PRISM’s platform uses AI segmentation to identify and
highlight residential structures in a neighborhood.
NASA sponsored Entrepreneurs Challenge events in 2020, 2021, and 2023 to
invite small business start-ups to showcase innovative ideas and technologies
with the potential to advance the agency’s science goals. To potentially
leverage external funding sources for the development of innovative
technologies of interest to NASA, SMD involved the venture capital community in
Entrepreneurs Challenge events. Challenge winners were awarded prize money, and
in 2023 the total Entrepreneurs Challenge prize value was $1M. Numerous
challenge winners have subsequently refined their products and/or received
funding from NASA and external sources (e.g., other government agencies or the
venture capital community) to further develop their technologies.
One 2023 Entrepreneurs Challenge
winner, PRISM Intelligence (formerly known as Pegasus Intelligence and Space),
is using artificial intelligence (AI) and other advances in computer vision to
create a new platform that could provide geospatial insights to a broad
community.
Every day, vast amounts of remote
sensing data are collected through satellites, drones, and aerial imagery, but
for most businesses and individuals, accessing and extracting meaningful
insights from this data is nearly impossible.
The company’s product—Personal
Real-time Insight from Spatial Maps, a.k.a. PRISM—is transforming geospatial
data into an easy-to-navigate, queryable world. By leveraging 3D computer
vision, geospatial analytics, and AI-driven insights, PRISM creates photorealistic,
up-to-date digital environments that anyone can interact with. Users can simply
log in and ask natural-language questions to instantly retrieve insights—no
advanced Geographic Information System (GIS) expertise is required.
For example, a pool cleaner looking
for business could use PRISM to search for all residential pools in a five-mile
radius. A gardener could identify overgrown trees in a community. City
officials could search for potholes in their jurisdiction to prioritize
repairs, enhance public safety, and mitigate liability risks. This broad level
of accessibility brings geospatial intelligence out of the hands of a few and
into everyday decision making.
The core of PRISM’s platform uses radiance fields to convert raw 2D imagery into high-fidelity, dynamic 3D visualizations. These models are then enhanced with AI-powered segmentation, which autonomously identifies and labels objects in the environment—such as roads, vehicles, buildings, and natural features—allowing for seamless search and analysis. The integration of machine learning enables PRISM to refine its reconstructions continuously, improving precision with each dataset. This advanced processing ensures that the platform remains scalable, efficient, and adaptable to various data sources, making it possible to produce large-scale, real-time digital twins of the physical world.
The PRISM platform’s interface showcasing a 3D digital
twin of California State Polytechnic University, Pomona, with AI-powered search
and insights.
"It’s great being able to push the state of the art in this relatively
new domain of radiance fields, evolving it from research to applications that
can impact common tasks. From large sets of images, PRISM creates detailed 3D
captures that embed more information than the source pictures." — Maximum Wilder-Smith, Chief Technology
Officer, PRISM Intelligence
Currently the PRISM platform uses
proprietary data gathered from aerial imagery over selected areas. PRISM then
generates high-resolution digital twins of cities in select regions. The team
is aiming to eventually expand the platform to use NASA Earth science data and
commercial data, which will enable high-resolution data capture over larger
areas, significantly increasing efficiency, coverage, and update frequency.
PRISM aims to use the detailed multiband imagery that NASA provides and the
high-frequency data that commercial companies provide to make geospatial
intelligence more accessible by providing fast, reliable, and up-to-date
insights that can be used across multiple industries.
What sets PRISM apart is its focus
on usability. While traditional GIS platforms require specialized training to
use, PRISM eliminates these barriers by allowing users to interact with
geospatial data through a frictionless, conversational interface.
The impact of this technology could
extend across multiple industries. Professionals in the insurance and appraisal
industries have informed the company how the ability to generate precise, 3D
assessments of properties could streamline risk evaluations, reduce costs, and
improve accuracy—replacing outdated or manual site visits. Similarly, local
governments have indicated they could potentially use PRISM to better manage
infrastructure, track zoning compliance, and allocate resources based on
real-time, high-resolution urban insights. Additionally, scientists could use
the consistent updates and layers of three-dimensional data that PRISM can
provide to better understand changes to ecosystems and vegetation.
As PRISM moves forward, the team’s focus remains on scaling its capabilities and expanding its applications. Currently, the team is working to enhance the technical performance of the platform while also adding data sources to enable coverage of more regions. Future iterations will further improve automation of data processing, increasing the speed and efficiency of real-time 3D reconstructions. The team’s goal is to expand access to geospatial insights, ensuring that anyone—from city planners to business owners—can make informed decisions using the best possible data.
PRISM Intelligence founders Zachary Gaines, Hugo Delgado, and Maximum Wilder-Smith in their California State Polytechnic University, Pomona lab, where the company was first formed.
No comments:
Post a Comment