Deepesh Mehta

Indian Institute of Information Technology Allahabad

Papers

Total Citations

H-Index

About

Deepesh Mehta is a researcher at the forefront of robotics and machine learning, specializing in generative models for robotic manipulation. His work addresses a critical challenge in modern robotics: enabling robots to grasp objects effectively when training data is scarce. Mehta’s most-cited paper, "Generative model based robotic grasp pose prediction with limited dataset" (2022, 24 citations), introduces a novel framework that leverages generative adversarial networks (GANs) to predict optimal grasp poses from small, real-world datasets. This contribution is particularly impactful for real-world applications where collecting large labeled datasets is impractical, such as in manufacturing or assistive robotics. By combining generative modeling with geometric reasoning, Mehta’s approach improves grasp success rates while reducing data requirements, a significant step toward more adaptable and cost-effective robotic systems. His work has been recognized for bridging the gap between simulation and reality, earning citations from researchers in both academia and industry. Mehta’s research continues to push the boundaries of data-efficient learning, making him a promising voice in the next generation of robotics innovation.