It's been in the news that some Uber driverless cars ran red lights. One of Google's cars had a prang recently. Some estimates are that Lyft (a ride sharing firm) will transport most of its customers in driverless cars by 2020. Lyft is part owned by GM. Trucks are being tested in USA and Germany amongst other places. It's coming, and soon. Tesla experienced the tragic death of a customer driving whilst reportedly watching a DVD. The car's cameras were unable to cope with bright low sun if I recall - as any photographer knows, light sensors in cameras have limited capability to 'see' (way less range than the human eye) and this gets worse at night. Recently Digital SLRs like the Sony A7S have come on the scene with incredible light sensitivity that can make night scenes look like day ones - but these are very expensive cameras). With all these limitations, accidents will happen. A few weeks back it was reported that Mercedes self driving systems would be programmed to save the driver and passengers when there was a choice betweeen them and pedestrians. Where does that leave bikers? If anyone has insider knowledge they can share, I'd be genuinely be keen to be educated... How does this make you feel about riding on the road? ehat restrictions or habits will you adopt when riding to stay safe? I'm thinking: - night riding is out - avoid motorways in heavy rain - avoid riding against a low sun - filtering will be risky: no cameras dedicated to looking for filtering bikes... though this may not be any different from at present as humans don't see us either!
There are of course positives: - less road rage? - more predictable traffic flow and patterns... less erratic driving... - computers don't get bored, make mistakes (except through faulty programming or logic) etc - so will be safer within their boundaries of capability And I guess we'll need to re-learn how to read traffic flow. Probably future posts in forums will talk about different manufacturers' autopilot algorithms and behaviour... more to gas about!
When it's eventually released into the Wild it will be so much safer than random moron drivers who are drunk, on the phone, and jacking off while driving. It's harder to get that las 0.01% +log safety, and every crash will be publicised for sure. Will still be safer than humans though.
Will the camera be able to see that just behind you spot pulling out of an awkward junction and judge the speed of an oncoming bike with its headlight on more accurately? Until it can, its not to be trusted
There will no doubt be a steep learning curve during which mistakes will be made. The system will either then improve to a point where it is acceptable, i.e. better than it is now, or it will be sent back for further development. What would be totally unacceptable is a rush of accidents caused by driverless cars. I personally am not losing any sleep over this, I can't see this becoming an issue north of Kendal any time soon.
Interesting one that. Don't know. Lots of the testing is in USA with no push bikes on the road and not much filtering going on. But they have bikes on the road, and so the software must be able to do this. In fact, I'd imagine it does it better than humans. Whilst humans seem to find judging speed of approaching smaller objects like bikes, for a computer they are no harder - if they can judge a car, they can judge a bike just as accurately. What I wonder about is how computers can replace judgment though, as you point out. We can anticipate what drivers will do. We look at mirrors, look for car wheels turning, look for eye contact - these are signs of intention, and I doubt computers can anticipate any of this. They will rely on reaction times and prior programming to respond to events. Algorithms and AI will I imagine have some predictive ability, but based on prior situations encountered and based only on measurements of the environment, not intentions.
I read that currently the car manufacturers currently accept responsibility when their autopilot systems are driving. So car insurance will get more complex, and will depend on who's in control. Autopilots will be monitored remotely, or have black boxes onboard, probably both.
Broadly agree: like planes and trains, the systems will be more reliable than the humans. The issue I'm highlighting is that we need to know their limitations, like camera sensors and situations they can't cope with. That caused the Tesla accident, from what I've read. As I mentioned, we will also learn in time how to read the movements and reactions of auto piloted vehicles. And a huge plus will be that they will be consistent, unlike drivers. I agree the danger lies in the first 5-10 years of mass adoption, when the sheer numbers will inevitably reveal the extent of problems, and public opinion and legal cases will establish boundaries and test cases, and the R&D process will continue.
Its all round bad news IMO. Picture the scenario where you get knocked off your motorcycle....must be your fault as computers dont make mistakes? Judging by the front camera on my Avensis, its a disaster waiting to happen. False alarms telling me to brake, a warning telling me to clean the windscreen because the 'camera cannot see'!!! , all of this when visibility is perfect, I can see there are no hazards. It all needs a serious rethink, as does our acceptance of machines over humans. Machines do malfunction, they cant get out of the car to clean the lights, they dont have years of experience to 'read' complex situations. Imagine also, coming over the brow of a hill to find stationary traffic, the driverless car stops, well short. Unfortunately the car behind (with a human behind the wheel..) does not. Now then, an experienced driver would maybe edge forward, to allow the nugget behind the extra braking distance required before they rear end them. Been in that scenario many a time and prevented a rear end collision. My Toyota also has the ability to read road signs (speed), so I would expect driverless to have same tech.? Great, till the idiots my way decide to turn the speed signs around on the posts! Too many variables that a human would deal with better. Maybe its just me thinking also that we need to be careful of companies like GM and Google trying to do away with human interaction and jobs!??
Lots in there: and don't get me started on job destruction at the hands of tech companies... oh, too late! All good points. But one can only imagine this why the testing is going on, to solve these problems. Remember, computers will not necessarily solve these problems in the same way that humans do. They will use sensors, measurement, reaction time and so on to deal with situations. If you look at aircraft technology, it's pretty reliable. When it does go wrong it goes spectacularly wrong, of course, because computers can't deal with the unexpected (yet... AI may in future, but that's a big IF...). Nobody is talking about doing away with pilots any time soon!! The point about cleaning the kit is a good one. I'd imagine in future there will be disclaimers we have to sign that mean we need to keep the kit clean. Which is fair enough IMO: because remember, at least for now as private drivers, it will be our choice when to rely on the autopilot. If you switch it on, it's your responsibility that you have cleaned the camera lens/window/etc... if it's dirty or you suspect it is, switch it to manual... Glass tech is moving on too: so in future I imagine self-cleaning technology will exist and be affordable... but this trickle down of tech takes decades... FAR MORE IMPORTANT IMO is the issue of job losses... that's a whole other topic in itself. Drivers, roboticised factories (have you seen the Tesla factories?), policing (cameras etc), medicine (AI coming...), war (drones...) and many more scenarios are already happening. I'm really worried about this. The tech firms assume they have the right to kill jobs en masse, all in the pursuit of profit. But such disruptive technology is going to have a massive impact if unchecked. They are already making people "self employed" (not just Uber, but also online markets like Fiver and other 'pay per hour' services) without worrying about the consequences for who pays for retraining, education (freelancers take on the cost of their development rather than employers). The same companies operate tax free through shell companies and offshore legal structures, so that they don't pay the taxes that could otherwise help fund all the consequences of job losses. This has to change, but since the problem is global the question is how? So many questions... and meantime, the tech firms are just motoring ahead... and Trump is unlikely to stop them!
You've no doubt seen this from BMW: their vision of the future... doesn't look too bad! The BMW Motorrad VISION NEXT 100.
At least a driverless Uber cab couldn't turn up outside a hospital and leave a man in a wheel chair...in the rain..... "Sorry mate that won't fit in here" and before I could explain it folds down like a kids stroller he's off faster than the babysitter’s boyfriend. I got charged £5 for the pleasure.....I hate Uber cabs...they make Black Cabs and Addison Lee look respectable.
Making Addison Lee look respectable is saying something right there! Yes, and Ubers on autopilot don't require awkward polite conversation either. For some reason I've got images in my head of old "Airplane" movies with inflatable pilots... better leave that one there. Me neither in U.K., but did use them in Chicago last year - easiest way to hail a cab when there are no taxi ranks... I read this week that some states in USA are planning to subsidise Uner fares rather than invest in infrastructure and public transport systems. Uber are still making massive losses apparently as they are subsidising fares, so they will likely want to hasten driverless vehicles as fast as they can.
What can no one see how we are marching towards some kind of Walle world, combined to AI and that other movie with Will Smith
iRobot? Great movie. The future will either be fabulous or absolutely unbearable I reckon. Fabulous if robots take the hard work out of work, we all work 3 days a week for the same money, and inequality is banished to the annals of history. But if not, the alternative is not looking great.
Uber's business model appears to be based on attracting customers by charging less than taxis and attracting drivers by paying more than taxis. How is this even possible? Because Uber uses the money put in by investors to subsidise transactions heavily. This is why Uber has never come close to turning a profit - it makes massive losses every year. So why are investors willing to invest? Because they are attracted by the promise of profits to come, once competitors have been killed off; at that stage prices to customers go up, pay to drivers goes down, and Uber laps up the cream. Essentially Uber's commitments to investors, to drivers, and to customers are wholly incompatible with one another, and they obviously are not upfront about this. The whole thing is not quite a Ponzi scheme, but pretty close. In the meantime it provides a cheap and handy service so enjoy it while you can.
