Advances in AI, remote sensing, imaging and power systems mean that self-driving cars are coming, and sooner than you think. But we're blind to how embedded the car is in our society. We let it take over up to 60% of our real estate, emit 40% of greenhouse gasses, literally billions and billions of hours of our time, and human driving kills 45,000/year in the USA, a million worldwide, and injures far more, with accidents costing up to 4% of the GDP. And yet solving all this, changing all this is now a fairly tractable engineering problem. This calls for an Apollo project, which we do not because it is easy, but because it is hard.
Such a massive change in transportation means major changes in how we live and how we want to design our cities. In particular it means the end of public transit as we know it in under 15 years, but transit planners are still planning further in the future than that. What else does this, and other trends mean for the city of the future? (And yes, city of the future predictions are always risky.)
Here is a more detailed summary
A concept that was science fiction just a few years ago has rapidly come closer to reality -- the computer driven vehicle which is fast, efficient and most important, highly accident resistant. Human driven cars kill 45,000 per year in the USA, a million worldwide and injure millions more. Accidents cost around 4% of the GDP (NTSB). Congestion wastes 3.9 billion hours/year in the USA, and driving even more.
Great argument for transit, but as you watch videos of competitors in the DARPA grand challenge, where prototype driverless cars zoom down curvy roads, pass one another and move safely on urban streets, obeying ordinary signs and yielding to other vehicles, it becomes clear another answer is coming and many expect it to be here in under 10 years.
The self-driving car has many advantages aside from not killing millions of people. They will also be faster, rarely cause congestion, park themselves densely and travel faster without ever needing to slow for anything, and allow occupants to read or work rather than watch the road. Remarkably, they will also be vastly greener than existing cars and current figures for transit systems.
The computer-driven car enables something even more interesting -- the on-demand robot taxi. This taxi may be a commercial vehicle or a private vehicle not currently being used by its owner. This taxi can be the right vehicle for the trip. Most of the time it will be a single passenger, lightweight, short-range electric vehicle that's green as can be. Where needed, it can be a 12 passenger van or a long-range gas-powered SUV. Every trip is point to point, with few (if any) stops, in the right-sized vehicle for the job. Special vehicles for those with special needs will be immediately available.
In turn, this frees up private individuals to not own a vehicle at all, or to own the vehicle that is right for most of their trips -- in cities, probably a 1-2 passenger short range electric vehicle narrow enough to fit two to a lane. (Private owners may also elect to own a vehicle which will generate good revenue for them when not in use.)
While mass transit such as commuter rail can be more efficient when fully packed at rush hour, it is less efficient most of the day and fails the vast majority of other comparisons.
This suggests that there will be radical changes in the nature of transit by 2020. Current forms of transit may become obsolete. Private right of way will no longer make sense, or gain much advantage. Whatever happens, transit planning beyond 2020 needs considerable rethinking, and even planning to 2015 must not be based on current assumptions.
This talk will show videos of real computer driven cars in action, present figures for costs and energy use and examine some of the other implications of such vehicles to urban and rural transportation and urban design, as well as outline likely incremental steps to the final result.