Future planetary exploration missions demand high-performance, fault-tolerant computing to enable autonomous Guidance, Navigation, and Control (GNC) and Lander Vision System (LVS) operations during Entry, Descent, and Landing (EDL). LVS, demonstrated on NASA’s Mars 2020 mission, provides real-time navigation updates through onboard vision-based systems. Current research evaluates the deployment of LVS and GNC algorithms on next-generation multicore processors—including HPSC, Snapdragon VOXL2, and AMD Xilinx Versal—to achieve real-time image processing, object tracking, and trajectory computation with improved power efficiency and reliability.

Benchmarking results show up to 15× speedup in LVS Forward FFT and correlation steps on VOXL2 and Versal, and over 250× speedup in GFOLD, a convex optimization-based GNC algorithm. Beyond raw speed, fault tolerance is enhanced via HPSC’s dual-core lockstep and multicore redundancy, incorporating a real-time majority voting mechanism (ARBITER). This mechanism supports both “static” routines like GFOLD and “dynamic” closed-loop control scenarios such as EDL or Attitude Control Systems (ACS).

The GFOLD pipeline was decomposed into three computational stages to analyze fault observability and identify the most error-prone phase. Targeted fault mitigation at these critical points improves resilience to radiation-induced errors and hardware faults. This study presents a scalable, energy-efficient architecture capable of supporting mission-critical reliability for future missions like Mars Sample Return, Enceladus Orbilander, and Ceres Sample Return. It offers actionable insights into balancing speed, energy efficiency, and robust fault protection in next-generation autonomous space systems.