Max Robotics — Global Service Robot Directory + US Certification

About

Paloma Sodhi is a leading researcher at the intersection of robotics, computer vision, and agricultural automation, whose work bridges the gap between perception, state estimation, and real-world phenotyping. Her research spans three core areas: differentiable optimization for robot learning, high-throughput plant phenotyping, and novel sensing modalities for localization. Sodhi’s most impactful contribution is **Theseus**, an open-source library for differentiable nonlinear least squares optimization built on PyTorch (47 citations), which provides a unified framework for end-to-end structured learning in robotics and vision. In agricultural robotics, she pioneered **in-field segmentation and identification of plant structures using 3D imaging** (55 citations), enabling automated high-throughput phenotyping of energy sorghum crops—a critical step toward accelerating plant breeding through genetic-phenotypic correlation. Sodhi also advanced state estimation with **Incremental Constrained Smoothing (ICS)** (19 citations) and introduced **ground encoding** for robot localization using ground penetrating radar (24 citations), complemented by the **CMU-GPR dataset** (8 citations). Her work on **virtual occupancy grid maps** (46 citations) enables globally consistent SLAM and planning in 3D environments. With over 237 total citations, Sodhi’s research is distinguished by its practical impact, open-source contributions, and novel integration of optimization, perception, and agricultural robotics.