Safe Speed Forums

I had this idea on another thread discussing limitation in the current test, and an idea popped into my head that is so obvious and practical that I cannot believe that it hasn't already been implemented somewhere.

It requires the creation of a car simulator that should be as realistic as possible:

Driver set-up is in a real car interior placed in a moving platform set-up as widely used for aircraft simulation, roller-coaster simulation etc. This is perfectly practical, but not particularly cheap.

Car behaviour simulation is already available - the latest Grand Theft Auto and similar games provide very realistic simulation of car behaviour including on and over the limit characteristics. They would need a little extension to include the feed for the moving platform to simulate G forces on the driver, but this is probably built into the platform systems already and if not could be borrowed from existing aircraft and roller-coaster simuations.

A set of simulations of driving in real world conditions including day/night, sun/rain/snow/fog, empty/congested/normal traffic loads, single track country up to 4 lane motorway. Basic AI simulation of other driver behaviours including slow vehicles, tailgaters, considerate and inconsoderate behaviour, and non-drivers such as pedestrians, cyclists etc.

Such a system would have multiple uses:

As part of the driving test to check competancy before moving to on-the-road part of test. In fact all aspects of the on the road test could be included including a very realistic emergency stop that would require proper anticipation by the driver rather than the "when I tap the dash". The test can therefore include motorway and other aspects that are currently practical due to location issues.

As part of driving instruction particularly the early parts, but also can be used for advanced driving techniques and high speed training.

Entertainment - in the evenings why not let the system earn money from the public with entertainment uses such as simulated races around famous tracks - they do with computer games already this is just a much better I/O environment.

Safety analysis - yep rather than guessing about the best way of improving safety on roads the various options could be tested in advance with real drivers, in a wide variety of conditions and proper measurements about the impacts of things such as cameras or chicanes that are impossible to see on dark rainy nights can be made. It might even provide evidence that Speed Kills is a load of clap-trap.

The system would be produced in the tens of thousands and would have worldwide sales possibilities, so unit costs would be less than 10K, and at that cost could even be provided in schools and similar locations.

Anybody else think that this idea is a good one? I know IT and know that the IT aspects are perfectly practical. Is anybody else doing such a thing already - if so why isn't it widely available? Anybody want to go into partnership to build and sell such systems?

Ok, I have had a little look about on the net, and there are a few similar things around such as Green Dino which is in dutch, so I cant understand the detail, but got an award http://www.virtuality.nl/visnews/news_archive.html There are a few university eqivalents such as http://www.reflex.lth.se/carsimulator/traffic3/ and including a moving platform http://education.umn.edu/kls/research/hfrl/facilities/was.html.

A company is trying it http://www.vstep.nl/p_s_dsim.htm, but they haven't got a proper moving platform (vibration only) and haven't got the basics of force feedback onto even the steering wheel yet, and without that it isn't a decent simulation of anything.

I expect that the biggest problem is that most of the area is covered by other peoples patents, and that may prohibit the development of an effective system at an effective cost. The way forward might be to get the various specialist groups working together - such as MaxFlight (who make the Roller Coaster Simulators) http://www.maxflight.com/products/vr2002/default.asp.

yes and no.



many vehicle manufacturers, universities and the like already have such systems (ncluding moving platforms & simulated steering feel). i'm sure any of them would be willing to make them widely available if you're willing to pay enough.



no
to make it part of the test would require huge investment.
a closed road test circuit at each test centre would probably be cheaper.

I think this is a brilliant idea.

Ideally the driving would be on real roads, with real traffic behaviour prgrammed in. I suspect parts of this can be gathered from existing CCTV footage of motorways and city centres. A limitation will be the extent of the road network you have. Perhaps you just mark roads you can't drive down with no entry signs.

If possible, crash, skid etc feelings should be programmed in so if someone did loose it and hit a wall the simulator would give the apropriate (to a safe degree) tactile feedback. Also useful would be what the car handles like when you have a puncture.

Ideally I think the units should have the capacity for two simultaneous drivers, thus allowing head to head racing with your friends in an evening. More practically it could be used on a test, for example an instructor could be in the other car tailgating the student, to give a human interaction on the students reaction.

Another useful possibility would be the driving while lost - e.g. get the driver to navigate roudn an unfamiliar city to a certain car park, etc. Also it would allow interactions with different traffic not found in the locality - e.g. trams, the nearest ones to where I am are in Nottingham and Croydon - no way could a driving lesson get that far.

Also useful would be to simulate driving long journeys, e.g. get them to feel what their driving is like after a two hour drive. Or even driving while drunk (dunno how this would play, as it would require getting drunk in advance - could you tell someone they must get drunk?).

Lots of brain dumped ideas here, but hopefully you can make sense of them.

Chris

Estimate the visual field - I'm thinking that about 30 LCD projectors should do the job...

Assume a 7 foot screen at 12 feet from the driver in a circle, that's 75ft (circumference)*7ft of screen.

_________________
Paul Smith
Our scrap speed cameras petition got over 28,000 sigs
The Safe Speed campaign demands a return to intelligent road safety

30 is far too many - we only used 3 to give 120 degree forward vision for the Harrier systems simulator at BAe, but that was 20 years ago at just over VGA resolution, so assuming that we keep to a round 360 and then we only need nine. IIRC the Tour-de-France 360 film that I saw at Futuroscope (near Poitier France) used 12 projectors for a circular hall that was 20m plus in diameter (but the 360 12 camera setup that was used on a motorbike in the middle of the actual tour racing was a bit cumbersome). That was 10 years ago and much bigger than is needed here.

You only realy need high resolution in the forward 180 as you do not normally focus hard in the mirror or over your shoulder when driving. I would think that 3 HTDV quality projectors would do the front visuals fine, and it might be possible to do the whole thing with just 6, because realistically a driver has just 4 views - windscreen, right and left side and rear. The A pillars that everybody complains about because you cannot see the bikes are covenient join points. So 6 projectors: 3 for front to give high resolution front image as is required for hazard perception and one each for sides and rear.

The graphic set-up doesn't need to be perfect, just good enough, and if you look at the quality of the latest games, they are certainly good enough for this. If you doubt the ability of consumer level graphics cards, then you need to look at some of the screen shots or clips from modern games consoles such as PS3 or XBox. The level of realism being displayed is plenty good enough for a driving simulation, and they can now output to HDTV.

So what do we need - high powered processor for the main simulation (1) with subsiduary processors for graphics (2) and motion, audio, force feedback and input collection (3). 6 graphics cards (7 if you include the control console).

The movement platform only need a 6 cylinder platform as I wasn't planning on doing realistic rolls when you crash - you can simulate a realisitic up to .8G force just by leaning with limited lateral movement during the transistion stages. This is just 6 pneumatic cylinders and a control system not very expensive components - it is the control of the cylinders that is complex, so it is probably worth buying in the experience to get a mapping from the G force aspect of the car simulation to the cylinder positions.

It will take time to develop and perfect the synchronisation, but all of the necessary elements are already available and on the market - the biggest investent is of course where the money is, which is why computer games is pushing the graphics capabilities, and that technology should be used.

Only a few people have considered road safety and driving as a potential market. As I see it, the biggest problem is likely to be simulator sickness caused by a mismatch between visual simulation and physical sensation, and with a full motion base and good synchronisation

I think you're way way under in your estimates of visual detail required. Flight simulators are one thing - we're trying to see things like runways. But driving requires us to look as far ahead as possible - imagine the graphics resolution required to scan a mile ahead in a motorway situation - but that's just normal driving when the landscape permits.

Basically you need to match the resolution of the eye in all the areas that a driver might wish to look. If I remember rightly the centre portion of the eye (fulvia? something like that) has 6 million light sensitive cells mapped onto an area the size of a ten pence piece at arms length. OK - maybe we don't need to go that far, but I reckon it's at least ten times worse than you're suggesting. I'm starting to think my 30 projector estimate is well under.

But even with your estimates it's a massive project, with a multi-million pound development budget and a unit cost well into 6 figures (say £300,000) + development amortization.

_________________
Paul Smith
Our scrap speed cameras petition got over 28,000 sigs
The Safe Speed campaign demands a return to intelligent road safety

What about VR headsets? After all you can only look in one direction at a time. These are available as consumer grade products and some come with head tracking as well. (http://www.i-glassesstore.com/) An uprated version could be designed for the purpose. Some form of polar coordinates projection system could be used. Instead of having pixels arranged in columns and rows they're in concentric rings (Same number of pixels in each ring) This gives very high resolution in the very centre with the res. dropping off as you move away from the centre - much like the human eye.

The idea sounds good. Various hazardous situations could be simulated and driver reactions investigated.

I think the hardware will be the biggest problem.

Just read this thread after posting something similar under Brake and young driver thread. Simulators are used for planes so could be used for cars. They would be expensive but then so are lives and insurance for young people. Could be entirely self funding if set up in a way that the driver successfully completing say 100 hours on a simulator gained a lower insurance premium.

in the same way my advanced driving doesn't ?