autonomous mobility

A Course Correction for Autonomous Mobility

In the transportation world, there has always been consensus that there is nothing worse than the Single-Occupancy Vehicle (SOV). Well, now there is.

The Arrival of the Zero Occupancy Vehicle

Afew years ago I wrote an article about the Zero Occupancy Vehicle (ZOV) — how gridlock will be augmented by empty autonomous vehicles (AVs). There is simply nothing about a vehicle being autonomous that makes it more likely to achieve higher occupancy. In fact, the current trajectory of AV deployment roadmaps and our transportation policy response ensures its average occupancy will be lower.

Why Occupancy Matters

Car occupancy has always been pretty stable, with an average of 1.1 occupants at peak travel times. Outside of a limited number of areas where there are high-occupancy vehicle (HOV) lanes or toll discounts, little effort has been made to incentivize a behavioral change. A small upward change of average vehicle occupancy (say 1.2 or 1.3) would completely eradicate traffic congestion in our cities. Similarly, a small downward change (say 1.0 or 0.9) will be devastating.

Vehicle throughput versus person throughput

Now consider our driverless utopia — a fleet of cars that we can summon at the touch of a button. Consider that these cars will be empty most of the time — repositioning, waiting, traveling on deadhead return trips. Consider that travel demand spikes predictably 2 times per day, that cars are idle 95% of the time, and that most people live in suburbs or exurbs. How do empty AVs sitting in traffic impact our largest transportation problem? They make it much worse. Faith in driving pattern algorithms alone reducing traffic is misplaced given the relative sensitivity of congestion to occupancy versus to driving skill. And so a correction of priorities is required.

Level 5 is the Wrong Destination

Today the dominant framework for understanding the development of AVs is the SAE Levels of Automation. It has been wholeheartedly adopted by everyone from auto companies to the highest levels of transportation authorities, including the US Department of Transportation (USDOT). It is a staple of almost every presentation you will see on AVs. And level 5 autonomy is presented in our media as the pinnacle of disruptive mobility — as though we are more likely to carpool simply because a car is autonomous.

SAE Levels of Automation

However, it’s important to note that SAE stands for the Society of Automobile Engineers. The prism through which we view the development of autonomous mobility is an engineering framework. It simply describes a roadmap for the technical features of a vehicle, as they progress through stages of development. Its relevance is limited to auto company development teams and their competitive positioning, not to transportation authorities and their planning priorities.

We already have a long-standing hierarchy with which to guide our transportation planning decisions. It prioritizes pedestrians and cyclists over transit, which it in turn prioritizes over cars. The SOV sits appropriately on the bottom rung. And yet now our traditional frameworks and rules are being bypassed in an attempt to accelerate the introduction of autonomous mobility. AVs are not discussed in relation to their level in a hierarchy. This risks diverting our focus from planning a smart, efficient transportation ecosystem.

Which way to intelligent autonomous mobility?

A Crossroads for Transportation Authorities

Governments today compete over deregulation measures for AV deployments, creating an express lane for the SAE vision. USDOT’s latest policy document, now a “de facto global standard” for AV deployments, has prioritized modernizing or eliminating “outdated regulations that unnecessarily impede the development of automated vehicles”. It describes “voluntary technical standards” and “voluntary guidance”, as well as an AV “exemption process” from Federal Motor Vehicle Safety Standards. USDOT’s new Nontraditional and Emerging Transportation Technology Council furtheraims to cut any red tape that may impede AV deployments.

All this is part of an understandable step towards enabling technological advancement, but it is a huge mistake to omit a strengthening of core transportation planning principles. While USDOT briefly acknowledges this risk in its voluntary guidance for state and local governments to “consider the potential for increased congestion, and how it might be managed”, it is negligent in applying traditional measures for mitigating this risk. Federally-funded transit or vanpool projects, for example, are required to regularly submit NTD reports on performance, including vehicle occupancy.

Reinventing the Pyramid

And so we risk a transportation system that elevates AVs, including those with zero occupants, to the tippy-top of our planning hierarchy. AV manufacturers are accelerating towards this goal, as evident in our vehicle-centered connected infrastructure terminology such as “vehicle-to-everything (V2X)” and a vehicle “green wave”. To avoid a likely negative impact on traffic congestion, transportation planners should commit to a renewed emphasis on occupancy-based planning for all modes, new and old — a re-affirmation, rather than an-upending, of the existing rules of the game.

Extending the transportation hierarchy.

There is simply no excuse today for not enforcing occupancy-based performance measures for all transportation modes, including AV deployments. A vehicle with zero occupants should not be treated the same as a vehicle with one occupant, or two occupants, regardless of its level of automation. The existence of reliable consumer technology for verifying occupancy means that there is no longer any excuse for federally sponsored AV deployments to be exempt from reporting on occupancy. Such a course correction will result in new opportunities for local and state agencies to safeguard and improve the performance of their road networks, and for auto companies to fully participate in solving the problem of traffic congestion.

Autonomous Cars Will Get Road Rage Too

Picture the scene. It’s a typical morning commute, marked by traffic congestion and frustration. Except this time it’s not drivers complaining to each other about the traffic, it’s autonomous cars. That sounds ridiculous, but it is the current trajectory for the future of personal mobility. The allure of Elon Musk’s vision for autonomous vehicles is understandable, particularly with proclamations that “within two years you’ll be able to summon your car from across the country”, all of the obvious safety benefits to come, and all the cool tech that goes with that.


However, a missing dimension when we paint this vision is our daily pattern of car utilization. The efficiency of our transportation system doesn’t improve simply because cars will drive themselves more efficiently than we can. A predictable tidal wave of traffic congestion will still hit all of our cities every weekday morning and evening. In fact, it will get worse.

The pervasive wisdom is that autonomous cars will appear and disappear as we wish, wherever we go. Our primary transportation resource (256 million cars and counting) will essentially live in the cloud. Personalized mobility on demand. Most articles written on the subject portray a transportation system that is seamlessly efficient, the ultimate product of the sharing economy. And a recent article in the Wall Street Journal explains that “Twenty-five years from now, the only people still owning cars will be hobbyists, hot-rodders and flat-earth dissenters.” Common perception is, therefore, that this version of the future is inevitable. It is not.

Today, the average occupancy of a car is 1.1 people at commute time. That’s the reason you waste a week of your time every year simply sitting in traffic. Research has shown that even a small (2–5%) reduction in traffic volumes can lead to a giant reduction in total traffic delay (25–27%). And yet, instead of our cities competing on commute-time occupancy scores for federal transportation funding, we continue to invest $80 billion annually on new roads and $87 billion in wasted fuel and lost productivity. Not to mention the harmful impact on local air quality, with the American Lung Association reporting that almost half of us are breathing air polluted enough to make us sick.

At the same time as making these investments, we celebrate the arrival of the autonomous car as the solution, with President Obama recently announcing $4 billion over the next 10 years on autonomous vehicle technology. That $4 billion is largely earmarked for safety features, a very worthy cause — but auto and tech companies need to also start investing in better understanding car utilization patterns. Without learning today how society can better use cars, the autonomous car of the future is unlikely to have a greater occupancy at peak travel times, and it will be riderless a large portion of the time, as it moves to and from the “cloud”, thereby driving occupancy rates down further.

Today, the average American household has 2.28 cars, each one sitting idle 95% of the time. For most Americans, there is no alternative. Most people live and work in America’s suburbs and exurbs; they don’t take Uber to get to work, and they don’t live near a transit stop. They’ve been left behind by a downtown mindset in mobility innovation and neglected in our collective vision for Shared Autonomous Cars, if we even have one. A recent MIT studyshows that the entire population of Singapore could be served by approximately 1/3 of the total number of passenger vehicles currently in operation — but getting to that point requires a giant collaborative effort, fueled by Tesla-levels of iteration and inspiration, and ultimately leading to societal behavior change.

There is, therefore, an urgent and fundamental need to change our long-established car utilization patterns. The same (high-occupancy) car that takes you and your neighbors to work must be the same car that takes your colleagues to off-site appointments during the day, and your family to a restaurant or Ikea or the beach in the evening.

At Carma, we're proud to be working with Toyota to make this a reality. Our fleet of Toyota hybrid vehicles are available to Toyota team members for a shared commute to work, and for private errand trips during the day. The logistics are challenging and we learn new lessons every day, but we believe that car utilization patterns are fundamental to the efficiency of our wider transportation system. Not just for the next generation, but for today’s.

Image uploaded from iOS (6).jpg

Surely it’s time for Silicon Valley to apply the same urgency to ensuring cars are shared and filled throughout the day as it has to summoning a car from 3,000 miles away. The conversation around autonomous cars has to start encompassing the logistics of how a car gets utilized for 24 hours, not just for 1. That technological and behavioral challenge is one that faces us today, not just in 10 or 15 years time. If we don’t change course, we’re heading towards a future in which zero occupancy vehicles further clog up our already congested roads. And nothing is more likely to trigger robots taking over the planet than having robot cars stuck in traffic every day, right?

A version of this post first appeared in Fast Company (Co.Exist).

Emmett Murphy is Chief Product Officer at Carma