Aleksei Valenkov

Russian State Scientific Center for Robotics and Technical Cybernetics

Papers

Total Citations

H-Index

About

Aleksei Valenkov is a researcher advancing the frontier of 3D scene understanding and open-vocabulary object grounding. His work centers on enabling autonomous agents to interpret complex, context-dependent natural language queries within three-dimensional environments. Valenkov’s major contribution, the 2025 paper "Beyond Bare Queries: Open-Vocabulary Object Grounding with 3D Scene Graph," tackles a critical limitation of existing CLIP-based methods: their inability to resolve ambiguous descriptions that require reasoning about object relationships. By integrating 3D scene graphs, his approach allows agents to locate objects described not just by name, but by their spatial and semantic context—a leap toward more intuitive human-robot interaction. With 11 citations in its first year, this work has quickly garnered attention for addressing a practical bottleneck in embodied AI. Valenkov’s research bridges computer vision, natural language processing, and robotics, offering a pathway to systems that can navigate and act upon nuanced human instructions in real-world spaces. His contributions are particularly relevant for applications in autonomous navigation, assistive robotics, and augmented reality, where understanding the “where” and “why” behind a query is as crucial as recognizing the object itself.