OK, long distance = 5kms.
Running is more like brisk walking, which is fine and safer. Story here.
The key here is that the robot was untethered – the battery that powers it was carried. That means on one charge such a robot can make deliveries of light objects to neighbouring suburbs. That it then needs a recharge is not an issue – swap batteries or just wait, robots aren’t paid by the hour.
Expect a mix of delivery bots in the future. Wheels are more efficient if the terrain allows it, let’s call them Rollers. And we will have Runners that can cope with other scenarios, and take shortcuts, and climb stairs. Don’t be surprised if both types are speed limited to 5kph. For more urgent deliveries we will have self-driving cars, we could call them Rapids. They could rendezvous with the other bots. For example, an apartment building could have its own Rollers or Runners that meet the Rapid on the street and take over the delivery.
Because of the many scenarios that non-human deliveries can involve, the winner won’t be the robot manufacturers, but the logistics company that controls the network of cars and robots. One network will be more efficient than many, and first-mover advantage could be enough to get to that monopoly. Code costs nothing to run, but it will need human oversight, especially for manually guiding cars and bots that get stuck, and protecting them from bot-haters and thieves. A single network would mean that stolen bots have no value elsewhere.
Tip: invest in the first company in each country that looks like pulling off such a network venture.