There are some critical data points that are missing here. Ones that would help me understand if your stance on and use of public shadow driving is proper. I am hopeful you will provide this information. As it would either differentiate you from almost everyone in the industry who is doing this wrong and will harm people for no reason or it would actually make you one of the worse offenders as it relates to misinformation, misleading people and providing false confidence.
The key issue being what you don’t say about the dangers of public shadow driving, even with your beneficial hands on approach. And the untenability and dangers of the practice overall. Without this your defense of public shadow driving, admitting it is dangerous, giving people the belief that somehow it is unavoidable and that your method makes it safe could be extremely misleading. The key here is exactly how much public shadow driving is needed, why and how it is done. Not just involving your driver but the public around you. If you maximize proper simulation and test tracks. If you use public shadow driving less that 1% of the time and when it is done it is done so properly then all is well.
I would like to focus on your statement here — “Right now, human safety drivers are a crucial component in the safe development of autonomous driving systems.” Exactly what does this mean? What % of the time is it needed versus proper simulation or test tracks? What exactly do you not believe you can do in proper simulation or on a test track and why? Please provide specific scenarios. What simulation products do you use? Have you read about my concerns with simulation products in this industry and addressed them?
While I certainly agree your hands-on approach to shadow driving is a better approach do you agree there are scenarios where that nor any monitoring or alarm system can provide the time to regain proper situational awareness and do the right thing? When you use shadow driving in the public domain where do you go and what precautions are taken to protect the public? Does your shadow driving involve the public?
Do you agree that it is a myth that public shadow driving is the best or only solution to create a fully autonomous vehicle? That it is impossible to get anywhere close to a driverless state using this method? That no one can drive and redrive or stumble or restumble on enough scenarios nor afford over $300B to do so? And that this method will kill people? Not only through running shadow mode in thousands of accident scenarios thousands of times but those that involve handover?
If you have the right answers to these questions and concerns, I suggest you edit this post as well as your certification paper to include them. I will then in turn delete this post and champion your approach.
For more information please refer to my articles here
Uber isn’t the only Autonomous Vehicle maker who should fear killing a child
The Crash of the Autonomous Vehicle Industry
Impediments to Creating an Autonomous Vehicle
The Dangers of Inferior #Simulation for #Autonomous Vehicles