Ethan Fahnestock
I'm a Ph.D. student in MIT's Robust Robotics Group advised by Professor Nicholas Roy.
I work on helping robots navigate in unfamiliar environments by learning to exploit distant visual cues, prior overhead data, and environment structure. I'm particularly interested in how learning can help robots reason through their actions and reliably operate in less structured settings.
In summer 2025, I interned at Aurora, a self-driving truck company, working on joining their perception and planning models. I demonstrated that richer perception features improved trajectory ranking. Previously, I studied robotics and physics at the University of Rochester.
Selected research
All publications-
2026LOCI: Leveraging Semantic Maps for City-Scale Cross-View Localization
In submission to IEEE Transactions on Field Robotics (T-FR)
Combining visual observations and semantic maps to localize robots across city-scale environments.
-
2025Far-Field Image-Based Traversability Mapping for A Priori Unknown Natural Environments
IEEE Robotics and Automation Letters (RA-L)
Learning from nearby terrain to predict traversability farther away and guide navigation through unknown environments.
-
2024Learning Semantic Traversability Priors Using Diffusion Models for Uncertainty-Aware Global Path Planning
ICRA Workshop on Resilient Off-road Autonomy
Using diffusion models to infer unseen terrain from partial observations and plan under uncertainty.
-
2019Language-Guided Semantic Mapping and Mobile Manipulation in Partially Observable Environments
Conference on Robot Learning
