Field Calibration of Hyperspectral Cameras for Terrain Inference
Nathaniel Hanson, Benjamin Pyatski, Sam Hibbard, Gary Lvov, Oscar De La Garza, Charles A DiMarzio, Kristen Dorsey, Taskin Padir
AI summary
Problem
RGB cameras cannot distinguish critical intra-class terrain variations like water content, while hyperspectral imaging is highly sensitive to changing ambient light, making reliable field deployment difficult without frequent recalibration.
Approach
The authors built HYPER DRIVE, a mobile robot platform that uses onboard point spectrometers and a learned up-sampling network to dynamically calibrate hyperspectral images under varying outdoor lighting.
Key results
- Designed and deployed HYPER DRIVE, a field-ready hyperspectral imaging system for mobile robots
- Developed a data-driven joint calibration method that dynamically corrects for illumination changes
- Successfully estimated soil moisture content and vegetative health indices from a moving platform
- Demonstrated computationally efficient differentiation of visually similar terrain types
Why it matters
Enables autonomous field robots to accurately assess critical terrain properties for safe navigation and agricultural applications without relying on controlled lighting or static calibration targets.
Abstract
Intra-class terrain differences such as water content directly influence a vehicle’s ability to traverse terrain, yet RGB vision systems may fail to distinguish these properties. Evaluating a terrain’s spectral content beyond red-green-blue wavelengths to the near infrared spectrum provides useful information for intra-class identification. However, accurate analysis of this spectral information is highly dependent on ambient illumination. We demonstrate a system architecture to collect and register multi-wavelength, hyperspectral images from a mobile robot and describe an approach to reflectance calibrate cameras under varying illumination conditions. To showcase the practical applications of our system, HYPER DRIVE, we demonstrate the ability to calculate vegetative health indices and soil moisture content from a mobile robot platform.