Software-based redundancy methods offer a solution to achieve high reliability with commercial-off-the-shelf (COTS) hardware, providing a significantly reduced development cycle as well as lower SWaP and cost. By contrast, conventional hardware-based methods for the use of high performance applications-specific processors in space are very costly and introduce a long delay between terrestrial and spaceflight use. This proposal investigates the use of Astrobotic’s Software Defined Reliability product, A-SDR, for the reliable use of COTS coprocessors in space through a specific study of a Texas Instruments digital signal processor (DSP) IC.
A-SDR is a lightweight, low-overhead software consensus framework designed to detect and correct transient single event effects with minimal hardware requirements. With its software-based mitigation techniques, A-SDR can elevate this powerful multicore DSP IC to be flight-reliable, delivering 160 Gigaflops of computing power with a typical power consumption of less than 15 Watts. This capability is simply not available in space today and would have a broad impact on the autonomy of small spacecraft missions, and Astrobotic has identified a specific spaceflight application that requires this advanced computing performance to serve as a proof of concept: autonomous free-flying navigation in LEO and on planetary surfaces. Unlocking these high-performance computing applications will enable a new era of autonomy in spaceflight.
Low cost, high performance, and reliable space computing is a cross cutting technology for both engineering and scientific needs. The A-SDR+DSP solution is broadly applicable to LEO, GEO and Deep Space environments and a range of spacecraft. The lower cost and faster availability of commercial ICs will allow for spacecraft designers to develop their missions with more certainty and correspondingly less risk.
The Air Force Research Lab (AFRL) notes, “The need for on-board processing will be dramatically greater than the processing performed presently. This is due to the desire for greater levels of on board autonomy…” The Space and Naval Warfare Systems (SPAWAR) and DARPA’s Tactical Technology Office have also been identified as potential DoD customers. The largest potential market is for computing in autonomous automobiles and UAVs, technologies on the verge of wide commercial deployment.