Aleksandr Medvedev
Russian State Scientific Center for Robotics and Technical Cybernetics
Papers
1
Total Citations
2
H-Index
1
About
Aleksandr Medvedev is a leading researcher at the intersection of robotics, computer vision, and natural language processing, with a primary focus on enhancing the adaptability of autonomous systems. His most influential work introduces **LERa (Look, Explain, Replan)** , a novel framework that empowers robotic agents to recover from execution failures using visual feedback. By integrating a Visual Language Model, LERa allows a robot to "look" at its environment, "explain" the discrepancy between its plan and reality, and dynamically "replan" its actions—moving beyond the limitations of purely text-based Large Language Models. This contribution directly addresses a critical bottleneck in instruction-following robotics: the inability to handle real-world perturbations. While his 2025 paper has already garnered early citations, signaling its immediate relevance, Medvedev’s broader impact lies in pioneering a paradigm where robots can reason visually and adapt in real-time. His work is foundational for developing truly resilient, human-interactive robots capable of operating in unstructured environments, marking him as a rising innovator in embodied AI.
Research Focus
Key Achievements
Top Papers
- 1LERa: Replanning with Visual Feedback in Instruction Following2 citations · 2025