Spacecraft autonomy is a very complex subject matter to most of the people having a variety of of logic-based algorithms set to handle circumstances. It all needs artificial intelligence (AI) has in the inputs to the algorithms falling within the pre-defined mission scope.

Nowadays, the mission requirements are increasing with much greater navigational precision from spacecraft, that requires high-class AI algos.

A sensing and perception capability is required either it is high-precision space navigation to small comets and asteroids; EDL or RPO. Conventionally, these technologies were under public sector domain but nowadays private sectors are also active in these technologies like autonomous satellite servicing, lunar landing, and research in vision-based AI and machine learning.

Vision-based sensing systems, are considered most appropriate but they have huge amounts of developing cost and are resource intensive. The algorithms translate a raw image into data under the supervision of engineers with specialized expertise. These algorithms are then verified involving complex physical testbeds with robots on tracks of spacecraft and asteroids. Sometimes, the testbeds are also flown in orbit.

Once the validation process is done, the implementation is started. The processing resources are optimized to make them survive the hostile radiation environment of space. The algorithms are then distributed between FPGAs and computer processors. This split can make the design more complex.

NASA’s Raven that is an on-orbit testbed vision-based sensing systems

Source: EDN

Many competitions just as Google Lunar XPRIZE are motivating companies for lower costs of extraterrestrial landing technologies.

But how is it possible? For this Matlab and Simulink is being used for algorithm development. In this way, high level apps are developed rather than re-invention of image processing routines. These languages also support rapid prototyping and can be integrated with navigation, and controls models for early system-level validation. Code is generated automatically on both processors and FPGAs, using these languages and test benches are created for system verification.

Segmentation can be done in MATLAB

Source: EDN

The importance of ML is quite evident from this discussion. To learn about any of the ML or AI techniques, consult Logic Finders as we specialize in ML techniques.

The link to company profile is just below, in case you want any of these services or consultation

https://www.logicfinder.net/wp-content/uploads/2019/05/Company-Portfolio.pdf