GAO faults DoT for not testing Autonomous Vehicles to ensure minimal capabilities
Please reference this article and the associated report
I want to focus on this part — “The Transportation Department, for its part, said it concurs that a comprehensive plan will eventually be needed. But in a prepared statement published alongside the GAO report, a department official said such a plan is “premature,” because of “the nature of these technologies and the stage of development of the regulatory structure.”
It is an absolute myth that creating test scenarios in any way negatively impacts AV technology advancements or actually has anything to do with that technology in the first place. There is almost no correlation between the tech involved and creating test scenarios to ensure that tech is doing what it should. As a matter of fact the minimal acceptable scenarios should have already been created. Especially geofenced subsets prior to fielding of these vehicles. That to include Waymo in Phoenix and GM and others very dangerous use of handover in their L2/L3 vehicles. DoT, state and local governments have been abdicating their responsibilities by not creating minimal test criteria. As a matter of fact if the AV makers had this test set they would know what the end state needs to be. At least at a minimum.
The second myth is that the majority of the test scenarios will come from public shadow driving. The majority of which is done for AI. It is impossible to drive the trillion miles and spend the over $300B to do so. (Not to mention the lives that will be lost needlessly when that AI migrates from benign scenarios to those that are complex and dangerous.) Not only can you never come close to driving the miles how would you KNOW that you experienced all the road patterns, degraded sign types, environmental conditions etc? And every perturbation of them without some form of expert driven checklist?
The third myth is that the commercial sector will find best practices without government involvement. Someone tell me when that has EVER worked. The FAA and NASA can clearly show that did not happen in air travel. Right now the AV industry is the Wild West. There is no coordination on any of this. To make matters worse almost every one of them is misrepresenting their capabilities so they can grab the cash or glory. This is producing a false sense of confidence in the public and clearly the government. And as I said most of the AV makers are using the wrong process to produce the technology in the first place.
The solution, as Waymo’s massive paradigm shift shows is to stop using handover and public shadow driving and move to simulation. The other part of the solution is to create a Scenario Matrix that assures all the minimal capabilities are there. Including the geofenced cut outs along the way.
There is an article on this where Senator Dianne Feinstein articulates the point I am trying to make very well — “I’m strongly opposed to it,” Feinstein told Bloomberg Government Dec. 14. “I do not want untested autonomous vehicles on the freeways which are complicated, move fast and are loaded with huge trucks.”
For more on this and links to references I cite please see my article
Autonomous Levels 4 and 5 will never be reached without Simulation vs Public Shadow Driving for AI