Re: Autonomous Vehicles NOT Ready for Prime-Time
Posted: Thu Mar 07, 2019 2:06 am
Your Kizashi Owners Club and Forum
https://www.kizashiclub.com/forum/
I am submitting that "real world scenarios" can only be created by actual public roads. I understand your concerns about this but it's rather self-evident to me and the engineers behind automated driving that they have to run these cars no other place but public roads for them to have any validity and merit in testing and data acquisition for development. While not near perfect by any means, they are not up against perfection either. In fact, as I have repeatedly stated in every engagement here, the human driver is far more flawed as a whole. You argued that automated systems are useless in inclement weather conditions where snow and ice can impair sensors, cameras, etc. I agree, however, what I also consider is that human drivers are also very bad at driving in snow and ice. In fact, simple rain causes accident rates to increase regardless of location. In essence, why have an expectation that automation must, "BE guaranteed to function correctly 100% of the time (short of failed componentry) UNDER NORMAL OPERATING CONDITIONS" when humans are nowhere near this standard either and that's just an accepted reality? In fact, humans cause so many accidents, insurance is a normalized industry and is required almost everywhere. Again, automation should have a high standard, but the expectation of perfection is odd when those demanding it seem to accept the pathetically low standards of the status quo.Ronzuki wrote: My position absolutely NEVER assumed any such thing. My position IS that exhaustive testing with real world scenarios must be performed, and passed repeatedly, prior to giving it the old college try on public roads.
Good luck to the prosecution trying to prove that the human in that automated Uber was liable instead of Uber (which quickly settled out of court). I think this would be an easy case for a good defense attorney, especially given the facts of the case: the vehicle was automated and was supposed to be driving itself and the victim was jaywalking at night. No doubt that human was lousy, but criminally responsible?
Then please enlighten me how autonomous vehicles are to be tested for valid real-world driving. Please offer a realistic alternative.Ronzuki wrote: No, I really don't think you truly grasp and understand my concerns, especially as an automation engineer for 30+ years.
I'm not saying autonomous vehicles are currently superior to the average sober and attentive human driver. What I'm saying is that their dangers are over-hyped on the media given how many millions of miles they have cumulatively ran and continue to cover. They are also improving by the day as they gain more data from real-world testing. Also, Tesla and Uber are just two of over several different companies operating automated driving just in the US.Ronzuki wrote: Your unsupported opinion that autonomous cars are safer than humans is complete conjecture. What REAL WORLD data could you possibly have to support your position when they vehicle can not sense an object in front of it in clear weather?
Oh, I get it. You seem to have trouble that I don't agree with your assessments. After all, you have yet, after numerous repeated inquires, made an opinion about a comparison of automated driving to the dangers of human driving. It's odd given you just complained about sharing the road with automated "experiments" yet neglect that hundreds of people die each day in just this country alone sharing the road with drunks and distracted drivers. This is what I mean by your steadfast acceptance of the status quo when they present a far greater risk today than the current state and especially, the future, of automation.Ronzuki wrote: Still not getting it are you? What makes you, or anyone else (elon the arrogant) think I want to share the road, day in and day out, with someone's unproven science project? What makes you or anyone else think I want to pay for all of this BS, be in the car I purchase or the ridiculous taxes I have to pay for public works that will be greatly affected by thoughtless and typical government oversight (waste)?
I already pay the highest gas taxes in the country and the roads are deplorable, the traffic is deplorable and there's no relief in sight. Again, government that can't maintain the road systems we have in a manner that would provide the basis for autonomous vehicles to even operate?
That's a hilarious response given you first start off telling me you're the expert, then tell me I'm not, and then avoid my question, and then state that I should answer my own question. Why did you avoid my question? You even tried to flip it on me despite you're the "expert" on automation here. Honestly, answer my question. How would you test for automated driving technology other than running them on public roads?Ronzuki wrote: I live, eat, and breath how uncooperative these modern whiz-bang 'systems' (very loose term) and technology behave in the industrial environments which they are supposedly designed to operate in. Environments that are far more controlled and predictable (in theory), as you've noted earlier, than public roadways. Again, what makes you, someone who has never engineered anything from a blank piece of paper and blank application development screen, think that anyone could possibly automate a car and have it function under all normal operating conditions, affordably? No, you first, please, enlighten me. For me to explain to you the processes' required, and the utilization of the experience of what works, what doesn't and what is simply not a good idea to pursue, just because you can, would take months of writing. I've worked with vision systems any many, many different sensing technologies to attempt to automate some not so very repetitive processes, and they are never as reliable as the customers mistakenly believed they should be. I've explained this all before elsewhere. Rarely do they remain online long enough to garnish a ROI and ultimately are abandoned. Hence, waste of time and money. Public roadways are an even worse 'not so very repetative process'. Understand it takes years before a 'new academic engineer' is knowledgeable enough for us to allow said engineer to electrically design and program something that is functional, safe, reliable, cost effective and can be commissioned to provide years of 24/7 service w/o causing downtime and loos of profits. Otherwise, we're out of business in the blink of an eye. We're not talking about an iphone or some kiddie toys here either.
Yes, but it's still a fact that lettuce has killed more people than automated cars last year. If your concern is public safety, then where's the rage against lettuce? Automated cars have cumulatively ran hundreds of millions of miles and their fatality rates for all cars combined internationally are in the single digits. While it's hard to compare them statistically to human drivers due to limited data for a fair comparison, I think it's reasonable to say we should be more fearful of other human drivers than automated vehicles on the road. Automated vehicles will get better and better as they collect more and more data. They will also have less deviance from the standard in operation among them. The drunk or distracted driver doesn't really improve given their negligence and, oh, there are millions of them at any given time. But hey, that's normal so whatever, right!?Ronzuki wrote: How many heads of Romaine lettuce are harvested and consumed? How many AVs are running around. Numbers and percentages my friend. Manipulate them properly if you want to have those conversations.
Which company? I recall reading about one that did VW conversions too but wanted to use their own motors and package them with a full restoration so the products were $85K. Ridiculous. I figure this guy has a better business model. I also wonder if his supply is a reflection of Tesla's strong sales recently in CA, decimating the sales of cars including the Prius and BMW 3 Series. Also, that's pure speculation that he gets wrecked Teslas as a result of failed automation, especially given the fact that not all Teslas have Autopilot. Your line of reasoning is akin to saying that there are more Toyota Camrys available for parts in junkyards so they must be less safe than Chevy Aveos when talking about a guy who restores Toyotas with junkyard parts.Ronzuki wrote: Saw a program on TV about a shop in CA retro fitting old vehicles w/ electric motors and batteries...VW vans seem to be a thing for him. Guess where he gets the motors and batteries? From wrecked Teslas of course. He's got a backlog of projects, and apparently an ample supply of parts, to keep him busy for at least the next 6 years he claims. At least they're good for something. So the Tesla experimentation on public roads is going well it would seem. Hope they're 'learning' a lot.
Absolutely.Ronzuki wrote:Now, does anyone notice any similarities between today's automated aircraft business and where automated vehicles are headed?