Julia Programming for Real-Time Flight Control on Embedded Platforms

The Julia programming language was designed to provide good performance and good productivity. Julia's high performance is well known for desktop/application-class platforms, but the large runtime, garbage collector, and potential for JIT pauses have historically made it unsuited for real-time applications or use on extremely constrained hardware. Using the example of a quadcopter controller, we show the deployment of Julia into two different hard-real-time environments: * Running on an application-class processor (Raspberry Pi CM4) attached to the PX4 low-level flight controller through Zenoh, and * Running directly on the microcontroller integrated with PX4 and the NuttX RTOS. We validate the real-time performance of Julia under both compilation pathways and the toolchain used to deploy Julia onto the two platforms. Finally, we show that the controllers achieve real-time performance on real hardware with a flight demonstration on a small quadcopter platform.