Kourosh Hakhamaneshi
Papers
2
Total Citations
13
H-Index
2
About
Kourosh Hakhamaneshi is a researcher advancing the frontier of robotic learning, with a focus on enabling autonomous agents to master complex, long-horizon tasks through data-driven skill acquisition. His work centers on hierarchical imitation learning and human-in-the-loop systems, addressing the critical challenge of generalization in robotics. In his highly cited paper "Skill Preferences" (2021), Hakhamaneshi introduced a novel framework that extracts reusable robotic skills from large offline demonstration datasets, then refines them using human feedback—overcoming the performance limitations of purely generative models. This approach has garnered 7 citations for its practical pathway to solving challenging tasks. Complementing this, his work "Hierarchical Few-Shot Imitation with Skill Transition Models" (2021, 6 citations) tackles the dual goals of long-horizon planning and few-shot generalization by learning skill transition dynamics, enabling agents to compose behaviors for unseen scenarios. Together, these contributions establish a foundation for scalable, data-efficient robotic learning. Hakhamaneshi’s research is particularly notable for bridging offline skill extraction with interactive human guidance, a paradigm that promises to make robots more adaptable and capable in real-world environments.
Research Focus
Key Achievements
Top Papers
- 1
- 2Hierarchical Few-Shot Imitation with Skill Transition Models6 citations · 2021