Labelbox > Case Studies > NASA’s Jet Propulsion Laboratory Leverages Machine Learning for Extraterrestrial Life Search

NASA’s Jet Propulsion Laboratory Leverages Machine Learning for Extraterrestrial Life Search

Technology Category

Analytics & Modeling - Machine Learning
Sensors - Camera / Video Systems

Applicable Industries

Cement
Education

Applicable Functions

Product Research & Development

Use Cases

Intelligent Urban Water Supply Management
Leakage & Flood Monitoring

Services

System Integration
Training

About The Customer

The customer in this case study is NASA’s Jet Propulsion Laboratory (JPL), a leading center for robotic exploration of the solar system. JPL has been involved in space exploration since the beginning of the Space Age and has sent spacecraft to all of the planets in the solar system. The Ocean World Life Surveyor (OWLS) project at JPL is currently preparing for the possibility of sending a spacecraft to one of the moons in our solar system where water has been discovered. The Machine Learning Instrument Autonomy (MLIA) group at JPL is responsible for developing machine learning models to aid in this exploration.

The Challenge

NASA’s Jet Propulsion Laboratory (JPL) is on a mission to find signs of life in our solar system, focusing on the presence of water, a vital element for life. The Ocean World Life Surveyor (OWLS) project at JPL is preparing to send a spacecraft to either Europa, a moon of Jupiter, or Enceladus, a moon of Saturn, where ice and water vapor have been discovered. The spacecraft will be equipped with microscopes to collect video data from water samples, looking for evidence of microbes. However, sending this microscopy data back to Earth is a complex and costly task due to the vast distance. Traditional compression methods are inadequate, and the energy cost of downlinking the data is extremely high. The Machine Learning Instrument Autonomy (MLIA) group at JPL faced the challenge of building a machine learning (ML) model that could identify videos most likely to contain signs of life, capture short clips, and prioritize them for downlinking back to Earth.

The Solution

The MLIA team decided to use traditional ML methods such as decision trees and SVMs to build a model that could fit onboard a spacecraft's computer, find and track moving particles, distinguish life-like motion from drifting or jostling, and efficiently explain its decisions. The team also worked closely with scientists to review model outputs and verify the model's accuracy. To create their algorithm, the team labeled a large, diverse set of data sourced from various field water samples and lab-grown specimens. However, creating and maintaining an in-house labeling solution was challenging. To overcome this, the MLIA team turned to Labelbox for both the training data platform and workforce. The team uploaded video datasets collected from Digital Holographic Microscopes, and the Labelbox Workforce team annotated these assets quickly, enabling the MLIA team to focus on other tasks and move the project forward much faster.

Operational Impact

The use of Labelbox for data labeling and training has significantly improved the efficiency of the MLIA team at JPL. The team was able to quickly upload video datasets and have them annotated by the Labelbox Workforce team, freeing up their time to focus on other tasks and move the project forward. The MLIA team is now on track to complete their current ML project by the end of 2021, which will result in an ML-powered system that can identify, track, and classify life-like movement directly from microscopy data. The team is also working to bring other ML teams at JPL to Labelbox, so that the training data platform can be a general solution for the entire lab. This project is a significant step for researchers planning for future space excursions, although launching a complete mission to either of the moons could require another decade of research and development.