Ankit Goyal
Papers
10
Total Citations
732
H-Index
7
About
Ankit Goyal is a robotics and AI researcher whose work sits at the intersection of robot manipulation, task planning, and spatial reasoning. He is best known for developing **ProgPrompt**, a landmark framework that leverages large language models to generate programmatic, situationally aware task plans for robots — a contribution that has garnered over 600 citations across its versions and established him as a notable voice in LLM-driven robotics. His research on 3D object manipulation has produced influential systems including **RVT** and **RVT-2**, which use multi-view transformers to enable precise robotic manipulation from few demonstrations, and **IFOR**, an end-to-end approach to object rearrangement using iterative flow minimization. Goyal has also advanced spatial understanding through the **Rel3D** benchmark and explored 3D multiview pretraining for manipulation with **3D-MVP**. Earlier in his career, he applied computer vision to construction safety, developing methods for detecting hazardous human-robot interactions on job sites. Across his body of work, Goyal consistently bridges perception, language, and action — pushing robots closer to operating effectively in complex, unstructured real-world environments.
Research Focus
Key Achievements
Top Papers
- 1ProgPrompt: Generating Situated Robot Task Plans using Large Language Models508 citations · 2023
- 2
- 3ProgPrompt: Generating Situated Robot Task Plans using Large Language Models44 citations · 2022
- 4IFOR: Iterative Flow Minimization for Robotic Object Rearrangement34 citations · 2022
- 5RVT-2: Learning Precise Manipulation from Few Demonstrations33 citations · 2024
- 6
- 7
- 8RVT: Robotic View Transformer for 3D Object Manipulation6 citations · 2023
- 9
- 103D-MVP: 3D Multiview Pretraining for Manipulation3 citations · 2025