Jan 21
At the urging of CMU, Pennsylvania wants to remove “safety drivers” from driverless vehicle development and testing
Reference article — Pa. transportation secretary, senator unveil bill to allow self-driving cars to be tested without someone behind the wheel — https://www.post-gazette.com/news/transportation/2022/01/05/pennsylvania-self-driving-cars-autonomous-vehicles-bill-wayne-langerholc-yassmin-gramian-uber-carnegie-mellon-aurora/stories/202201050130
A Pittsburgh Post-Gazette editorial voicing their concerns — https://www.post-gazette.com/opinion/editorials/2022/01/10/Safety-first-Self-driving-industry-needs-to-earn-trust/stories/202201080012
For those who have read my articles before, you know I believe this is an incredibly bad idea. Now though, it is more personal. We are talking about significantly increasing the risk to me, my family and the community I live in. I live in Pittsburgh, PA. Second only to California in autonomous vehicle development.
What I say now is not alarmist, hype, or exaggeration. (Unlike the modus operandi of the industry.) This decision will injure, and fatally injure, citizens of this state needlessly if it is allowed to go forward. The reason for this is the driverless vehicle industry has made several very flawed assumptions regarding the effectiveness of their development and testing approaches and the technology being used to accomplish them. Those being general and deep learning are mature enough to be effective, the real-world and human test subjects must be relied on. And there is no simulation technology, including gaming technology, that can replace most of that. Of course, none of that is correct.
Regarding machine learning processes. The general and deep learning technology being used must experience driving scenarios to “learn” them. That involves experiencing driving scenarios, including many crash scenarios, over and over again. Possibly thousands of times each until the highly inefficient machine learning process succeeds. To make matters worse, each scenario must be very specific because there is little inferences available in current general learning processes. That includes very specific locations, object types and colors, time of day and weather. Even the existence of shadows. This means this process takes hundreds of billions of miles to execute in the US. That in turn requires hundreds of billions of dollars to fund. This all means the process is untenable from a time, cost, and safety point of view.
The solution is to use aerospace/DoD/FAA simulation technology to remedy the real-time and modeling fidelity issues associated with gaming simulation technology to alleviate most of the overall safety, time and cost issues. (In addition to improving general and deep learning approaches.). Where real-world testing will need to be done, particularly to inform and validate the simulation when test tracks cannot be used, there should be proof it is required. And those events should be run like a movie set. Not just haphazard involvement of the public. Said differently, we should utilize progressive engineering due diligence to ensure we use human test subjects, in and around the vehicles, as little as possible. Not as the go to approach. What CMU should be doing here, as the leading autonomous learning institute in the world, one that wields significant industry influence, is more engineering due diligence and homework on the right approach and simulation technology. And then advising the state, and everyone else, accordingly.
Below are a couple articles that explain my POV in more detail.
The Autonomous Vehicle Industry can be Saved by doing the Opposite of what is being done now to create this technology
How the failed Iranian hostage rescue in 1980 can save the Autonomous Vehicle industry
SAE Autonomous Vehicle Engineering magazine editor calling me “prescient” regarding my position on Tesla and the overall driverless vehicle industry’s untenable development and testing approach — (Page 2)
· https://assets.techbriefs.com/EML/2021/digital_editions/ave/AVE-202109.pdf
SAE Autonomous Vehicle Engineering Magazine — Simulation’s Next Generation
· https://www.sae.org/news/2020/08/new-gen-av-simulation
My name is Michael DeKort — I am a former system engineer, engineering, and program manager for Lockheed Martin. I worked in aircraft simulation, the software engineering manager for all of NORAD, a software project manager on an Aegis Weapon System baseline, and on C4ISR for DoD/DHS
Industry Participation — Air and Ground
- Founder SAE On-Road Autonomous Driving Simulation Task Force
- Member SAE ORAD Verification and Validation Task Force
- Member UNECE WP.29 SG2 Virtual Testing
- Stakeholder USDOT VOICES (Virtual Open Innovation Collaborative Environment for Safety)
- Member SAE G-34 / EUROCAE WG-114 Artificial Intelligence in Aviation
- Member Teleoperation Consortium
- Member CIVATAglobal — Civic Air Transport Association
- Stakeholder for UL4600 — Creating AV Safety Guidelines
- Member of the IEEE Artificial Intelligence & Autonomous Systems Policy Committee
Presented the IEEE Barus Ethics Award for Post 9/11 DoD/DHS Whistleblowing Efforts
