Although the title. Finally Doing It, sounds like a 1970s B-movie, the latest session of the Smart Driving Car Summit featured superstars of the autonomous vehicle world and blockbuster content on that topic. And hearkening back to the decade of stagflation, when another nascent technology was on the verge of hitting the mainstream, the fight to bring autonomous vehicles to market will most likely occur on a city-by-city basis, similar to how cable television rolled out on a market-by-market basis.
The focus of the session was on the rollout of driverless, shared vehicles (for this article, simply driverless). Featuring a mix of different size manufacturers/integrators (Ford, Local Motors) and service providers (May Mobility, Waymo). With Princeton’s Dr. Alain Kornhauser driving the panel as moderator, it was a fast-paced overview of the challenges and opportunities of bringing autonomous mobility as a service to market.
As Kornhauser often points out, the goal should go beyond merely providing another way to get from A to B, but it should be to improve the quality of life while providing safe and affordable mobility.
Parallels with the Cable Franchise of Yesteryear #
Like the cable television and more broadly the telecom business, Ford’s John Rich suggests that long-term, the driverless is simply too capital-intensive with too much intellectual property for there to be more than a few winners. In the short and near-term, however, there appears to be room for many players as they perfect their technological and business approaches on a market-by-market basis.
In many ways, driverless rollouts are analogous to the franchises issued by local authorities in the 1970s. In the case of driverless, the franchise is the Operational Design Domain; or the area served and under what conditions the service operates.¹ This process requires a close working relationship between the driverless operator and municipal or institutional entity.
This has led to 275k revenue-generating rides over their four existing deployments and they are set for three more launches in 2021. Olson believes there is a path for smaller companies to use a disruptive approach to carve a space in the market. He indicates that as a smaller company they have only spent 2% of Waymo’s capital.
Waymo’s Dan Smith describes two different business models that they have, that are centered around their Waymo Driver. One business model, Waymo One, is a commercially available driverless service currently serving the Phoenix-area. Waymo Via is the Waymo Driver applied to the trucking world, which enhances and doesn’t disrupt the truck driver. Collectively, the Waymo Driver has driven over 20+ million real miles, 20+billion simulated miles, and 100k miles without a safety driver.
The Suppliers – A Holistic Approach #
The suppliers to driverless operators are working closely with operators and municipalities to understand what is necessary to create robust, reliable, and safe driving platforms. Echoing others on the panel, Ford’s John Rich points out that each city has its own unique rules, layout, and behavior of the local citizenry. Maintaining a relationship with and sustaining trust with the cities and their citizens is key to advancing the technology.
Jay Rogers’ experience as a U.S. Marine Infantry Company Commander taught him that having the right kind of vehicle in the right situation at the right time can save lives. Rogers, the CEO/Co-Founder of Local Motors translated this need into a company centered on empathy, open-source, co-creation, local production, short hardware cycles, and sustainability. Rogers says that this approach allows Local Motors to develop much faster (12 months, instead of 5 to 7 years for a typical vehicle) and at a much lower cost ($5M versus typical $800M vehicle development costs).
The result has been various versions of the mostly 3D-printed, ever-cute Olli. More than a vehicle, the production is unique in that it takes place in a microfactory, which is a 10,000 square foot facility located in the markets where the vehicles are sold. It takes about 8 hours to 3D print a vehicle. Of course, this is just one part of a production process that integrates third-party components into the final assembly. Local Motors indicates that its distributed production technique is much more efficient requiring only 5,800 kWh to produce a vehicle versus 9,500 kWh compared to centralized factories.
Although Rogers didn’t explicitly say where Olli fits into this range, he estimates that today it costs $200k to $500k per vehicle to replace and remove the driver; implying that shared rides are necessary for the first practical commercial applications for driverless. From an operational point of view, the vehicles will have to be electric to minimize total lifecycle costs.
No one left behind #
Unlike cable television’s expansion in the 1970s, which focused on markets that would be willing and able to pay for a new product, the first markets for driverless could be those that cannot afford today’s mobility options. Serving the mobility underserved is certainly the emphasis of Dr. Kornhauser and he made it clear to the panelists that they would find a welcoming market in Trenton, NJ.
Rich says that every city that Ford works with is conscious of both transportation and food deserts. Their experience in Detroit and Miami has informed them of some of the huge problems driverless can solve. Delivering packages and food in neighborhoods without low-cost grocery stores or retail outlets is another potential benefit driverless brings.
Waymo is looking at hub-to-hub, highway driving as one of the early use-cases. This offers the potential to bring safety advancements to this segment. Automation that assists the driver potentially makes her job less stressful.
Building on his point that sharing is a requirement from an economic standpoint, Rogers predicts that shared, electric, autonomous travel could drop to one-fifth of today’s cost. Ultimately, this means costs of less than 20 cents per mile. Because of these low costs and the new opportunities that will result, Rogers believes there will be unimaginable upward economic mobility.
Self-Driving Claims Muddy the Public Perception of Driverless #
A safety-first approach is a common denominator for the companies represented by the panelists. Reiterating his comments from the February 4th Smart Driving Car Summit session, Waymo’s Smith says, “It starts with safety, as outlined in their methodology papers.” He explains that the decision to take the driver out is an enormously difficult, but incremental and studied decision. It’s not a leap, it’s a progression.
Rogers stresses that saving lives is the goal. That goal should create a sense of urgency to create an autonomous driving system that is safer than humans and reduces the mostly human-caused mayhem.
Olson says the priorities of May Mobility are
- Creating a good passenger experience and more broadly be “a good citizen on the roadway” as it interacts with pedestrians, bikes, and other vehicles.
- Maximizing autonomy
Rich states that, “There are no shortcuts, no cheating, no using someone else’s homework.” You have to do the work, using data. The claims by Telsa of fully self-driving cars have been confusing to the public and investors and are hurting the driverless industry.
Expanding on Rich’s comments, Waymo’s Smith explains that you don’t want people to conflate the problems associated with automation that assists the human driver with what is required for driverless. People need to understand their role and what they need to do when they are behind the wheel, regardless of the technology (anti-lock brakes, lane departure, automatic cruise control, etc.)
Kornhauser succinctly ties the discussion together stating, “Until Elon accepts responsibility for driverless, it isn’t.” This statement reinforces Kornhauser’s view that one sign of readiness is when the CEO makes the decision to launch a commercial driverless service.
Seeing Around the Corner – How Much Fixed Infrastructure Is Necessary? #
The big advantage of driverless compared to the early days of cable television is that very little fixed infrastructure is necessary. This not only allows the ODD to be redefined as needed to meet market conditions, it means that it doesn’t take years and years of planning and construction to start service.
One possible infrastructure, DSRC, is a purpose-built radio network for communicating to vehicles, pedestrians, bikes, and, potentially, anything that moves. This technology-approach, given approval by the FCC 20+ years ago, is quickly being usurped by C-V2X, which is based on 4G and evolving 5G technologies (see DSRC vs C-V2X).
Olson, Rich, and Smith are strongly in the camp that one cannot rely on an external communications network, such as DSRC or C-V2X, to drive. As Olson says, vehicles need to be accountable for themselves. Connectivity is helpful in understanding location, routing, and orchestration for improving efficiency, but that is different from using it for sensing or commands. The functionally also doesn’t necessarily mean it comes through a purpose-built infrastructure in the 5.9 GHz band.
Rogers sees a strong need for V2X. He says that they are deploying DSRC radios in Olli and that they encourage partners to install the associated infrastructure. Diana Furtchgott-Roth, former Deputy Assistant Secretary for Research and Technology for the U.S. Department of Transportation, points out that the ability to see around corners is one potential V2X use-case.
Echoing what was reported in the 2/4 Smart Driving Car session A Culture of Safety Is AV Table Stakes, V2X seems like a nice-to-have and not a must-have. That’s not to say that those saying they don’t need it today will never use it as an option. Smith indicates that if V2X can enhance their autonomous driver and do so in a timely and reliable way, they might add it to their sensor suite.
To the point about reliability, opening a new sensor input provides another potential attack vector from a cybersecurity standpoint. Regarding cybersecurity, Marjory Blumenthal Senior Policy Researcher, RAND Corporation, asks what is being done to vet open-source software?
Rogers’s response is that repository approaches like Github have proven to be resilient and cited examples like the Linux Kernel and its robustness. Further, Rogers suggests that the DoD has shown an open-source approach is superior for creating mission-critical code. The topic of cybersecurity is worthy of its own session or more.
Stay tuned, as next week the conversation will continue as the Dispatcher’s Michael Sena will lead a panel that addresses Who Will Build, Sell and Maintain Driverless Cars?
¹ Added 2/14/21 – John Rich spoke of the concept of Geo-Net to describe the roads within the ODD that their driverless vehicle will navigate. Simply, they may choose not to drive some roads because of the road conditions (e.g. blind intersections, poor markings, potholes, etc.), construction, or other factors. Alain Kornhauser describes the concept at 26:47 in the video, below.