A Framework for Autonomous Near-Earth Object Precovery

Abstract

Near-Earth Objects (NEOs) that are sufficiently large and close enough to Earth pose a risk of disastrous collisions. The detection of these objects represents the critical first step in mitigating potential impact risk. One of the primary methods for identifying moving objects is to link three separate tracklets of detections, which can take up to multiple weeks. With the introduction of powerful observatories like the Vera C. Rubin Observatory and NEO Surveyor, supplementary methods of detection are needed to accommodate the dramatic increase in new detections. Precovery is a detection process that finds an object in archival observatory data to refine its orbit, reducing the need for follow-up observations. Target objects are generally too faint for older observatories to detect, but they can be revealed with image processing. Precovery is currently used on a case-by-case basis but has the potential to become another fast and reliable method for identifying NEOs. This paper presents a proof-of-concept framework for automating precovery through a streamlined software pipeline. The developed Python framework integrates astronomical online tools and software packages. The pipeline starts with an object’s initial orbital elements and physical parameters from NASA, propagates backward in time, filters for observable states, and queries archival images with the predicted positions. The system then uses image timestamps and observatory locations to re-propagate the object and achieve more accurate states, which are used to retrieve image data for object detection. The pipeline was validated using three well-characterized NEOs of different Near-Earth orbit classes, and the results were compared with NASA’s JPL Horizons ephemeris data. These comparisons showed agreement between the predicted position and the actual position, with offsets below 0.1 arcseconds. This level of precision on constrained objects is a strong benchmark for the framework’s performance. The pipeline is also equipped to handle orbit uncertainty through covariance propagation. The current framework delivers the images and predicted locations; future work will incorporate image processing to complete the automation process. This framework for a fully autonomous precovery system reduces the manual nature of current precovery methods. The streamlined precovery process could augment other detection algorithms in confirming the thousands of NEOs that powerful new observatories like Vera C. Rubin will discover. This proof of concept demonstrates the potential for precovery to quickly quantify NEO populations and support planetary defense efforts.