Academic Projects

Scamp Series

SCAMP, the Super Complicated AI Mission Payload, is redefining onboard intelligence by integrating advanced neural networks directly into NASA's core Flight System (cFS). Through high-altitude balloon flights and upcoming rocket tests, SCAMP tests cFS's ability to manage multiple AI accelerators—like Coral TPUs and Hailo chips—simultaneously, running deep learning models for tasks like image segmentation, anomaly detection, and system health monitoring. SCAMP Version 1 validated real-time neural inference at the edge of the atmosphere, while Version 2 scales up with multiple accelerators working in tandem. Looking ahead, SCAMP Version 3 will integrate a custom FPGA-based TPU alongside existing hardware aboard a sounding rocket, pushing cFS to orchestrate even more complex AI workloads. With each mission, SCAMP proves that cFS can reliably run advanced AI in orbit-like environments, paving the way for autonomous, adaptive spacecraft.

• Learn More

Terp Raptor

The ~350-meter Potentially Hazardous Asteroid (PHA) 99942 Apophis will make a historic close approach to Earth on April 13, 2029, passing within ~31,634 km of Earth's surface—closer than geosynchronous satellites. Such an event is estimated to occur only once every 7,500 years, offering a rare opportunity for planetary science and defense studies. The Terrapin Engineered Rideshare Probe for Rapid-response Apophis Profiling, Tracking, Observing, and Reconnaissance (TERP-RAPTOR) is a 12U CubeSat mission concept by University of Maryland students to perform a flyby of Apophis during its perigee. Designed for launch from a Geosynchronous Transfer Orbit (GTO) rideshare, TERP-RAPTOR aims to deliver exceptional scientific returns at a low cost. This mission promises to be a pivotal observer of Apophis's 2029 close approach, alongside other spacecraft deployed to study this extraordinary event.

• Learn More

Vertex

The Vehicle for Extraterrestrial Research, Transportation, and Exploration (VERTEX) employs a distributed software architecture using the Robotic Operating System (ROS) to integrate and manage data across its components for decision-making. Organized as a directed network of nodes, the system enables both automated and manual control modes. In automated mode, the Intelligence Sector processes sensor and image data to inform the Automation Agent, which determines actions and coordinates intermediate controllers for calculations and low-level controllers for hardware actuation. Manual mode allows users to select specific suspension and steering controllers for precise operation. Key features include Apriltag detection for locating astronauts relative to the vehicle and a modular design that separates computational tasks from direct hardware control. This approach ensures adaptability, efficiency, and reliable performance, supporting VERTEX's extraterrestrial exploration objectives.

• Learn More