JPL is exploring semiconductor technologies to enable dramatic (orders of magnitude) increases in processing capabilities to meet a wide spectrum of JPL needs and requirements. This includes autonomous operation, machine learning and artificial intelligence, high-speed wireless infrastructure, robotics, and multiple sensor input and fusion. Areas of research are focused on a wide array of technologies that can be used across all of JPL’s space environmental conditions and mission classes.
Radiation-Hardened Computing Hardware
This focus develops high-performance, low-power computing capabilities for use in extreme environments, where reliability requirements are paramount and power is at a premium. Initial developments have provided greater than two orders-of-magnitude increases in processor throughput, memory bandwidth, and input/output bandwidth compared to the current state of the practice in radiation-hardened space electronics, at similar power levels, with an order-of-magnitude improvement in radiation and fault tolerance.
Advanced Software Technologies
Current software technology does not allow the generation and validation of the tens of millions of lines of complex parallel codes required for the autonomous systems envisioned for next generation space systems. New approaches are being developed that will enable the rapid, low-cost generation of the verifiably correct codes required for these future systems.
Advanced Commercial Chipsets and Platforms
Mobile phone and next generation automotive system on a chip (SoC) platforms, based on highly scaled (less than 16 nm) commercial complementary metal-oxide semiconductor (CMOS) technologies, provide multiple giga-ops of processing power, often with less than 1 Watt of total power consumed. These platforms are based on heterogenous computing architectures (CPUs, GPUs, and DSPs all integrated on one piece of silicon). JPL is partnering with these chipset providers to adapt and leverage existing solutions to meet next generation mission and project opportunities.