Sensors

---

AXOLTL and Cloudstone have collaborated to develop an innovative automated cloud segmentation pipeline tailored to their ground-based imaging sensors, leveraging advanced AI and ML techniques to enhance satellite data accuracy for space operations. AXOLTL aims to overcome limitations in traditional cloud detection methods, with plans to deploy the improved model on an edge device for real-time analysis to support the SDA TAP Lab's mission.

Arete is actively developing a portable neuromorphic SDA telescope. The prototype system, called Base2, is capable of real-time satellite tracking with high temporal fidelity and is currently being integrated into the UDL.

Millennial Software closes operational kill chain gaps with its TMDB, a central data layer enabling seamless, accurate transfers for CCDM interrogation and hostility analysis. CoCoWatch provides co-orbital and co-planar analysis for real-time space situational awareness of mission-critical assets. We are expanding support for Welder's Arc with an Event Database, Launch Site and Vehicle Databases, CCDM behavior detection, and Grappler PEZ/WEZ capabilities.

Maxar will show its automated non-Earth imaging and UDL messaging sequence. Resulting collections support both persistent monitoring of poles/oceans/solar exclusion zones as well as very high reduction space object characterization.

EdgeRunner's solution accelerates decision making when dealing with unknown/potentially hostile satellites by analyzing images, locating objects of interest within images, identifying their features, and comparing them with data in the Target Model Database (TMDB) to provide actionable recommendations.

PSAI developed artificial intelligence and machine-learning methods to aggregate reliable data from millions of records on spacecraft capabilities. JCO operators can now view satellites' launch data, instrument array, and life expectancy, gaining critical context for avoiding surprise and responding appropriately to orbital threats.

Turion Space has created an innovative suite of machine learning algorithms capable of analyzing resolved spacecraft imagery to extract insights such as shape, classification, solar panel configurations, power consumption, and more. During Cohort 5, the system was enhanced to detect features like optical tubes and parabolic antennas, providing detailed payload insights such as the aperture and angular resolution of optical tubes, the diameter of parabolic antennas, and power output metrics.

Rhea Space Activity has developed the UCT Killer Application for the SDA TAP Lab to automatically process vast amounts of uncorrelated track data and identify objects of interest hidden within. An updated orbital state estimate of these objects is then computed and passed to other members of the TAP Lab via an easy to use web hosted API.

In this video, we highlight the Uncorrelated Track Processor by LSAS-Tec, which processes uncorrelated data from ground stations and generates candidate orbits using custom algorithms and orbit determination software. This data is integrated into Subsystem 2: State Estimation of the Welders Arc system, enhancing the ability to track and differentiate objects in the GEO belt.

Pulsar Space Systems, a Space Weather solutions company, built an orbit propagation system integrating Orion Space Solutions' Dragster atmosphere model utilized to update aging TLEs and identify maneuvers below the noise floor of traditional models.

KMI UCT Processing

PREDATOR POD by Nominal Systems accelerates the reacquisition of critical targets with a high-fidelity, automated probability-of-detection service for any ground or space-based sensor—EO, radar, neuromorphic, infrared, or passive RF. By precisely calculating when and where targets likely can be detected, it empowers operators to conserve sensor resources and quickly reacquire targets to enhance situational awareness.

Katalyst Space Technologies' ARC interrogates data from various sources and phenomenologies to produce an evaluation for each resident space object (RSO) that demonstrates the degree to which it exhibits camouflage, concealment, deception, and maneuver (CCDM). A database of objects and indicators with an objects of interest (OOI) list is provided to the user to help avoid operational surprise.

StarDrive is transforming space operations with AstroShield, an advanced defense and optimization platform that protects critical orbital assets. Leveraging CCDM (Camouflage, Concealment, Deception, and Maneuver) strategies, AstroShield automates collision avoidance, detects threats using machine learning, and executes precise defensive maneuvers. It reduces optical and radar signatures, monitors proximity operations, and delivers rapid-response alerts to ensure spacecraft security in contested environments. StarDrive is also developing a fully reusable SSTO Space Plane to revolutionize satellite deployment, providing rapid, cost-effective, and sustainable access to space. Together, these innovations showcase StarDrive’s commitment to secure and resilient space superiority.

We provide a software module with fast vector handling and comparisons to conquer large amounts of high dimensional data. This helps alleviate overhead when comparing and allows the parent system to focus on more important processes.

Space traffic is growing at an unprecedented rate, creating an urgent need for advanced traffic control systems that can keep pace Our team is thrilled about OrbitGuard's potential - what began as a Space Force initiative has evolved into a groundbreaking solution at a crucial moment. With commercial space activities expanding exponentially, the next evolution of OrbitGuard will be vital for ensuring safe, efficient operations in an increasingly crowded orbital environment.

LaunchSense: Predicting and Detecting Foreign Space Launches

Introducing a GPT-powered solution for accurate and reliable space launch prediction and detection.

The University of Colorado Boulder Data Exploitation Lab for Trusted Autonomy has been developing not classified launch detection for the SDA TAP Lab since the inaugural Project Apollo Cohort using publicly available imagery. CU Boulder has developed a capability named OWLAT - Open-source Weather imagery Launch Alert and Tracklets that provides not classified launch alert messages that can be distributed without restriction to the Project Apollo Cohort. This micro-service aggregates data from publicly available sources and then runs image processing and detection algorithms to declare a launch. If a detection is declared, a not classified launch alert message with launch time, launch location, and launch confidence is distributed to the cohort.

Scout is creating an RF jamming weapon engagement zone detector. This cohort we automated our microservice with a 24-hour look-ahead window, integrated data from the TMDB, and added endpoints for the next subsystem to pull our results.

Modeling threats and simulating possible Weapon Engagement Zones (WEZ) in defense of Resident Space Objects (RSOs) for threat and opportunity situations ensures that decision makers understand the tradespace for taking appropriate offensive or defensive action. The MapLarge Team has integrated our Multi-Parameterized WEZ Prediction Services into the Welder’s Arc architecture to automate future on-orbit actions.

Ten One Aerospace has developed a physics-informed ML framework for co-orbital threat classification and characterization, based on simulated image generation, high fidelity maneuver simulation, and common RPO shapes. Our algorithm enables threat identification and inferred intent for space domain awareness.