In regards to your questions about the distribution chain and the “last mile” in particular:
I’m certainly not as well-versed on the innovations of distributors, but given that autonomous semi trucks are well on their way to commercialization, and warehouses such as Amazon’s (linked within my article) have already embraced heavy autonomy, it stands to reason that advances in the last mile (delivery to the consumer) will happen rapidly. This would be my (again, uneducated) take on how the last mile might develop over the next five years, starting around 2020 based on the timeline of my original article:
Autonomous delivery to your preferred location: This one is already criminal for not being a reality, but if you order something that is scheduled to be delivered during the day, you can request (or change your request) so that the vehicle delivers to your place of work rather than your home. What makes it a more viable option than what’s possible today is the ability for AVs to recalibrate and optimize driving schedules on the fly, which is not something human-powered logistics are efficient at handling. The delivery vehicle would be partitioned so that only your package would be unlockable in the event you or some other human party were taking delivery of it directly. This is not only a rather obvious solution, but is also a patented design Google has submitted, so it seems like a fair conclusion to make in the near-term.
Proliferation of “lockers”, to use Amazon’s marketing term: These safe deposit boxes for packages already exist in cities like NYC, but as autonomous vehicles become a reality, the Amazons of the world will have more incentive to plant (or subsidize the planting of) more lockers, as well as enhance them to the point that a vehicle could effectively drive into or over top of one and dump its contents accordingly. In the same vein, consumers with connected homes could grant temporary access to the delivery vehicle so that it could, for example, open the garage door, drop the package, and re-close the garage. Eventually you might purchase your own “locker” and get the same effect, since your garage will likely become a more properly-walled room of your house once you give up vehicle ownership. There’s also an obvious pricing scale to be seen there: those who can afford their own locker pay more for increased security and convenience, while those who can’t get will cheaper delivery costs with the trade-off of having to utilize a “public” package locker in their building or down the block.
The integration of delivery drones into the vehicle design: For those hard-to-reach places, or for scenarios where it would be more efficient to finish a delivery as the crow flies, having a delivery drone atop or inside of the autonomous vehicle makes a lot of sense. A few folks have heard that Amazon is already testing drones as delivery vessels, and it’s plausible to assume they would pair such solutions with larger vehicles that could more easily handle the long, redundant part of the haul. The AV gets packages to your neighborhood, and the drone takes the packages from the AV to your doorstep, or locker, etc.
In regards to your other questions, my answer is that technology improves. This technology in particular has improved at a startling pace. You can find many an article about autonomous vehicles being unable to cope with poor visibility conditions. Those articles and opinions have a short shelf life, as many other commercial technologies have proven to their nitpicking naysayers over time.
To this day, electricity still goes out in extreme weather conditions. I don’t see anyone publishing articles clamoring for a return to whale oil. Autonomous vehicles could foever remain inoperable during low-visibility conditions and still be a vast improvement over human-driven traffic.