MOCI: Structure-From-Motion in Low Earth Orbit
Abstract
The Multi-view Onboard Computation Imager (MOCI) is a 6U CubeSat being developed by the University of Georgia Small Satellite Research Laboratory (SSRL) and is sponsored by the Air Force Research Laboratory University Nanosatellite Program (UNP). MOCI utilizes the standard CubeSat bus for the main avionics stack, and will function as an on-orbit processing testbed.
MOCI's primary science objective will be to use custom Structure-from-Motion (SfM) algorithms to produce 6.7-m Digital Elevation Models (DEMs) of ground targets on-orbit and in near real-time using a Nvidia GPU with minimal flight heritage, as well as a custom optical payload.
Additional mission objectives include performing object classification and tracking of predetermined ground targets up to an approximate 10-m resolution, as well as passively computing changes of terrain in contested spaces with sensor and data fusion. Ground targets used for DEM creation through the onboard SfM algorithms will primarily be large landscape features, such as mountains or valleys. Ground targets used for object classification and tracking will include a wider variety of objects, including grounded airplanes and sports fields.
The COTS GPU onboard MOCI has been integrated with a custom peripheral PCB to further increase performance. Other custom PCBs have also been developed by the SSRL and are integrated into the avionics stack with other commercial-off-the-shelf (COTS) boards. The GPU that will be flown, and Nvidia Jetson TX2i, was originally designed for terrestrial applications, and therefore may cease operation due to radiation effects prior to the end of MOCI's operational lifetime. Expectations for the TX2i lifetime and radiation hardness are detailed, and are based on experimental results. Measures are taken to ensure a potential TX2i failure does not end MOCI's operational lifetime, and are detailed. Due to the nature of MOCI's scientific objectives, a COTS ADCS is used to achieve the precise pointing requirements and high-accuracy slews required by the SfM algorithms on board.
The MOCI engineering model is currently undergoing system level testing by a team of approximately 40 undergraduate students at the UGS SSRL, with the flight model expected to be completed by Q2 2025 and an anticipated launch readiness date of Q3-Q4-2025. Following the completion of the flight model, environmental testing will take place at Kirtland Air Force Base in Albuquerque, New Mexico.
The UGA SSRL will perform all on-orbit experiments, and will operate the satellite over the course of its operational lifetime. SSRL mission operators will evaluate the effectiveness of the in-house SfM algorithms and have specified several "stretch goals" to test onboard artificial intelligence (AI) over time. Other focuses for MOCI mission operators include testing the optical system auto-calibration over time.
MOCI: Structure-From-Motion in Low Earth Orbit
Utah State University, Logan, UT
The Multi-view Onboard Computation Imager (MOCI) is a 6U CubeSat being developed by the University of Georgia Small Satellite Research Laboratory (SSRL) and is sponsored by the Air Force Research Laboratory University Nanosatellite Program (UNP). MOCI utilizes the standard CubeSat bus for the main avionics stack, and will function as an on-orbit processing testbed.
MOCI's primary science objective will be to use custom Structure-from-Motion (SfM) algorithms to produce 6.7-m Digital Elevation Models (DEMs) of ground targets on-orbit and in near real-time using a Nvidia GPU with minimal flight heritage, as well as a custom optical payload.
Additional mission objectives include performing object classification and tracking of predetermined ground targets up to an approximate 10-m resolution, as well as passively computing changes of terrain in contested spaces with sensor and data fusion. Ground targets used for DEM creation through the onboard SfM algorithms will primarily be large landscape features, such as mountains or valleys. Ground targets used for object classification and tracking will include a wider variety of objects, including grounded airplanes and sports fields.
The COTS GPU onboard MOCI has been integrated with a custom peripheral PCB to further increase performance. Other custom PCBs have also been developed by the SSRL and are integrated into the avionics stack with other commercial-off-the-shelf (COTS) boards. The GPU that will be flown, and Nvidia Jetson TX2i, was originally designed for terrestrial applications, and therefore may cease operation due to radiation effects prior to the end of MOCI's operational lifetime. Expectations for the TX2i lifetime and radiation hardness are detailed, and are based on experimental results. Measures are taken to ensure a potential TX2i failure does not end MOCI's operational lifetime, and are detailed. Due to the nature of MOCI's scientific objectives, a COTS ADCS is used to achieve the precise pointing requirements and high-accuracy slews required by the SfM algorithms on board.
The MOCI engineering model is currently undergoing system level testing by a team of approximately 40 undergraduate students at the UGS SSRL, with the flight model expected to be completed by Q2 2025 and an anticipated launch readiness date of Q3-Q4-2025. Following the completion of the flight model, environmental testing will take place at Kirtland Air Force Base in Albuquerque, New Mexico.
The UGA SSRL will perform all on-orbit experiments, and will operate the satellite over the course of its operational lifetime. SSRL mission operators will evaluate the effectiveness of the in-house SfM algorithms and have specified several "stretch goals" to test onboard artificial intelligence (AI) over time. Other focuses for MOCI mission operators include testing the optical system auto-calibration over time.
