This one:

The challenge becomes managing the systems that manage drones. And maybe the systems that manage the systems that manage the drones.

Researchers and regulators have already recognized that managing future drone traffic will require far more than identification and tracking. Proposed UAS Traffic Management (UTM) systems include services for flight authorization, monitoring, conflict resolution, emergency response, and dynamic rerouting. In many ways the challenge resembles managing a transportation network rather than simply controlling aircraft

One drone is relatively simple.
Ten drones require coordination.
One hundred drones require automation.
One thousand drones require sophisticated management systems.
Hundreds of thousands of drones create an entirely new infrastructure challenge.

(Part two of a series.) For now, most drones still feel occasional. A neighbor flies one over the park. A real estate agent photographs a property. A police department uses one to monitor traffic after an accident. Firefighters deploy one during a wildfire. A stadium uses aerial camera coverage during a sporting event. Somewhere nearby, Amazon is probably testing another delivery route. Individually, none of these seem especially disruptive. Together, they may eventually become something very different. One of the reasons drones still feel manageable is that there are relatively few of them operating simultaneously in most places. The systems remain fragmented, experimental, and somewhat isolated from each other. But infrastructure problems often arrive gradually and then all at once. Roads worked well when there…

Most robot demonstrations, especially humanoid robots,  are carefully staged. The floor is clean. The lighting is controlled. The sensors are calibrated. The battery is fully charged. The software has been updated. The Wi-Fi works. Someone nearby knows how to reboot the system if things suddenly become “unexpected.” Then the robot walks across the stage carrying a box, folding a towel, or waving at the audience, and everyone focuses on the machine itself. But the robot is often the easy part. What matters is everything around it. That may be one of the biggest misunderstandings in the public discussion about robotics. We tend to imagine the machine as an independent mechanical worker — something like Rosie from The Jetsons — operating largely on its own intelligence…