Martin Kamme: OKAPI: Orbits Praktikum
I was fortunate enough to receive sponsorship from The Europe Center and the Stanford Club of Germany to pursue a 9 week remote internship in Germany during the summer of 2021. I focused my application on Germany’s growing interest in the space debris problem and was thus led to the NewSpace start up OKAPI: Orbits. I connected with Christopher Kebschull who believed I could be a good fit with the company, and he generously gave me the opportunity to contribute to their platform during the summer months.
NewSpace and OKAPI
“NewSpace” is a recently coined term that generally refers to the commercialization of the Space Domain. It brings along numerous benefits such as lowering launch prices, driving innovation, and transitioning away from the traditionally government dependent model. However, the rapid increase in the space population is crowding popular orbits and quickly raising the risks of on orbit collisions. A Low Earth Orbit satellite travels at 7 kilometers per second, which means that a collision with another satellite results in a catastrophic scatter of space debris. The most famous example of such an event is the 2009 Iridium collision with a Cosmos satellite, and collisions are becoming more common with the International Space Station being struck as recently as May of this year. The prevention of further collisions requires knowledge of debris and active satellite locations, a field known as Space Situational Awareness (SSA).
OKAPI: Orbits is a team of experts in the fields of Space Situational Awareness and Space Surveillance and Tracking that provides software for the safe operation of satellites and maintenance of a sustainable space environment. Their platform offers a multitude of services such as: orbit propagation, orbit determination, maneuver optimization, and conjunction analysis. As an intern, I was tasked with analyzing the performance of their Orbit Propagation and Orbit Determination algorithms.
The OKAPI: Orbits team has developed a highly accurate Orbit Propagator called NEPTUNE that takes the initial satellite state (i.e position and velocity) as input and can propagate that into the future. Unfortunately, the initial satellite state isn’t the only factor that governs the behavior of a satellite in space; aerodynamic drag, the uneven distribution of Earth’s gravity, and satellite maneuvers all affect the movement of a satellite in orbit. The NEPTUNE propagator has tunable parameters that account for these factors, but it is not impervious to errors that grow over time. My first task was to analyze the error behavior and compare it to NEPTUNE’s prediction of the error growth.
Figure 1 presents the meter level error statistics in the tangential location of the Sentinel-1b spacecraft over the duration of a week. The figure plots the average tangential error, the standard deviation of the propagation errors as calculated from the difference of the NEPTUNE states and ESA provided states (light gray shading), and the NEPTUNE propagated error (light blue shading). I was given a multitude of parameters to tune and through a series of structured testing, this was the best performance I was able to generate with the NEPTUNE orbit propagation. The NEPTUNE propagated error tracks the real error quite closely and can compensate for maneuvers as shown with the vertical red lines.
Orbit propagation can produce very accurate predictions of satellite state but is prone to errors that grow over time. Thus, it is important to also be able to determine the satellite orbit from measurements. My other task at OKAPI was to analyze their Orbit Determination algorithm. As with Orbit propagation, Orbit Determination is a very complicated procedure with many inputs that affect performance. In essence, an initial guess of the satellite state is given, and the algorithm uses measurements to manipulate that initial guess to what it believes is the best estimate of the state.
Figure 2 presents an example of this routine where residuals are the differences between the determined orbit state and the measured state. The results in Figure 2 are quite poor, meaning that the Orbit Determination algorithm failed to converge. This could be due to multiple factors, and my analysis consisted of identifying these factors and manipulating the inputs to produce better results. I concluded that unknown satellite maneuvers were the biggest contributors to poor Orbit Determination results and created a procedure of manipulating the inputs to generate significantly better convergence.
Due to the Covid-19 pandemic, I was unable to work in the OKAPI office. The OKAPI team was very accommodating and seamlessly incorporated me into their workflow. Communication was orchestrated through Microsoft Teams and my responsibilities were always clearly explained and my questions were quickly and effectively addressed by my team members. I am fortunate enough to carry a European passport, so I situated myself at my grandparents’ home in southern Sweden and was able to work the entire 9 weeks in the same time zone as my colleagues. Despite the remote situation, I thoroughly enjoyed my challenging work as well as getting to know the team through the daily meetings and weekly Coffee Breaks.
OKAPI provided a fantastic opportunity to work on difficult problems with a highly motivated team that shares my passion about Space Debris and Sustainable Space. My GRIP experience honed professional skills for my future career and fostered valuable friendships. I am sincerely thankful for the sponsorship from The Europe Center and the Stanford Club of Germany and for the amazing team at OKAPI: Orbits who made my summer of 2021 such a rewarding experience.
My European passport permitted me to travel in Europe during Summer 2021, and I was able to visit OKAPI separately from GRIP. Here is the team that is working on a more sustainable space environment.