Stanford Robotics Seminar ENGR319 | Autumn 2026 | Skydio: Autonomy for a Family of Flying Robots

Skydio drones are used extensively in public safety for crime prevention, emergency response, and situational awareness. Examples include locating injured individuals, monitoring crime scenes, de-escalating tense situations, and providing first responders with critical real-time information, often arriving at scenes faster than traditional emergency services.

Skydio drones employ a multi-sensor approach that integrates data from various sensors and cameras. They can intelligently identify and exclude GPS signals from obstructed satellites, building a more accurate and reliable position estimate even in environments where standard GPS is degraded.

Skydio's autonomy is driven by sophisticated AI, including deep learning models for object tracking (like the 'Shadow' feature), real-time environment mapping, and advanced navigation algorithms. They also leverage powerful onboard computing and cloud-based processing for complex tasks.

Skydio emphasizes a 'fly-train-evaluate' loop, where real-world data is continuously collected, used to retrain and improve their AI models, and then redeployed. This iterative process, combined with robust hardware and cloud infrastructure, ensures their systems are always up-to-date and performing optimally.

The Skydio X10 utilizes a 'Megadock' system, a specialized docking station with a robotic arm. This allows for fully autonomous, ground-based launch and landing, freeing the drone from carrying the weight of landing gear and optimizing its flight efficiency and endurance.

Skydio is moving towards a model where a single operator can manage multiple drones. They are developing systems for high-level command and control, allowing operators to coordinate fleets through high-level instructions rather than direct, one-to-one piloting.

Skydio drones are water-resistant and can fly in rain, though certain features like obstacle avoidance may be temporarily disabled. However, operations are limited by severe weather, particularly high winds, and the system will alert users to avoid flying in conditions that exceed safety parameters.

Skydio secures infrastructure by deploying autonomous drones from docking stations that patrol designated areas 24/7. These drones can detect unauthorized personnel or vehicles and alert operators, with future plans for custom object detection (like specific equipment or platforms) using open-vocabulary models.

Skydio has invested heavily in a 'Transparency Dashboard' feature. This allows every police department deploying their drones to publish detailed information about each flight, enabling public oversight and trust in how the technology is being used.

Skydio sees large language models (LLMs) as a key future technology, enabling more natural language interaction with drones. Their goal is to allow users to command drones with high-level spoken or text-based instructions, making drones feel like an extension of the user.