Valts Blukis
Papers
17
Total Citations
915
H-Index
9
About
Valts Blukis is a robotics researcher whose work sits at the intersection of natural language processing, task planning, and robot manipulation. He is best known for pioneering the use of large language models in robot task planning, most prominently through **ProgPrompt** (2023), which demonstrated how LLMs can generate programmatic, situated task plans for robots — a contribution that has garnered over 500 citations and reshaped how the field approaches instruction following. His research consistently tackles the challenge of enabling robots to understand and act on high-level human instructions, spanning persistent spatial semantic representations, few-shot object grounding, and natural language feedback for plan correction. Beyond language grounding, Blukis has made significant contributions to robot motion generation through **CuRobo** (2023, 82 citations), a parallelized GPU-accelerated framework for collision-free motion planning. His earlier work on socially competent navigation using braid group topology (2017) reflects a longstanding interest in human-robot interaction. More recently, his contributions to **RVT-2** and **RoboSpatial** advance precise 3D manipulation and spatial reasoning in vision-language models. Across his portfolio, Blukis demonstrates a rare ability to bridge foundational AI research with practical robotic deployment.
Research Focus
Key Achievements
Top Papers
- 1ProgPrompt: Generating Situated Robot Task Plans using Large Language Models508 citations · 2023
- 2CuRobo: Parallelized Collision-Free Robot Motion Generation82 citations · 2023
- 3Correcting Robot Plans with Natural Language Feedback67 citations · 2022
- 4
- 5ProgPrompt: Generating Situated Robot Task Plans using Large Language Models44 citations · 2022
- 6
- 7RVT-2: Learning Precise Manipulation from Few Demonstrations33 citations · 2024
- 8
- 9
- 10