Papers
35
Total Citations
1,613
H-Index
18
About
Jesse Thomason is a pioneering researcher at the intersection of natural language processing, human-robot interaction, and embodied AI. His work fundamentally addresses how robots can understand and act upon human language in real-world environments, spanning grounded language learning, vision-and-language navigation, and task planning. Thomason's most celebrated contribution, **ProgPrompt**, demonstrates how large language models can generate situated robot task plans without exhaustive domain engineering — a breakthrough that has garnered over 500 citations and reshaped thinking about LLM-driven robotics. His earlier foundational work on learning to interpret natural language commands through human-robot dialog (137 citations) established interactive, adaptive approaches to language understanding that avoid costly annotated corpora. He co-created influential datasets including **Vision-and-Dialog Navigation** and **TEACh**, which challenge agents to navigate and complete tasks through conversational grounding — critical benchmarks for the embodied AI community. His multi-modal grounded language learning research uniquely extended robot perception beyond vision to haptic, auditory, and proprioceptive signals. Contributing to a comprehensive survey of Vision-and-Language Navigation (91 citations) further cements his role in shaping this rapidly growing field. Through platforms like BWIBots and richly collaborative datasets, Thomason consistently bridges theoretical AI advances with practical, deployable human-robot interaction systems.
Research Focus
Key Achievements
Top Papers
- 1ProgPrompt: Generating Situated Robot Task Plans using Large Language Models508 citations · 2023
- 2Learning to interpret natural language commands through human-robot dialog137 citations · 2015
- 3Vision-and-Dialog Navigation119 citations · 2019
- 4
- 5
- 6TEACh: Task-Driven Embodied Agents That Chat89 citations · 2022
- 7Learning multi-modal grounded linguistic semantics by playing I Spy70 citations · 2016
- 8
- 9
- 10ProgPrompt: Generating Situated Robot Task Plans using Large Language Models44 citations · 2022