Toki Migimatsu
Stanford University, Corvallis Environmental Center, Intel (United States)
Papers
15
Total Citations
440
H-Index
9
About
Toki Migimatsu is a robotics researcher whose work sits at the intersection of natural language processing, manipulation learning, and task planning, with a focus on enabling robots to understand and execute complex instructions in human-centric environments. A central theme across their research is bridging the gap between high-level human communication and low-level motor control. Their Concept2Robot framework (2020–2021, 134 combined citations) demonstrated that a single multi-task policy could map natural language instructions and visual scene inputs directly to robot motion trajectories, marking a significant step toward flexible, instruction-following robots. Building on this, Text2Motion (2023, 197 citations) advanced the field further by enabling long-horizon task and motion planning from natural language, combining symbolic reasoning with feasibility verification. Migimatsu has also contributed to articulated object tracking, symbolic state estimation, and skill sequencing through works like STAP and Grounding Predicates through Actions, collectively underscoring a commitment to robust, generalizable robot autonomy. With over 400 cumulative citations, their research has meaningfully shaped how the robotics community approaches language-conditioned manipulation and sequential decision-making.
Research Focus
Key Achievements
Top Papers
- 1Text2Motion: from natural language instructions to feasible plans197 citations · 2023
- 2
- 3
- 4Learning to Scaffold the Development of Robotic Manipulation Skills22 citations · 2020
- 5Category-Independent Articulated Object Tracking with Factor Graphs16 citations · 2022
- 6STAP: Sequencing Task-Agnostic Policies13 citations · 2023
- 7Grounding Predicates through Actions13 citations · 2022
- 8
- 9Text2Motion: From Natural Language Instructions to Feasible Plans10 citations · 2023
- 10