Thursday, June 9, 2016

Smart Cities Proposal Shows How Hard it is To Envision, Much Less Create, Any Part of a Smart City

Denver’s proposal to the  U.S. Department of Transportation as part of its Smart City Challenge might show why smart city programs are such complicated undertakings, under the best of circumstances. The DoT program will award one U.S. city about $40 million for a smart city project, with an expected award of up to $10 million also provided by Vulcan.

In fact, it probably is a bit of a misnomer to talk of “smart cities.” Instead, there are lot of potential ways intelligence, big data and sensor networks can actually produce valuable outcomes.

But those real outcomes are going to be few and far between for some time, because the infrastructures do not yet exist.

What Denver wants to do first is “establish a robust data management and sharing platform that will connect disparate data sets from multiple agencies.”

That information will be made available in the form of mobile apps and at kiosks, integrating information about  the city’s five car sharing and three ride sharing companies with the bicycle program and the city bus and rail services.

Denver also proposes to electrify taxi and City vehicle fleets and introduce wireless charging (including for transit buses). Useful, but not necessarily “smart.”

The proposed project also will test autonomous vehicles, introducing autonomous elements to fleet and transit vehicles while testing autonomous vehicle business models. Also useful, but not something that will affect consumers.
Denver proposes an integrated data system that would draw real-time information from many sources, providing a detailed picture of travel through the city.

That, in turn, is supposed to support programs that give residents and commuters access to more transportation options, initially. Useful, but essentially an “uber app” (in the sense of amalgamating and integrating other exists information sources, not in the sense of the ride-sharing app).

Eventually, the information system is seen as supporting “smart vehicles” and autonomous vehicles.  to the street grid and to each other, helping to pave the way for self-driving cars. Again, useful, but not something that will be seen or touched by citizens and users.

Some elements focus squarely on making it easier for residents of low-income neighborhoods — especially those without credit cards or smartphones — to connect with ride-sharing services such as Lyft, check out B-cycle bikes or find other ways to fill in transit gaps.

There will be some visible changes: charging stations and information kiosks. Lots of sensors likely will be deployed to gather and update the information base. But most consumers will “see and use” a new transportation app.

The project would convert a significant percentage of the city’s fleet vehicles to electric power. That leads to a greener city, but is it necessarily a “smart city” development?

More electric vehicle charging stations would be added. An additional 15 miles of new bike lanes would be added per year. Both good things, one might argue. But hardly “smart.”

Many of the other outcomes are process related, such as creating a policy and regulatory environment inviting for automated vehicles, or create plans for an 80 percent reduction of greenhouse gas emissions by 2050, supporting efforts to increase bike and pedestrian commuter mode share to 15 percent by 2020, or reduce single-occupant vehicle mode share to 60 percent by 2020.

The project also implements a safety program to reduce and ultimately eliminate vehicle-related crashes, injuries and fatalities.

The point is that the “smart” part of the projects involve collecting and making information available.

Most of the other activities are not necessarily “smart,” but green. There is nothing wrong with that.  But it does suggest how hard it will be to create even one element of a smart city (in this case, smart transportation).

No comments:

Costs of Creating Machine Learning Models is Up Sharply

With the caveat that we must be careful about making linear extrapolations into the future, training costs of state-of-the-art AI models hav...