Sombit Dey

Indian Institute of Technology Kharagpur, Naver (South Korea), Sofia University "St. Kliment Ohridski"

Papers

Total Citations

H-Index

About

Sombit Dey is a robotics researcher whose work spans agricultural automation, autonomous navigation, and vision-language-action models. His most cited paper, "A Proposal of FPGA-Based Low Cost and Power Efficient Autonomous Fruit Harvester" (2020, 4 citations), introduces a novel robotic harvester that integrates computer vision, deep learning, and a unique end-effector design for efficient fruit plucking—demonstrating a practical, low-power approach to precision agriculture. In "Learning Whom to Trust in Navigation" (2023, 3 citations), Dey tackles the challenge of terrestrial robot navigation by developing a dynamic switching mechanism between classical SLAM-based planning and neural methods like reinforcement learning, enhancing robustness in uncertain environments. His recent work, "ReVLA: Reverting Visual Domain Limitation of Robotic Foundation Models" (2024–2025, 2 citations), addresses a critical bottleneck in open Vision-Language-Action models by improving their adaptability across diverse visual domains, a key step toward generalist robotics. With a focus on bridging hardware efficiency and algorithmic intelligence, Dey’s contributions are shaping the next generation of autonomous systems that are both cost-effective and cognitively flexible.