Harjatin Singh Baweja

Papers

Total Citations

H-Index

About

Harjatin Singh Baweja is a researcher at the intersection of agricultural robotics, computer vision, and reinforcement learning. His work focuses on developing autonomous systems for high-throughput plant phenotyping, particularly for sorghum—a critical crop for bioenergy and food security. Baweja’s most impactful contribution is **StalkNet**, a deep learning pipeline that enables the automated measurement of plant stalk count and width directly in the field. With 62 citations, this work addresses a major bottleneck in agriculture: replacing slow, labor-intensive manual phenotyping with rapid, precise computer vision. He extended this research with a deep learning-based stalk grasping pipeline (13 citations), integrating perception with robotic manipulation for in-situ harvesting or inspection. Beyond agriculture, Baweja has explored fundamental challenges in robotics, notably in **reinforcement learning without ground-truth state** (9 citations), where he developed methods for learning manipulation policies directly from raw sensory inputs—particularly valuable for handling deformable objects where traditional state estimation fails. His work bridges practical agricultural automation with core advances in robot learning, demonstrating how deep learning can transform both field operations and robotic control.