Recently Waymo began notifying the public about major shifts it has been making regarding how it develops and deploys autonomous vehicles. Prior to this paradigm shift Waymo followed the path the vast majority of the industry is unfortunately still following. That being the use of vehicle handover (L2+/L3) and public shadow driving for AI, engineering and test. As well as hyping their technologies capabilities and when that technology will be complete. Instead of following this path Waymo is loudly and forcefully making monumental changes because they want to save lives with their technology, not needlessly put people at risk, actually bring the technology to fruition and frankly, to save the industry. Waymo and its leadership chose to take a direction that far too few do. They held their egos and pride in check, defied peer pressure and allowed objective data to drive them through a sea change. These acts demonstrate many examples others should follow. Examples of courage, ethical fortitude, honesty, humility, good business acumen and engineering excellence.
The first thing Waymo realized is that the path they and most AV makers are on was putting the public in jeopardy needlessly. That path involves the vehicle steering on its own short of full autonomous driving (L4). Where the vehicle can “handover” to the driver. Often with them unaware this is necessary. Especially in critical situations. When this process is used during development to train the vehicle there is even more of a danger because the system doesn’t know what most of the proper things to do are yet. Waymo knows that when these companies move that public shadow driving for AI from benign to more dangerous scenarios lives will be lost needlessly. They know that the L2+/L3 vehicles some are using for ride services or selling to the public cannot be made safe by any monitoring and notification system. They know that these practices will result in the first loss of a child or family in a vehicle using these very flawed practices. (I am aware that Joshua Brown lost his life due to this and that as an adult he was someone’s child). Waymo also knows that you can never create an autonomous vehicle using public shadow driving because you cannot drive the one trillion miles required to stumble and restumble on the massive amount of scenarios you have to get through to train the system thoroughly. Or spend over $300B to do so. Waymo is also aware that at some point the world will figure all of this out. Probably when tragedy strikes. When that happens the industry will be under severe scrutiny. That scrutiny will lead to the public, press, governments, insurers and others become aware of these issues. They will realize the industry plans to use them as needless Guinea pigs to drive thousands of dangerous scenarios thousands of times over. That they are doing so in the use of a largely unnecessary process that will never result in an autonomous vehicle. And that they are not using the alternative which is aerospace level simulation. All of this will lead to mistrust and far more regulation and delays than if the industry self-policed and got on the right track.
Instead of continuing down this very counter-productive, unethical, hypocritical and untenable path Waymo made assessed the situation properly and made significant course corrections. They stated they would skip L2+/L3 and replace most public shadow driving with aerospace level simulation. Recently the CEO John Krafcik called Tesla out on this by name. Declaring Tesla and others are using the wrong path. Instead of letting his competition go down the wrong road, Waymo chose to make them aware of the issues. To nudge his competition in the right direction. He did this not just because he realizes that the pending tragedies the others will cause will negatively impact the whole industry but because Waymo doesn’t want to see anyone hurt by the wrong approach. If you are a true believer that this technology is meant to save lives there would be no other position to take.
Waymo is also taking the high road by refusing to participate in the perpetual one-upmanship loop the majority of the industry is engaging in by exaggerating their capabilities and time to market timelines. A practice that is meant to give their audience a false sense of confidence, boost egos and secure funding ahead of the competition. (Some have taken this to extremes. Tesla by saying they would have full autopilot this year. Intel admitting it made a fake ad using LeBron James. And comma.ai who sells a kit, with software the user can modify, without most of the critical sensor systems. In addition to making the shadow driving issues far worse this approach will make hacking and weaponization easier). Waymo also admitted that after 8 years they are just getting to engineering and testing L4 vehicles in significantly geofenced environment. That being the use of a modern, well-lit, well-marked, low complexity gridded street environment like Phoenix. (And that is only up to a light rain. This is due to an industry wide sensor capability gap not Waymo’s AV technology).
(I am not naive. I am fully aware that there are other benefits of making these public announcements. Benefits that will probably help separate Waymo from the pack. To avoid the untenable path of shadow driving one trillion miles. And to avoid being sucked into the massive backlash that will come when avoidable tragedy strikes. While some may argue those are Waymo’s true intentions, I do not believe that is the case. I believe Waymo’s intentions are the right ones. I believe the strong conviction they convey is genuine).
Hopefully, in the end, Waymo’s very public paradigm shift and vocal declarations of what the wrong approach is, and what it will cause if not changed, will rise above the din of the massive the echo chamber before that first child or family comes into harm’s way. And other AV makers spend too much time and money on the wrong approach. This should drastically hasten the timeline to true autonomous technology so that the world can benefit from it.
Please find an article I wrote explaining all of the issues in the AV industry as I see them. As well as ways forward. I have updated it to include the most recent articles on Waymo’s position.
Letter to Congress — Handling of minimum standards for Autonomous industry
My name is Michael DeKort. I am a former systems engineer, engineering and program manager for Lockheed Martin. I worked in aircraft simulation, the Aegis Weapon System, NORAD and on C4ISR for DHS. I also worked in Commercial IT. Including cybersecurity. I also received the IEEE Barus Ethics Award for whistleblowing regarding the DHS/USCG Deepwater program post 9/11. •http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=4468728
I am leading an effort to pull together a group of autonomy and mobility community professionals who wish to ensure the approach to gaining this technology can actual be achieved and is done in a responsible and safe manner.
Professionals for Responsible Mobility
Please let me know if you would like to join our group