Skydio Director Explains Drone Autonomy Stack: Edge-Cloud Split, Maps as World Models, and Limits of End-to-End Learning

In his talk at the AI Engineer World's Fair 2026, Suchet Bargoti, Skydio's Director of Inspection and Mapping, began with a live demonstration rather than slides. From a laptop, he launched docked drones in San Mateo and Colorado, and had one autonomously track a car. This setup introduced his central thesis: scaling "drones as infrastructure" requires solving the operator attention bottleneck and distributing intelligence between edge devices and the cloud.

Bargoti argued that the traditional model of assigning a skilled pilot to each drone does not scale as alerts and tasks increase, creating cognitive load. Consequently, Skydio's autonomy stack keeps immediate safety actions, such as obstacle avoidance and basic tracking, on the drone's edge hardware. Heavier reasoning, long-term planning, and Vision-Language Model (VLM) analysis are offloaded to the cloud. This split relies on high-quality video transmission, requiring optimized encoding to ensure sufficient visual information reaches the server even under bandwidth constraints.

A key mechanism described is "maps as world models." Drones do not rely solely on local perception but integrate building, road, and power-line vector data into a global map for path planning. Crucially, this map is dynamic: when the fleet observes environmental changes, such as new construction sites, these observations feed back into a map synchronization process. The updated map is then shared across all drones, creating a data flywheel that allows the system to self-maintain as the environment changes.