Papers
18
Total Citations
417
H-Index
11
About
Anthony Simeonov is a robotics researcher whose work spans robot manipulation, object representation learning, and robotic actuation. He is perhaps best known for developing **Neural Descriptor Fields (NDFs)**, a category-level SE(3)-equivariant object representation that enables robots to understand and manipulate objects across diverse poses and configurations — a paper that has garnered over 138 citations since 2022 and significantly advanced the field of generalizable robot manipulation. His subsequent work on Local Neural Descriptor Fields extended this framework to handle unfamiliar objects in unstructured environments, pushing toward more capable household robots. Simeonov has also made meaningful contributions to robotic actuation, particularly in modeling supercoiled polymer (SCP) artificial muscles, including hysteresis compensation techniques that improve control precision in compliant robotic systems — work cited nearly 70 times. Beyond representation learning and actuation, his research touches on imitation learning pipelines (JUICER), motion planning with neural networks, 3D scene imagination for affordance prediction (MIRA), and lifelong learning through language-guided planning with LLMs. His early work even ventured into acrobatic robotics. Together, this body of work reflects a broad yet cohesive vision of building robots that perceive, learn, and act more like humans across real-world settings.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5JUICER: Data-Efficient Imitation Learning for Robotic Assembly21 citations · 2024
- 6
- 7Motion Planning Networks15 citations · 2019
- 8MIRA: Mental Imagery for Robotic Affordances13 citations · 2022
- 9Lifelong Robot Learning with Human Assisted Language Planners11 citations · 2024
- 10Stickman: Towards a Human Scale Acrobatic Robot11 citations · 2018