Dmitry Savransky

"ACCIS: Autonomous Cross-Calibration for Imaging Satellites"

ABSTRACT -- We present a fully autonomous, image-based, cross-calibration concept for constellations of down-looking satellites. This method seeks to remove the requirements of pre-determined ground control points, the majority of manual, ground-based, pre-flight cross-calibration, and large portions of the ground segment processing from the problem of data harmonization across multiple space-borne sensors. The approach is based on continuously applied dynamic filtering to estimate each spacecraft's position, attitude, and camera state, encoding a geometric calibration. The filter is applied to both conventional measurement sources such as GPS and star trackers, along with invariant features extracted from the primary mission imaging data. Satellites pass their measurements and current filter states across the constellation such that subsequent filter iterations are run on augmented states, sharing data between multiple constellation elements, eventually leading to overall network consensus on the geometric calibration. This approach can also allow for the potential autonomous flagging of transient events, when state estimates based on the same scene at different times diverge from one another.

We present a detailed description of the overall concept, along with the results of the first two years of project work. These include the development and demonstration of a new, higher-order unscented estimator (HOUSE), the creation of an integrated simulation framework for analyzing the concept, and preliminary results from structural, thermal and optical performance (STOP) analysis of a representative SmallSat model to generate realistic process noise distributions for the simulation. We will also discuss plans for year three of the project and beyond.

SPEAKER -- Dr. Dmitry Savransky is an Associate Professor in the Sibley School of Mechanical and Aerospace Engineering at Cornell University.

Professor Savransky received his PhD in Mechanical and Aerospace Engineering from Princeton University, where he was a member of the High Contrast Imaging Laboratory. After completing his doctorate in 2011, he became a postdoctoral fellow at the Lawrence Livermore National Laboratory where he assisted in the integration and testing of the Gemini Planet Imager and participated in the planning of the Gemini Planet Imager Exoplanet Survey.

Professor Savransky's research is comprised of a variety of problems related to the detection and characterization of extrasolar planets. These include investigations of advanced control and estimation for active wavefront control systems; computer vision and machine learning applications for automated optical system alignment and astronomical image post-processing; optimal scheduling for autonomous space observatories and ground-based surveys; and statistical analysis of large astronomical surveys.

Next Week: "Gasoline Powered Lighting, 1860s-1920s", Donald Linebaugh, Uinversity of Maryland

Engineering Colloquium home page: https://ecolloq.gsfc.nasa.gov