Food delivery robot casually drives under police tape through active crime scene
vice.com
vice.com
>In the video, someone standing near the tape raises it for the robot to go under.
so the robot was being operated by a human, and the human operator was following the instructions of the authorities on scene. it's a fun headline, but a pretty boring story
There are two reasons why it makes sense economically:
- to deliver by moped you need to live in a reasonable commuting distance to the place where you are delivering. The robot driver can operate from a much lower cost of living area, or even a different lower cost country.
- even if you don’t do that and your drivers all operate from a high cost of living area the idea is that your software engineers will keep updating the software and you need less and less human supervision.
The first day maybe all your robots are remotely operated 100%. A month later you have a model to drive the robots automatically on the simple straight bits. What you can do then is to have a driver operate maybe two robots. When one finishes with the simple task it can do on its own it waits for the operator.
The more fully autonomous you can make the software the more robots a single operator can operate at a time, and the beter your margins will ve. This way your level of autonomy is not a binary perfect or nothing proposition but a sliding scale where as you get better you earn more money.
if you're just hiring scooter drivers, you're going to have to make money the old-fashioned way
Perhaps ML can read these scenarios better than humans.
EDIT: to keep the HN value high here's a link to a case where ML does better than humans in reading the scenario - lung cancer detection in xrays.