What Military Drones Can Teach Self-Driving Cars About Remote Supervision Safety
Key Takeaways
- ▸Self-driving cars frequently struggle with routine situations like construction zones and unpredictable pedestrians, requiring remote human supervision to prevent crashes or traffic-blocking freeze events
- ▸The U.S. military has been remotely supervising unmanned aerial vehicles since the 1980s, experiencing numerous early accidents due to poor interface design, inadequate training, and communication delays
- ▸Some U.S. self-driving companies now employ remote operators based in the Philippines, raising concerns about applying lessons from decades of military UAV research
Summary
In an opinion piece for IEEE Spectrum, Professor Mary (Missy) Cummings from George Mason University draws critical parallels between military drone operations and self-driving car safety challenges. Cummings, a former Navy fighter pilot and early UAV researcher, argues that the autonomous vehicle industry is repeating mistakes the U.S. military made in the 1980s and 1990s with unmanned aerial vehicles (UAVs). Self-driving cars frequently struggle with common scenarios like construction zones, school buses, power outages, and unpredictable pedestrians, often requiring remote human supervisors to intervene when the vehicles freeze or behave unpredictably.
The article highlights that self-driving companies now employ human "babysitters" to remotely supervise autonomous vehicles, with recent revelations showing some U.S. companies use operators based in the Philippines. This approach mirrors the military's early UAV remote supervision model, which experienced numerous accidents due to poorly designed control stations, insufficient training, and communication delays. Cummings spent thousands of hours in the 1990s researching how to improve UAV remote supervision interfaces, generating valuable insights about safely managing remote operations.
The core argument emphasizes that decades of military research on UAV operations offers a deep knowledge base that could significantly improve autonomous vehicle safety. Issues such as latency in communication, operator fatigue, interface design, and training protocols have been extensively studied in military contexts. By applying lessons learned from military drone operations—where failures were thoroughly analyzed and design improvements systematically implemented—the autonomous vehicle industry could potentially avoid repeating costly and dangerous mistakes, ultimately leading to safer deployment of self-driving technology on public roads.
- Military research on UAV remote operations has generated extensive knowledge about communication latency, operator fatigue, interface design, and training that could improve autonomous vehicle safety
Editorial Opinion
This cross-industry perspective offers a sobering reminder that technological hubris can be dangerous—and expensive. The autonomous vehicle industry's apparent failure to systematically apply decades of military research on remote supervision represents a missed opportunity to leverage hard-won safety lessons. If AV companies continue deploying remote supervision systems without incorporating the military's institutional knowledge on operator fatigue, communication latency, and interface design, they risk not only repeating past mistakes but doing so on public roads where civilian safety is at stake.
