Bakaeva Eva
Papers
1
Total Citations
1
H-Index
1
About
Eva Bakaeva is a researcher at the intersection of embodied AI, natural language processing, and human-robot interaction. Her work addresses a critical bottleneck in deploying intelligent agents in real-world settings: handling ambiguous instructions. She is the lead author of the influential "AmbiK: Dataset of Ambiguous Tasks in Kitchen Environment" (2025), which provides a benchmark for evaluating how large language models (LLMs) detect and resolve task ambiguity during physical task execution. By systematically capturing real-world kitchen scenarios where user commands are underspecified or context-dependent, Bakaeva’s dataset enables researchers to test and improve LLM-based planners beyond idealized lab conditions. Her contributions are foundational for building more robust, user-friendly embodied agents that can ask clarifying questions or infer missing details—a key step toward safe and effective home robotics. While her work is early in its citation trajectory, the AmbiK dataset is already recognized as a vital resource for the growing community studying ambiguity in situated language understanding. Bakaeva’s research promises to bridge the gap between rigid instruction-following and the fluid, context-aware communication that human environments demand.
Research Focus
Key Achievements
Top Papers
- 1AmbiK: Dataset of Ambiguous Tasks in Kitchen Environment1 citations · 2025