Junwon Seo

Agency for Defense Development, Carnegie Mellon University

Papers

Total Citations

H-Index

About

Junwon Seo is a rising researcher in the field of autonomous robotics, specializing in off-road navigation and environmental perception. His work focuses on enabling mobile robots to operate safely and reliably in unstructured, outdoor terrains—a challenge far beyond the scope of conventional on-road systems. Seo’s major contributions include pioneering self-supervised learning methods for traversability estimation, as demonstrated in his 2023 paper "Self-Supervised 3D Traversability Estimation With Proxy Bank Guidance" (10 citations), which allows robots to learn from past driving experiences without manual labeling. He has also advanced semantic mapping under uncertainty, introducing evidential reasoning with Bayesian Kernel Inference in his 2024 works (5 and 2 citations), enabling robots to construct reliable maps despite noisy sensor data in off-road environments. These innovations directly address critical gaps in robotic autonomy, improving safety and adaptability in applications like agriculture, search-and-rescue, and planetary exploration. With a growing citation record and a focus on uncertainty-aware, self-supervised systems, Seo is establishing himself as a key contributor to the next generation of robust, field-deployable robots.