汽车 > 车联网博客



NHTSA: Driverless, Directionless in DC

by Roger Lanctot | 4月 10, 2016

The National Highway Traffic Safety Administration (NHTSA) held a public hearing in Washington on self-driving cars last week.  As reported by the Verge,  the voices of skeptics and critics overwhelmed supporters of self-driving car technology thanks in part to the absence of representatives from Google, Tesla, Apple, Lyft, Uber or Volvo.

http://tinyurl.com/qcxyqhb - Voices Clash at Fist Public Hearing on Self-Driving Cars - theverge.com

The pressure is building on the U.S. Department of Transportation to define policy on automated driving to provide a framework for domestic development as well as to align the U.S. with the rest of the world.  While it was a government agency, DARPA, that can largely take responsibility for the introduction of the concept, the future of automated driving is now in the hands of the private sector and regulators.

Fans of driverless cars tout the potential for reducing traffic congestion and related toxic emissions as well as the prospect of reducing or eliminating the economic cost of vehicle crashes.  Societal benefits are touted as well including making mobility available to the handicapped and improving social mobility and better access to employment opportunities.

Unfortunately, the opportunity posed by driverless cars appears to have caught the USDOT completely off guard and its flat-footed stance on the subject leaves it teetering dangerously toward a defensive position that might lead to a premature application of the brakes.  The closest thing the agency has to a position on the subject is a feeble pivot several years ago from a focus on surviving crashes to a focus on avoiding crashes.  NHTSA has also provided for rule exemptions for automated vehicle developers.

While avoiding crashes is a laudable objective, the goal was presented without a roadmap for its achievement.  The only research agenda currently on NHTSA's docket is the agency's pursuit of a mandate for vehicle-to-vehicle communications based on Wi-Fi technology - an initiative with little or no relevance to the self-driving car debate.

Multiple safety advocates, including the outspoken gadfly and former regulator Louis Lombardo, have demanded that President Obama and his Transportation Secretary Anthony Foxx commit the country to a Vision Zero agenda for drastically reducing and ultimately eliminating crashes and their related fatalities.  The U.S. is fourth in the world for the number of highway fatalities and ranked embarrassingly high on the rate of those fatal crashes vis-a-vis miles driven.

The latest statistics from 2015 show the amount of driving is up, highway fatalities are up and even the rate of fatalities is rising.  All trends are moving simultaneously in the wrong direction.

The agency has to ask itself:

  1. Are self-driving cars part of the problem or part of the solution?
  2. Will the pursuit of self-driving car technology take us closer to mitigating or eliminating crashes and crash-related fatalities and injuries?
  3. What are the agency's goals in facilitating the development and adoption of self-driving car technology?

Holding hearings is helpful, but it is less so without context.  I don't blame the self-driving car leaders for taking a pass on speaking in DC last week, but they abandoned the stage to those in opposition who raised the specter of killer cars and terrorism on the highway.

The greatest challenge in guiding the development and deployment of this technology will be, in the words of Lou Lombardo, "proving a negative."  How can the agency prove the efficacy of self-driving car technology and all of the precursor technologies such as lane-keeping, adaptive cruise control, blind-spot detection and automatic emergency braking?  What sort of research agenda will be sufficient to reassure the general public of the benefits of self-driving cars?

Agency direction is especially important here given the fact that researchers are routinely pointing out the inclination of drivers to turn off various forms of safety and emission reduction systems from lane keeping alerts to start-stop functions.  While consumers demand safety there is no groundswell of driver demand for self-driving vehicles.

The car companies themselves are a bit at a loss as to the positioning of the technology.  Most car makers appear to be in reaction mode in response to Google's brash self-driving experiments on public roads.  This situation puts car makers and NHTSA in the same defensive crouch when it comes to self-driving cars.

The car companies have hied to test facilities like Mcity in Michigan and others popping up around the country.  These artificial environments provide safe liability-free testing zones for the incumbent car makers, while upstarts Google and Tesla continue to operate on actual highways and local streets.

Car makers need some regulatory cover so that they may too get out on the open roads and contend directly with Google and Tesla, but how?  Can NHTSA help make this happen or will NHTSA seek to rein in the insurgents?

It may be time for NHTSA to consider shifting its focus from a nearly irrelevant emphasis on V2V technological development toward a closer scrutiny of sensor-based collision avoidance technologies now being applied to self-driving cars.  The recently concluded Nvidia GTC event in San Jose revealed the increasing role being played by both multifunction cameras on cars and Lidar technology in conjunction with deep learning systems gathering and interpreting images gathered from millions of miles of driving.

The ultimate challenge of proving a negative, proving that sensors and sensor fusion systems can learn and can be taught to avoid collisions lies in the increasing volume of data being collected by cars in real world and test scenarios throughout the world and in the U.S.  Simultaneously, the data is being developed to determine what kind of infrastructure elements need to be added to support self-driving systems.

The first step, though, is for NHTSA to determine if it wants self-driving car technology on the road and, if so, what it wants from this technology.  DARPA was looking to create driverless convoys in order to avoid casualties related to improvised roadside explosives.  Thankfully, we don't have that objective to worry us on our highways.

It's time for a U.S. Vision Zero objective to help clarify the self-driving car conversation currently characterized by fear, suspicion and doubt.  It would be unfortunate if the skeptics and critics were somehow successful in impeding development and adoption of technology that could start reducing the alarmingly high rate of crash-related injuries and fatalities in the U.S. and abroad - while also stimulating the economy and ultimately reducing the cost of vehicle ownership - or possibly eliminating vehicle ownership.

It may be that the truly scary vision for the industry and for regulators is not a car without a driver, but a car without an owner.  Inherent in such a prospect is the car company as a utility rather than primarily a manufacturer and that transformational prospect is an even greater challenge for the USDOT to confront.  But confront that prospect it must.

Previous Post: Nvidia GTC: Paradise by the Dashcam | Next Post: Huawei HAS 2016: Cloudification of the Car
Leave a comment