John Leonard (MIT) did polls on driverless cars at the 2018 NSF NRI PI meeting. Here I summarize the poll results. The people participated in the polls are the robotics professors in the U.S. They would be considered as the experts in autonomous driving. Most importantly, most of they don’t have any financial tie with the auto or autonomous driving industry.

When do you think you will be able to ride in a fully driverless car from MIT to Boston airport.
2018-2020 — 6
2021-2025 — 13
2026-2030 — 30
2031-2035 — 10
after 36 — 10

When do you think you will be able to ride in a fully driverless car from MIT to Boston airport in all weather conditions (airplane-take-off OK conditions)
2018-2020 — 0
2021-2025 — 3
2026-2030 — 9
2031-2035 — 8
after 2036 — 38
Not in your lifetime — 12

Which is the most difficult technical challenge for self-driving car
Adverse weather — 16
Interacting with other people in other vehicles — 13
Interacting with other people outside of vehicles — 8
Maintain HD map — 1
Road construction — 0
Reducing the cost of sensors and computing — 1
Perception algorithm — 25

Which industry sector will be most disrupted by self-driving car
Taxi, Uber, etc — 8
Long-distance trucking — 32
Short-haul trucking — 0
Urban real estate — 3
Car dealer — 8
Insurance — 11

The polls were done very quickly during John Leonard’s talk. Not much time was given to the people who participated. Anyhow, the majority think L5 would take 20+ years. L4 commercialization is in 10 to 15 years range. The most challenging problem is still perception. Trucking business would be most affected, even more than taxi and insurance. But the majority may not be right all the time.