Publications

Google Scholar

2026

IMPACT: Intelligent Motion Planning with Acceptable Contact Trajectories via Vision-Language Models

Yiyang Ling∗, Karan Owalekar∗, Oluwatobiloba Adesanya, Erdem Bıyık, Daniel Seita (∗ equal contribution)

IEEE International Conference on Robotics and Automation (ICRA) · 2026

IMPACT uses Vision-Language Models to let robots plan contact-rich motions in cluttered environments by inferring which contacts are acceptable, improving task success while minimizing harmful collisions.

2023

One-shot learning for methane leak detection in remote areas using hyperspectral data

Karan Owalekar, Shailesh Deshpande

IEEE International Geoscience and Remote Sensing Symposium (IGARSS) · 2023

A one-shot approach to methane detection in hyperspectral imagery: a contrastive model compares an observed spectrum with a reference methane signature to predict methane presence, outperforming traditional approaches in accuracy and speed on AVIRIS data from the Aliso Canyon leak.

Triplet-Loss Driven Optimization for Improved Methane Leak Detection

Karan Owalekar, Shailesh Deshpande, Arpan Pal

13th Workshop on Hyperspectral Imaging and Signal Processing: Evolution in Remote Sensing (WHISPERS) · 2023

Extends the one-shot detector with a triplet-loss objective that learns spectral embeddings for sharper separation between methane and background signatures, improving accuracy and speed over the earlier one-shot method on the same dataset.