Earth Mission Control Engine: A Multimodal AI Driven
Approach to Climate Intelligence in Virtual Reality
Name
debonet-ldebonet-meng-eecs-2026-thesis.pdf
Size
22.78 MB
Format
Adobe PDF
Checksum (MD5)
c6d7860bf89b4ec79e2e3bb29aa1498d
Author(s)
De Bonet, Lucas
Advisor(s)
Newman, Dava J.
Date Issued
February 2026
Publisher
Massachusetts Institute of Technology
Abstract
Decision-making in the face of climate impacts requires timely and contextual access to relevant information such as Earth Observation (EO) and socioeconomic data, but interpreting and applying this data for human-scale, place-based insight remains difficult. Earth Mission Control (EMC) is a collaborative Virtual Reality (VR) based decision-making platform designed to bridge this gap. This thesis explores how multimodal Artificial Intelligence (AI) models can advance EMC’s capacity to support hyperlocal decision-making by making it more versatile, informative, personal, and context-aware for users navigating climate data, and for producers creating vignettes to convey the meaning of this data. This thesis contributes the architecture and implementation of the EMC Engine, a backend system that wraps around the existing functionality of Earth Mission Control, enhancing its flexibility, versatility, and usability with key features: A storyboarding process for creating vignettes, the ability to generatively create scenics using AI skyboxes, a dynamic cloud-hosted library of EMC vignettes, a vector asset database for Retrieval Augmented Generation (RAG), the ability to generate entire vignettes from scratch, and an embodied LLM assistant within the EMC experience. This work aims to help users and producers navigate and produce content for the data-rich environment of EMC, which will accelerate informed decision making around climate impacts and solution pathways. Evaluation assesses system performance, acceleration of production workflows, and deployability. Overall, these contributions position EMC to be a more adaptive and scalable platform, enabling faster and more content-aware creation of vignettes, as well as more hyperlocal understanding for end-users.
MIT Department
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
Terms of Use
In Copyright - Educational Use Permitted
Copyright retained by author(s)
Persistent DSpace Link
