With the number of objects in space growing at an ever-increasing rate, the task of maintaining space domain awareness using the traditional systems is becoming more and more untenable. Even making use of recent commercial SSA solutions such as ExoAnalytics global sensor network may still not be sufficient to meet the escalating needs, especially as our adversaries increase their in-space maneuvering agility and tactics. One possible solution is to work toward employing in-space capabilities that utilize relatively low-cost satellites that collaborate to provide whole greater than the sum of the parts approaches to augment the space object custody and tracking enterprise. These platforms would not only serve to provide the sensor hardware to perform observations, but would orchestrate the entire Tasking, Collection, Processing, Exploitation, Dissemination (TCPED) loop autonomously, allowing operators of the ground enterprise to merely specify a collection of Resident Space Objects (RSOs) to monitor and get back curated state data for each RSO and immediate event notifications if there are significant changes. Orbit Logic and the University of Texas at Austin (UT) are teamed to develop a solution called Heimdall Onboard. The research will result in the ability of a team of satellites in a cooperating cluster to perform fully automated RSO custody maintenance of an operator-specified set of high interest objects. The logic required to a) determine an agile schedule to acquire images using onboard sensors, b) detect objects in the images, c) determine orbits for those detections, d) associate the orbits with objects in the locally-maintained onboard catalog, and e) adapt the sensor tasking plans dynamically based on uncertainties, sensor capabilities, and the relative viewing perspectives of the cluster members over time will all be performed on the flight computers of the cluster elements using decentralized, collaborative autonomy and data fusion approaches. Achieving the described capability will be accomplished by integrating several of UTs mature processing and fusion algorithms with Orbit Logics Autonomous Planning System (APS) a decentralized onboard autonomous multi-domain planning framework with flight heritage. UT will assess both Probability Density Function (PDF) and Labeled Multi-Bernoulli (LMB) approaches to decentralized object association in Phase I research to determine the most effective techniques given communication and processing constraints. UTs Expected Information Gain algorithm will support APSs multi-factor figure-of-merit optimization approach to develop deconflicted collection plans across the cluster that maximize RSO state knowledge. A Phase I initial prototype of Heimdall Onboard will integrate all involved software and allow its performance to be characterized when running on flight processor equivalent hardware.
