The limiting factor was Operator attention and issues with an environment.
In a closed, mapped environment like a campus with minimal street crossings. The robot can make its way to the restaurant, get the delivery and make the delivery, with out operator input or attention… even if people block the robot, it can navigate around and interact. After a couple failed attempts, it alerts an operator and then manual action may occur.
Some situations were a bit more complicated. I’ve had to navigate 4 robots, all at street crossings with different types of traffic. The safe thing to do is, take care of them one at a time, even if a couple robots miss the light.
Once a crossing light changes and things look safe, we would just initiate the crossing. The robot can navigate on its own.
This is how I assumed the ones near me operate - they are mostly independent and a live person takes over if it gets in trouble or encounters a tough situations. I can imagine one person being able to operate more than 5 if they have solid pathing.
In situations like this it is possible for an operator to manually organize the robots.
Before I left we were making great strides to allow 1 operator to be able to keep tabs on up to 5 robots at a time in certain neighborhoods.
Campuses, which are fully and thoroughly mapped, can probably have 1, maybe 2 operators at a time. Just watching and interjecting when issue arises.