotics learning using Hybrid Simulating Testbed3 citations · 2014
- 8Learning to Augment Synthetic Images for Sim2Real Policy Transfer2 citations · 2019