Traditional digital avionics systems have included ground and flight segments where the most intensive processing is often done on the ground for applications like optimization compared to more immediate lower latency requirements for flight control and flight management. Combined ground and embedded avionics are however being asked to do more based on exciting new uses of airspace from UAS to space- based ventures as well as expanding commercial, military and general aviation. New features may range from interactive agent and assistant features that enhance flight optimization and planning to new cybersecurity features for the post-quantum world.
To meet new demands, much like other ground transportation with intelligent systems infrastructure, avionics will not only require more scale-up, but also scale out (networking) and co-processing in the cloud to remain secure and to integrate new AI and machine learning features. The drive to integrate AI and autonomous features is a combined opportunity and challenge that can be met using high-performance computing industry GP-GPUs (General Purpose Graphics Processing Units) and QPUs (Quantum Processing Units). This is similar in concept to GP-GPUs used today for online AI services such as LLMs (Large Language Models). Both GP-GPU and QPU can complement avionics software systems which can benefit from machine learning, optimization and post-quantum security, best provided by cloud co-processors. This exciting new architecture of QPU (Quantum Processing Units) and GP-GPU will be reviewed and examples demonstrated and explained.
This tutorial will introduce you to the challenges, opportunities and the latest hardware and software tools and practices for hybrid computing using parallel GP-GPU co-processors using CUDA as well as cloud-based QPUs using CUDA-Q. You will emerge with a fundamental understanding of concepts and theory, both proven, and more emergent to consider for future projects.
