3D Reconstruction Intern

KBR•Moffett Field, CA

12h•$22 - $24

About The Position

High-resolution, near-real-time 3D modeling can greatly enhance lunar science surface operations. Rapid and accurate scene reconstructions support situational awareness while enabling scientists to visualize and analyze geophysical, geothermal, and geochemical data. With many upcoming orbital, landed, and human missions for the lunar polar regions, we are particularly interested in robust 3D volume rendering applications for the exploration of the Moon, including identifying and analyzing concentrations of water ice and other volatiles . This work is focused on the development and validation of an efficient, robust pipeline for the volumetric reconstruction of lunar polar scenes across multiple wavelengths. This will leverage advanced techniques such as Neural Radiance Fields (NeRFs) and Gaussian Splatting . The pipeline is designed to enhance planetary mission capabilities in areas like situational awareness, science prioritization, and in-situ sample analysis. A key application is the mapping of lunar volatiles, which could be integrated with the upcoming VIPER mission to support rapid and informed science decisions. Our methodology will be developed and evaluated using multispectral data obtained from the instruments aboard VIPER.

Requirements

Both c++ and python desired
C++17 (or equivalent) development knowledge is acceptable in place of python (Cmake, gtest, make, and other development tools would be expected)
Basic knowledge of GNU/Linux for use as a development machine is required
Python PyTorch
C++
Experience in Machine Learning
Experience in Computer Vision
Prior experience with common remote sensing data products is desired
Experience with GIS software is an advantage
Experience with multispectral sensors for remote sensing is desired

Responsibilities

Develop a robust and efficient software pipeline for 3D reconstruction, tailored to simulate lunar polar scenes within a controlled lab or analog environment, using novel volumetric rendering techniques across various wavelengths.
Demonstrate the efficacy of 3D reconstruction using multispectral instrument data in augmenting the efficiency of lunar surface science operations, particularly for VIPER or Artemis missions.