By David Browne

In 1991 Congress passed the Intermodal Surface Transportation Efficiency Act (ISTEA) authorizing $650 million to develop “technology for driverless cars to operate on an automated highway.” DARPA’s “Grand Challenges” beginning in 2004 and “Urban Challenge” of 2007 played significant roles in further stimulating the development of self-driving vehicles. Since then, driven by the multiplying pace of progress of technological advances in lasers and sensors, vehicle to vehicle communication, vehicle electrification, and ride sharing, the initial stages of the transformation of the transportation (people and goods) and automotive industries have already begun. Over 15 companies are currently testing autonomous vehicles in California. Google’s test vehicles have logged over 2 million miles, and a controversial test market in Pittsburgh with AV rides being offered by Uber (with a backup driver) began in August 2016.

Autonomous Vehicles (AVs) may be characterized by their classification relating to the level of their automation, i.e., the assumption of the current human roles by automated technologies. These automated capabilities have been described by several manufacturers going on the record with projected milestone dates for availability of such capabilities (see stages, below). Market offerings of such technologies must first pass review processes including testing, guidelines, regulation, state and likely national legislation. Thirteen states to date have passed their first legislation related to automated driving, with 17 others “under consideration” as of April 2017.

SAE has defined six levels of automation (0-5) as follows (simplified):

Level Name Automated System Role Human Role
0 No Automation None All driving functions
1 Driver Assistance Driver-assist features including adaptive cruise control, preemptive braking, lane centering, parking assist, driver observation & alerts Responsible for all core driving functions
2 Partial Automation Partial driving automation, e.g., steering, acceleration, deceleration Responsible for monitoring roadway environment, ready to assume control w/ or w/o system warning
3 Conditional Automation Most driving functions and roadway monitoring are automated. System-designated request for human intervention State of ongoing readiness to assume control and intervene in response to system request
4 High Automation All driving and monitoring functions are automated. Operation limited to selective environments, e.g., defined shuttle routes Human control unnecessary. Steering, pedals, and gear shifting generally unavailable
5 Full Automation All driving functions and environments without a human driver Human input to navigation but without any vehicle control. Possible opt-in / out by human operator


Stage 1, Driver Assistance: Now widely available, with such features becoming standard in more mid and low-end vehicles, while initially offered only in premium models and brands.

Stage 2, Partial Automation: First introduced by Tesla on 10/9/ 2014 via its “Autopilot” system on Model S and then X, with regular “over the air” software updates. According to a recent report by NHTSA, crash rates for Tesla vehicles have dropped 40% since Autopilot was first installed (DOT-NHTSA Investigation: PE 16-007 closed 1/19/2017, Jeff Quandt reviewer).

Stage 3, Conditional Automation: Company spokespeople have indicated Tesla’s “Level 3 or 4 capability will be available in 2018.”

Stage 4, High Automation: Google / Waymo has indicated it will have a Level 4 capability in 2020; Volvo and Ford have been quoted as reaching this milestone in 2021.

Stage 5, Full Automation: Elon Musk was quoted on 5/2/17 as saying this fully automated capability will be “demonstrated by the end of 2017 and enabled in 2019” (Electrek).

At each stage of this evolution, the role of OEMs, Fleet Operators, Integrators, and Service Providers within the industry ecosystem will continue to evolve, as will the continuum of ownership/sharing, sales and service channels, and the value contributed by each stage.


The policy and legal regulations will continue to evolve – with many concerned that a patchwork of local rules could slow testing and commercialization of certain classes of vehicles. In response, NHTSA released a Federal Automated Vehicles Policy Statement (FAVP) in September 2016. While neither comprehensive nor binding, it sought to build a foundation for the collaborative development of future policy. Four sections of this policy statement addressed guidance relating to vehicle performance, federal and state roles (discouraging states from regulating design and performance of AVs), a reiteration of its regulatory tools (including exemptions), and new tools and authorities.

Safety, Liability, and Insurance

Risk is clearly beginning to shift from “drivers” to those who provide vehicles, technologies, and software. The legal framework must dramatically evolve before responsibility and liability will become clear, especially in automation levels 2 and 3.

On May 25th a six car fleet of Tesla vehicles ran in a pilot program by Tesloop, a ride-sharing service transporting customers between cities in Southern California, was offered a new auto insurance policy by Farmers Insurance at a 25% risk premium discount vs. the previous policy (a “test partnership”).

A recent report by KPMG predicts the personal auto insurance sector could shrink by 60% within 25 years due to autonomous technology (10/15 White Paper / insurance).

Approximately 90% of collisions today are believed to be caused by human error. A recent study by McKinsey and Co. suggests that by mid-century, the penetration of AVs could cause vehicle crashes in the United States to fall from second to ninth place in terms of their lethality ranking among accident types and save over $190 billion / year in total system costs caused by auto accidents. While automation will never be flawless, technologies including AI and machine learning grow more promising each day. While the incidence of collisions and the average speed of those collisions are expected to fall dramatically, consumer expectations are also expected to dramatically change, with a corollary lowering tolerance for even lower levels of (AIS) injury severity deemed acceptable in the AV environment of the future.  Continuous innovation in safety technologies, including both active and passive safety systems, will remain critical in both near and longer terms, to both reassure the still highly skeptical marketplace about the potential of safe AV ridership in the short-term and meet the heightened expectations around improved safety outcomes in the future.

Current Market Attitudes and Readiness

Finally, based on a range of survey data collected in 2016 and aggregated by the Governors Highway Safety Association, a majority of drivers currently say AVs:

  • Make them worried (57%)
  • Will not reduce crashes and fatalities (46%)
  • Should allow a driver to take control (80% / 96% – two studies).

Re: AVs, a majority of drivers also say they:

  • Would not use AVs if they were available today (75%)
  • Would not be likely to drive in an AV if it were available in 10 years (46%).

*This is the introduction to an ongoing series.