Papers
7
Total Citations
85
H-Index
5
About
Hyemin Ahn is a leading researcher at the intersection of human-robot collaboration and embodied AI, whose work bridges natural language understanding, motion generation, and physical human-robot interaction. Her most impactful contribution is the Interactive Text2Pickup (IT2P) network (32 citations), which enables robots to resolve ambiguous human commands during object pick-and-place tasks, pioneering more intuitive human-robot dialogue. Ahn has also made seminal advances in motion retargeting with her self-supervised shared latent embedding (S³LE) method (15 citations), allowing humanoid robots to safely and naturally replicate motions from RGB video without paired training data. Her research extends to robust human motion forecasting using transformer architectures (11 citations), vision-based muscle activation estimation for tele-impedance control (9 citations), and generative models like Text2Action (8 citations) that synthesize human actions from language descriptions. Notably, she developed a generative autoregressive network for 3D dance synthesis from music (5 citations), demonstrating her versatility in cross-modal generation. With over 85 total citations across her core papers, Ahn’s work is shaping the future of robots that can understand, predict, and physically collaborate with humans in unstructured environments.
Research Focus
Key Achievements
Top Papers
- 1
- 2Self-Supervised Motion Retargeting with Safety Guarantee15 citations · 2021
- 3Robust Human Motion Forecasting using Transformer-based Model11 citations · 2022
- 4
- 5Text2Action: Generative Adversarial Synthesis from Language to Action8 citations · 2018
- 6Smartphone-Controlled Telerobotic Systems5 citations · 2014
- 7