Introduction
Mammals split their nervous system into two very different kinds of processing. The cortex handles deliberate, high-level work: perceiving the world, building a model of it, and deciding where to go. It is powerful, but slow, and it can be wrong. The involuntary nervous system (spinal reflexes, the brainstem, the autonomic system) works without conscious thought. It keeps you balanced and pulls your hand off a hot stove before you feel the pain. It is fast, bounded, and always running.
The Branes.AI platform follows the same split:
| Repo | Biological analogue | Responsibility | Time scale |
|---|---|---|---|
cortex | Cerebral cortex | Perception, world building, path planning | ~5–100 Hz, best effort |
reflex | Brainstem, spinal cord | Stabilization, trajectory tracking, optimal control, fault recovery, safety envelopes | ~100 Hz–8 kHz, hard real-time |
racing-drone | The whole organism | Mission and flight controller assembled from both | — |
Design principles
Section titled “Design principles”- Determinism over throughput. Every update has a bounded, constant execution time. Nothing on the hot path allocates, blocks, or throws.
- The reflex layer never depends on the cortex to stay safe. Commands from the cortex are setpoints to track, never something the vehicle needs in order to stay stable.
- No dynamic reconfiguration. Gains and limits are fixed when a controller is constructed. Changing them means building a new controller through the lifecycle, never mutating one in the hot path.
- Type-generic numerics. Controllers are templates over the scalar type, so mixed-precision and custom-arithmetic studies use the same code that flies.