Forums Forums National News Driverless cars

I can't think of any reason why your phone would lose GPS in Wales or Scotland over England though it might lose GPRS etc. The cars don't run solely on GPS in fact some of them don't use GPS at all for the autonomous stuff. The GPS system does go down sometimes and has planned outages

Even hardware goes wrong sometimes.

https://www.theregister.co.uk/2018/03/21/nvidia_titan_v_reproducibility/

Years ago I got (at work) a very expensive Hewlett Packard scientific calculator. It worked just fine and made my life a lot easier. One day, I did a simple multiplication of two three or four digit numbers and got an answer that was wildly wrong. I phoned them and they admitted that it was a hardware fault. I was the third person in the UK to have stumbled on it. They knew the reason and it would malfunction ONLY on those numbers. In due course we received a letter warning us of the malfunction. It had taken some years for this fault to be found. You can imagine the result had it been a computer controlling the steering of a car.

Add to this the FACT that with millions of lines of code, there WILL be software faults that cannot in practice be either found or rectified, and they can also cause crazy behaviour.

As Mr Wolmar observes in his book, there are huge problems with supervised automated systems - in this context, cars that may require driver supervision under certain circumstances.The supervisors tend to go to sleep! You can't really blame them. Sitting there with nothing to do for ages, it happens. When the car runs into problems it can happen with very little warning and the supervisors do not react fast enough or properly. The problem has emerged in the pilots of aircraft, railway engine drivers, and automated collision avoidance in ships. It will emerge in cars too.

I cannot for the life of me find anything to support the claim that these cars will be safer than humans, and my experience of real-time software over a number of projects suggests the opposite. I think the enthusiasts for these things are naive.

It depends on how the regulations change over time and whether the car will have a self-drive mode.

At present you must be sober and qualified to drive - even if you're letting the car drive, just in case you have to take over. Pretty much like with an instructor with a learner driver.

Whether in future cars with self-drive modes have a competency test before it lets the human drive is a possibility.

Computers are surely only as good as their hardware and programmers . I hope to God no-one who has ever had anything to do with setting up the Governments Inland Revenue system is ever let near computers for autonomous vehicles. Massive KNOWN glitches that accountants constantly point out to them , yet they just allow staff to plod on using the wrong figures that are being churned out, and STILL argue the toss when you quote their own rules to them which show without doubt youre right!

All that kind of stuff doesnt inspire me with confidence that I could trust a car operated by a computer - Id be sat beside it on the edge of my seat , and a nervous wreck by the time I got to my destination.

https://www.theguardian.com/technol...death-woman-failure-fatal-crash-arizona. ( Vid of the collision)

You know when you look at a laptop screen, it depends exactly on the angle its opened at, as to how light or dark the 'picture' is ---
Looking at the video , theyre saying the woman just stepped out with her bike, and with or without technology assisted driving, (the tech they agree was faulty), nothing could have stopped a collision.
However, does that mean then that just because this video shows the scenes as being too dark for the human driver to have seen the woman in time to react , that that was actually the case? The report says the technology failed in a big way, but how can they be sure that the actual drivers view at the time , wasnt lighter and that had they been concentrating and not bored, they might just have had a couple of seconds longer to swerve or brake? Probably was too close anyway, but just wondering in principle if camera view playback is defo going to be identical to a drivers original view ? Also, it would depend on the quality of the dashcam in the first place -- apparently according to AutoExpress, some have had problems with adjusting to light and dark - although youd hope that absolutely top notch ones were used for autonomous cars

From what I can see She stepped out from the crown of the road after I presume crossing half the road. The auto side should pick up in any kind of light as in parking sensors, no picture or light required infra red measurement I think.The human eye should see her way off in the head lights. Human error and not just the dozy driver.

The driver wasn't looking at the road, it looks like she was playing with her phone.

I've got to admit I don't like semi-autonomous vehicles, drivers will be blasé and treat them as autonomous vehicles when they aren't. The car hasn't currently got an "autonomous mode" that the media keep on stating. Also the driver is the "driver" not the "safety driver" like they are stating.

In this instance the pedestrian would probably fall foul of the jaywalking laws.

Multiple failures by the driver and the pedestrian, the semi-autonomous car was in testing and while it may not have performed as expected, that is the reason it is in testing and has a driver.