Blog: Why Autonowashing Makes Me MADD

Article By : Colin Barnden

Have you heard the term “Autonowashing"? Conceived by researcher Liza Dixon and explained in her paper “Autonowashing: The Greenwashing of Vehicle Automation,” this is a must-know issue for anyone working on ADAS or AV technology...

If you haven’t previously heard the term “Autonowashing,” you have now. Conceived by researcher Liza Dixon* and explained in her paper “Autonowashing: The Greenwashing of Vehicle Automation” this is a must-know issue for anyone working on, or writing about, assisted, automated or autonomous vehicle technology.

Liza Dixon (Photo: PChua, Yeong Jiun)

Dixon defines autonowashing as:

The practice of making unverified or misleading claims which misrepresent the appropriate level of human supervision required by a partially or semi-autonomous product, service, or technology. Autonowashing may also be extended to fully autonomous systems, in cases where system capabilities are exaggerated beyond what can be performed reliably, under all conditions. Autonowashing makes something appear to be more autonomous than it really is.

The objective of autonowashing is to differentiate and/or offer a competitive advantage to an entity, through the use of superficial verbiage meant to convey a level of system reliability that is misaligned with the technical capabilities of the system. Autonowashing may also occur inadvertently, when one unknowingly repeats erroneous information about the capabilities of an automated system to another. Autonowashing is a form of disinformation, and it is, in a sense, viral.

Autonowashing is the logical outcome of the current mania of an AV tech industry — encompassing many high-profile and massively-funded suppliers such as Aurora, Cruise, Nvidia, Tesla, Uber and Waymo — engaged in a race to full autonomy.

I encourage any AV tech company which has exaggerated the capabilities of its systems and the suitability of probabilistic AI and machine learning for use on public roads and highways — that’s roughly all of them with the possible exception of Argo AI — to sincerely consider the concept of autonowashing and to completely rethink their PR and marketing messages. AV technology is one part of what will be a decades-long journey to improving safety on our roads and highways; it is not an instant panacea to end all human suffering.

I know I’m forever quoting Missy Cummings in my writing, but I listen carefully to everything she says and this quote from episode #18 of the Automotive News Shift podcast is particularly damning:

The entire driverless car community took what was basic research and made an assumption that it was ready to deploy. They were just wrong and they didn’t do their homework. They made the wrong decision that the technology was much closer to deployment than it actually is.

How did we end up here? How did something as important as road safety get turned into a race? Simple: It’s because it started as one.

The DARPA Grand Challenge
DARPA is the Defense Advanced Research Projects Agency and the present-day race to commercialize “self-driving” or “driverless” technology has its roots in a military challenge. The Wikipedia entry of the 2005 Grand Challenge and the video below give us some major clues how we ended up with autonowashing.

The 132-mile course was successfully navigated by five vehicles, four of which managed it in about seven hours, equating to an average speed of just under 20mph. Great. I estimate most 8-year old kids could have driven the desert terrain, which was a closed course.

With no finishers to the course in the 2004 Grand Challenge, one might erroneously interpret that as infinite progress in just one year and conclude that self-driving cars would be “solvable” in, say, 10 or 15 years. Google hired key members of the 2005 winning team to form its self-driving car project (now Waymo), many billions of dollars of investment poured in, the race to commercialization of the technology began and we can just hit fast-forward to where we are today.

Yet it seems no one stopped to think that completing a complicated challenge might not tell us anything about the applicability of the technology in a complex system. We discussed this issue in EE Times a couple of weeks ago.

Until any AV tech company can adequately explain how their perception and prediction AI will overcome emergent properties inherent in complex systems, I’m not expecting much in the way of real progress.

In my view, the greatest life-saving technology of the next 20 years won’t be anything “self-driving,” but vision-based driver monitoring systems (DMS), coupled with the application of human factors science to make human drivers into safer drivers.

It isn’t participants of the DARPA Grand Challenge that D.C.-based lawmakers should be courting, but human factors scientists. Recently, experts like Missy Cummings and industry bodies such as Partners for Automated Vehicle Education (PAVE) have been focusing a lot more on DMS.

All this autonowashing is making me MADD
DMS has a critical role to play in tackling not just human distraction and fatigue, but also intoxication. I had the privilege this week of talking to Mothers Against Drunk Driving  (MADD) about the suitability of DMS to detect for intoxication by alcohol.

Fatigue and intoxication can both be categorized under impairment and I believe DMS has a substantial role to play in the years ahead mitigating for the effects of drunk driving. The goal of human factors in driver monitoring is to deliver reliable, real-time understanding of the drivers’ cognitive state as it applies to accident risk.

I’m not advocating DMS-based ignition interlocks or any form of “nanny-tech” whatsoever — and a car cannot under any circumstances rat you out to the cops. However, DMS technology clearly has a role to play in detecting driver impairment and adapting the responsiveness of the safety systems accordingly. This could include:

  • Disabling the “highway-assist” capability of systems like GM Super Cruise or Ford Active Drive Assist if the driver is deemed to be impaired.
  • Increasing the sensitivity of the autonomous emergency braking and lane-keeping systems based on measurements of the driver’s impairment level.
  • Activating a non-overridable speed limiter using data from the traffic sign recognition camera if the driver is deemed to be impaired.
  • Increasing visual and audible alerts and using seat vibration to encourage the driver to stop driving and take a rest.

Mandatory installation of DMS is covered in the Moving Forward and SAFE Acts and efforts to mitigate for impaired driving are also covered in the RIDE (Reduce Impaired Driving for Everyone) Act. However, U.S. road safety policy is now several years behind initiatives from European bodies such as Euro NCAP and the updated European General Safety Regulations.

Autonowashing isn’t the way forward and Liza Dixon has made a significant contribution to the safety debate with the publication of her paper. If you already want to hear more about the subject, she featured on episode #198 of the excellent Autonocast podcast.

Hopefully U.S lawmakers can set aside party politics, agree upon suitably effective measures to make human drivers into safer drivers and vote that into effect very soon. They would be MADD not to.

*Liza Dixon, M.Sc. is a recent graduate of the Hochschule Rhein-Waal University of Applied Sciences Usability Engineering program, where she specialized in human-machine interaction of assistive and autonomous vehicle systems.

Leave a comment