Slashdot Mirror
Kroger Will Use Autonomous Vehicles To Deliver Groceries (theverge.com)
Starting this fall, Kroger will partner with driverless car company Nuro to deliver groceries using its autonomous vehicles. The Verge reports: A pilot will be rolled out to a yet-to-be-announced city later this fall. To start out, Nuro will use a fleet of self-driving test vehicles with human safety drivers to make deliveries for Kroger's grocery stores. Customers can track and interact with the vehicles via a Nuro app or Kroger's pre-existing online delivery platform. But if Nuro's human test drivers don't get out to help you, don't be mad because in our driverless future, we all need to pitch in and unload our own groceries.
Nuro is still tweaking its user experience, but for now it will go something like this: customers can place an order through Kroger's online delivery portal or using Nuro's forthcoming app. Kroger workers will load the items into Nuro's temperature-controlled compartments, at which point the vehicle will drive autonomously to its destination. Customers can track the vehicle throughout the trip using the app, and once it arrives, will need to meet the vehicle at the curb or in their driveway -- in other words, no more door-to-door service. They can use either a PIN code or some other verification system to retrieve their delivery. Nuro was reportedly working on a facial recognition system, but has since tabled that.
Nuro is still tweaking its user experience, but for now it will go something like this: customers can place an order through Kroger's online delivery portal or using Nuro's forthcoming app. Kroger workers will load the items into Nuro's temperature-controlled compartments, at which point the vehicle will drive autonomously to its destination. Customers can track the vehicle throughout the trip using the app, and once it arrives, will need to meet the vehicle at the curb or in their driveway -- in other words, no more door-to-door service. They can use either a PIN code or some other verification system to retrieve their delivery. Nuro was reportedly working on a facial recognition system, but has since tabled that.
I wonder how many failed deliveries they will have when people don't come outside to unload their own groceries.
Greed is the root of all evil.
but it seems weird to be ready to go ahead and start sending driverless cars (with safety riders, I know) all around at people's orders.
It's even stranger that a driverless vehicle that is transporting only *things* would be tested with a safety driver...I think I've seen companies developing, from scratch, driverless delivery vehicles that are never intended to carry a human inside. This makes sense for a lot of reasons.
Autonomous cars that transport people must satisfy two major safety requirements: 1) not kill or injure the people they are transporting and 2) not kill or injure other people on the road.
With delivery vehicles, you only really have to worry about number 2). That should make the job significantly easier. You don't have to have some sort of AI "ethics" that has to judge whether to protect the people in the car or the people outside the car, it can always just "sacrifice itself". Drive over a bridge to avoid hitting a pedestrian? OK.
Furthermore, cars that don't have to carry people are simpler to design structurally...they don't have to protect the passengers in case of a crash for example. These things can also be slower. Limit them to say 30 km/h, just schedule the deliveries properly, and you reduce a lot of potential for accidents. Etc.
So testing things first with a safety driver seems pointless, i.e. it's basically development of the wrong type of vehicle.
Alright, I should've read TFA first...but hey, this is Slashdot. Nuro is the company developing the driverless, passengerless, never-intended-for-a-human delivery cars I was thinking about. Which makes it even stranger why they are doing a different class of self-driving vehicle with a safety driver first.
People who are handicapped or otherwise unable to unload groceries may be unable to use this, although I expect there'll be an option to pay extra for a human to unload the groceries for you. Eventually there'll be an unloader robot inside the van, which can also load the van (rather than having a human load it). Endpoint tasks will be the killer app for robots, which are normally stuck in warehouses and factories; since if you have a human driver you might as well just pay that human to do the task at the endpoint, i.e. pay a professional to also drive there. Once we have fully-automated go-there-and-do-something systems, that can replace a huge swath of human jobs normally not considered for automation (plumbers etc.)
Corruption is convincing someone that the selfless ideal is the same as their selfish ideal.
When I was young, the smartly-uniformed milkman delivered all sorts of things up and down the garden path, and while he was doing that the bored horse moved the milk-float (some yards behind it) to align with the next house that had regular orders. People forget what was possible in a less techno-mad world.
With delivery vehicles, you only really have to worry about number 2). That should make the job significantly easier.
The majority of failure modes where a car kills its occupants also apply to killing individuals outside the vehicle. This includes vehicle to vehicle collisions, loss of control (traction), failure to recognize obstacles, insufficient time to react, other driver's unpredictable actions, and more. While you are correct that it does take some failure modes off the table, most of them still remain in some form or fashion.
You don't have to have some sort of AI "ethics" that has to judge whether to protect the people in the car or the people outside the car, it can always just "sacrifice itself".
That presumes it has sufficient situational awareness to make that possible. There is no evidence we have reached that happy state of affairs yet. Also bear in mind that people are not the only thing that need protecting. While lives should be paramount we don't need cars causing massive property damage either.
Limit them to say 30 km/h, just schedule the deliveries properly, and you reduce a lot of potential for accidents.
There isn't a single road in my town where the speed limit is as slow as 30km/h so all that would do is screw up traffic.
So testing things first with a safety driver seems pointless, i.e. it's basically development of the wrong type of vehicle.
Explain to me how you expect to develop a vehicle that can drive safely around humans before we've developed the robust logic and sensor systems to ensure they can drive more safely than a human. A vehicle that cannot keep itself safe reliably really shouldn't be on the road. Whether or not there is a human on board doesn't really change that fact. And since there isn't any way to actually test these devices safely around humans without having a human driver to step in when needed you need to account for that fact.
What do you mean? Hypocrony or ironocrisy?
#DeleteFacebook