When it comes to autonomous transportation, everyone wants to talk about the vehicles. How will they work? Will they be safe? Will anyone dare ride in them? What infotainment features might they have? These are important questions, and the public’s fascination with the autonomous pods that have been proposed, from rolling hotel rooms to Jetsons-like aircraft, is understandable.
However, the real social, economic, and environmental value that autonomous mobility might bring have more to do with the services that it will provide than with the vehicles themselves. The prospect of dramatically reduced traffic and pollution, and affordable accessible transport for all regardless of age, income, or ability, will only be realized if autonomous vehicles (AVs) are deployed in services that are shared, electric, and integrated with public transport. And in each of these areas the services will need to be carefully orchestrated.
Multiple studies have shown that if AVs are used primarily as personal vehicles, traffic will only get worse. Far worse. There are several reasons for this. First, it will be cheaper for the vehicles to return home than to park, at least doubling the distance cars travel each day. Second, people will be so comfortable in their private pods that they won’t mind long commutes, leading to more traffic and more low-density suburban housing. Third, there is concern that more people will use these vehicles in place of public transport, putting more vehicles on roads and further cut transit use.
It doesn’t have to be this way. There are also a number of studies, including a simulation Bestmile performed using a virtual fleet representing Chicago’s 31,000 daily taxi trips, showing that highly optimized pooled services could exponentially reduce traffic. Advanced orchestration algorithms can enable these services to be ultra-efficient, with little change in passenger convenience (ride times, wait times, etc.), and they can be economically sound for operators.
The algorithmic computing needed to enable a shared fleet to deliver predictable levels of service is extremely complex. To offer convenient pooled rides at scale requires processing hundreds or thousands of ride requests in real time, sending the right vehicle to the right place at the right time. The algorithms must consider the locations of every ride request, the number of passengers, the available vehicles, the vehicle locations, battery levels, traffic, construction, weather, geography, allowable ride times and wait times, future demand and more.
It is complex, but it can be done. In addition to our own research, studies by MIT and the University of Texas came to similar conclusions. Our study found that 400 shared vehicles could do the work of 2700 Chicago taxis with predictable ride times and wait times. An MIT study found that a fleet of 3,000 taxis could meet 98 percent of demand served by New York City’s 13,000 vehicles with an average wait time of 2.7 minutes. UT found that one shared autonomous vehicle could replace 10 personal autos with wait times between a few seconds and five minutes.
Integrated with public transport
With cities expected to grow by 10-15 percent in coming decades, shared autonomous fleets must complement, not compete with public transport. The impact of peer-to-peer ridehailing services on public transport is significant. They have increased traffic by some 180% and most riders use the services in place of staying home, walking, biking, or public transportation.
The ridehailing giants have announced partnerships with some cities to incentivize travelers to use the services to reach transit stations.
But let’s face it. There is no way to reduce urban congestion that does not include efficient, affordable mass transit. It is time for people, especially in the U.S., to understand that public transit is an investment in economic vitality, not a cost that drains city coffers and taxpayer wallets. Private businesses and homeowners are among the chief beneficiaries. In the United States, for example, a $10 million investment in public transportation generates $32 million in increased business sales, and property values for homes near transit hubs service are 42% higher.
Integrating new mobility services with public transport means more than investing in transit. It means designing services so that they synchronize with public transit schedules to remove friction from journey planning and delivery. This is one of the functions that algorithm-driven autonomous services can offer—not only synchronizing with transit schedules but communicating with public transportation networks in real time and adjusting to delays, larger or smaller than usual passenger counts, system disruptions and other fluctuations that can impact travelers’ convenience. Convenience, after all, is the highest factor in travelers’ decisions about transport options—higher even than price.
Electric vehicle sales nearly doubled between 2017 and 2018, and that trajectory is expected to continue. While still a small fraction of global vehicle sales, virtually every auto manufacturer in the world is betting on an electric future and investing accordingly. Just as there is no practical way to reduce congestion without some form of affordable mass transit in and out of cities, there is no practical way to improve air quality without the widespread adoption of electric vehicles. Reducing gasoline powered vehicles by a factor of 10 would certainly be a welcome reduction, but as GM CEO Mary Barra has called for, the best-case outcome of autonomous mobility is “zero crashes, zero emissions, and zero congestion.”
Autonomous mobility services are well suited to electric vehicles for a few reasons. The vehicles are easier to fuel without a human, the wear and tear on the electric drive train is much less than their gas-powered equivalent, and the fuel is less expensive. It is no wonder than most of the autonomous service trials in the world feature electric vehicles (puzzling exceptions are Uber and Lyft).
“Range anxiety” is one concern for autonomous electric fleets, and battery range is another factor that fleet orchestration algorithms consider. Does the vehicle have enough power to complete a mission? Does it have enough to complete the next mission, or to get to a charging station? Where is the nearest charging stations and is there a wait time due to other vehicles being charged? Fleet size has a direct correlation on vehicle range requirements, and battery range is a factor in designing an optimal fleet size.
There has yet to be a fully autonomous on-demand service deployed at scale. There are regulatory and technical challenges still to overcome. The technology is not ‘there yet’ for full autonomy. And regulators will need to determine acceptable safety criteria. However, there is widespread optimize that these challenges will be eventually overcome. While vehicle technology and policy are important, equally important is to understand how these fleets can perform when working together. For autonomous fleets to achieve the “zero congestion” goal of the autonomous dream, the fleets will need to be highly orchestrated.