Satellite Lost and Found

Expertly tracking debris objects in the congested space environment

19 Ratings

Imagine you are a satellite operator afraid that your spacecraft will collide with a newly spawned debris field. Should you perform an expensive maneuver to get your satellite out of harm’s way? The Multiple Frame Assignment Space Tracker (MFAST) software can create a catalog of debris objects, providing useful data to enable more confident predictions of collisions and inform satellite operations.


Imagine an errant piece of space debris collides with an operational spacecraft and creates a breakup spawning thousands of new debris objects. You need to establish orbits on these new fragments as quickly as possible to avoid a catastrophic snowball effect of further collisions. You task all available sensors to take observations of the debris field. Faced with the daunting task of associating these observations to individual debris objects, you pass the stream of measurements to the Multiple Frame Assignment Space Tracker (MFAST) software. MFAST generates candidate orbits from this data, in real-time, and improves the accuracy of these orbits as subsequent data streams are processed. Eventually, MFAST provides you with a provisional catalog of the new debris objects that can be used to predict and evade future collisions.


Operators need a time-efficient way to track and maintain a more comprehensive catalog of small, dim, deep-space, and hard-to-acquire debris objects. With the instantiation of new sensors able to see penny-sized debris, there will be a proliferation of newly detected objects. Although the current operational system meets present requirements, in the future it may struggle to manage a much-larger catalog with as many as half-a-million space objects. By failing to adopt new paradigms for tracking, such as the one used by Numerica’s Multiple Frame Assignment Space Tracker (MFAST) software, the U.S. may have a diminished ability to maintain super The central problem in tracking multiple objects in space is the data association problem of partitioning sensor observations into discrete objects. MFAST uses an optimization-based method for assigning data to specific space objects. This assignment formulation works in real-time, up to 100 times faster than competing methods. Changes in past data associations can also be made to improve current ones, significantly reducing the number of incorrectly associated observations.


In 2006, AFOSR engaged Numerica to develop a tracking solution specifically designed for space surveillance. A series of Air Force-funded SBIRs provided Numerica the opportunity to put this AFOSR basic research into practice by developing the MFAST software. Under the AFRL GEO Odyssey program, MFAST is being transitioned to operations to provide improved uncorrelated track resolution.

"Imagine if you could make an investment decision today based on tomorrow’s Wall Street Journal. MFAST does something analogous; it can change past decisions based on new data." — Aubrey Poore


The Multiple Frame Assignment Space Tracker (MFAST) software provides a real-time multi-sensor tracking system for space surveillance. Radar, optical sensors, and an optional prior catalog of space objects provide input data to MFAST. The MFAST software solves the problem of associating data to specific cataloged objects or to new objects. MFAST outputs an updated catalog of space objects. There are several key components contained within MFAST that have been customized for space: physics-based methods to quickly rule out infeasible hypotheses, new estimation and fusion techniques to improve orbital accuracy and covariance realism, and hypothesis scoring techniques based on a probabilistic Bayesian framework. Another innovative feature of MFAST not currently used in operations is the concept of a “sliding window,” which facilitates changes in past decisions to improve current ones and is especially valuable for resolving closely-spaced objects. The MFAST software is provided with a variety of user interfaces that can be customized to the specific tracking problem under consideration by the operator.


The MFAST software provides a valuable tool to the warfighter that acts as a “force multiplier” in certain time critical activities. Operators and analysts can process a higher throughput of sensor data through MFAST than was previously possible, thereby providing more timely and accurate catalog maintenance, breakup processing, and resolution of uncorrelated tracks.

The success from the development of MFAST software for space has resulted in a funded transition opportunity to harden the software for use in the Air Force’s Distributed Space Command and Control Center at Dahlgren (DSC2-D). The effort required to achieve this transition has resulted in new job creation within Northern Colorado.

Preservation of our nation’s space assets are of vital importance to national security, as well as global commerce, communication, and transportation systems. With the volume of objects in space expected to grow dramatically in the coming years, solidifying the U.S.’s positon of dominance in space will further ensure the preservation of global infrastructure and systems that make our global systems work.

In addition to the transition opportunity to DSC2-D, development of the MFAST has led to a number of commercialization successes across other Numerica business lines. An important result is that Numerica is developing long term staying power as more and more corporate knowledge is gained about the process of taking technology from early stage research to mature operational software – a process that also drives enhanced engagement with our scientists and engineers.

MFAST can establish orbits up to 10 times faster than existing methods in typical breakup scenarios. — Shawn Herman, VP of Integrated Defense Systems at Numerica Corporation

Scalability of MFAST

The problem of tracking all objects in different regimes of space using multiple radar and optical sensors, ranging from catalog maintenance to breakups to uncorrelated track resolution, is challenging. The MFAST solution offers the best potential at the system-level. The framework is also embarrassingly parallelizable and highly scalable, being able to exploit modern computing architectures and high performance computing.

Numerica Corporation

Fort Collins, CO

Numerica Corporation is a small business specializing in the research and development of advanced algorithms and scientific software for multiple target tracking and data fusion systems.

Joshua Horwood Joshua Horwood

Joshua Horwood

Senior Research Scientist

Jeff Aristoff Jeff Aristoff

Jeff Aristoff

Program Manager

Navraj Singh Navraj Singh

Navraj Singh

Research Scientist


Advanced Estimation and Data Fusion Strategies for Space Surveillance / Reconnaissance





For more exciting Air Force launch stories, visit



Remote-Controlled Recon


Seeing the Unseen