In the last couple of years, I’ve attended meetings focused on such topics as smart cities and vehicles of the future. The two are very closely connected. Proponents of autonomous vehicles (AVs) are striking fear into the hearts of city managers who worry these AVs will lead to the replacement of the private vehicle, a major source of revenue for cities in taxes and parking fees. They also worry AVs will take riders away from mass transit and clog the streets searching for passengers or places to recharge their batteries.
A big problem these vehicles face is recharging. Practically every AV proponent is assuming that the vehicles will be electric. In all the seminars I attended, I have never heard mention of how gas station robots would fuel autonomous cars.
Moving from gasoline-based transportation to electric-based transportation will require some serious infrastructure building. We might have enough electric power to charge cars, but the challenge is making the power available at the right place and at the right time. Who gets priority in the afternoon, air conditioners or electric cars? Where do we stash all the cars while they recharge, in converted parking garages?
But it’s not the electrical power infrastructure we want to discuss, it’s the smart city intelligent traffic system. Most larger cities have been building their intelligent traffic systems (ITS) by connecting traffic lights to a central facility and making them smart. Cameras at intersections can now monitor traffic and provide feedback to switch smart lights to help traffic flow more smoothly. These networks use city-wide fiber optic networks to connect intersections and move data.
Within the AV proponents, there seem to be two factions. One, predominantly based in Silicon Valley, believes that artificial intelligence (AI) can do all the driving. Another group, including many traffic engineers, believe the vehicles need to communicate with each other and the city infrastructure.
The AI believers should be having second thoughts. Several recent accidents show that AI can be fooled or cannot react quickly enough. In four cases, a car operating in autonomous mode crashed after a vehicle they were following swerved to avoid a blocked lane. Telemetry showed most of them actually sped up after the vehicle in front made an emergency avoidance maneuver and then crashed into the lane blockage without reacting. Two times the lane was blocked by a fire truck responding to an earlier emergency.
The ITS notes that if vehicles have vehicle-to-vehicle (V2V) and vehicle-to- infrastructure (V2I) communications, a vehicle stopped in a lane or swerving to avoid an obstacle could warn other vehicles of the lane blockage. If V2I is available, video surveillance would spot blocked lanes and warn all vehicles, even providing instructions on how to detour around the problem.
Providing V2V and V2I is complicated and requires cooperation among many groups, including cellular providers. Everybody assumes that V2V and V2I will work over cellular radio, often using it as another reason to rush 5G cellular, the latest proposal for cellular upgrades. 5G makes many assumptions, including the deployment of small cells, cellular antennas typically mounted on traffic lights, street lights or utility poles that serve relatively small geographic areas.
Both V2V and V2I require every car be equipped with cellular radio dedicated to the operation of the vehicle. To be useful, there must be standards as to what kinds of data are shared and how. All vehicles will probably need the equivalent of aircraft transponders that would broadcast the vehicle’s data to the infrastructure and other vehicles. A city ITS would have to digest this data and provide real-time feedback to cars.
When it gets down to this detail, cities get worried. Are they expected to foot the bill for all these systems? Today most have already developed much of the ITS infrastructure, but the additional data will probably require even more backbone capacity. Smart cities, we always point out, run on fiber. Adding autonomous vehicles to the current ITS programs and all the other dozen or more smart cities communications needs is a massive undertaking.
Fortunately, none of this will happen quickly. Sure, Silicon Valley is ready to go. The California Department of Motor Vehicles has issued more than 50 permits for on-street testing of AVs, but the recent crashes have put a damper on testing. It will take time to get going again. And it could take years for current groups looking at V2V and V2I to create standards.
Finally, at one meeting a speaker brought up an important point. In California—and the United States as a whole—the average vehicle is over 11 years old. If we started selling only autonomous electric vehicles tomorrow, it would still take decades before a majority of vehicles were autonomous.