Talking to your car

High-end cars of various makes are gaining ears and a voice, allowing the driver to give instructions to the radio, the satnav, an in-built mobile phone, or the ventilation system. All of which is great for keeping attention on the road, where it belongs, and away from fiddling with buttons while ploughing through the central reservation.

There is a tension between making these speech interaction systems simple, and making them functional. After all, nobody wants a conversation with their dashboard that goes like this:

You: “Put a call in to the office.”
Car: “I don’t understand the instruction.”
You: “Call the office.”
Car: “I don’t understand the instruction.”
You: “Call Bob Smith.”
Car: “There are two entries matching Bob Smith in your address book: Bob Smith Office; and Bob Smith Mobile.”
You: “Call Bob Smith Office.”
Car: “You are driving too fast to make a call.”
You: “Aaaaargh!!”

Various suppliers - VoiceBox Technologies for example - are working with the major auto-makers to bring smart interactive systems to market that will actually respond in a meaningful way, rather than making you want to kick the centre console.

It’s not as simple a problem as detecting a PC users’ voice. Firstly, the user’s tone of voice is likely to change with speed, due to increasing road-noise levels, which will create a two-fold problem of extracting the information from the background and then extracting the meaning. Also, drivers will often have the radio playing just when they want to give verbal instructions, which will also pollute the audio environment.

One of the commonest tricks employed to improve speech recognition in difficult circumstances is to adapt to the user’s voice over time. Again, in a car, this might create problems. Cars are often used by more than one driver within a family or within a firm. And cars are typically sold on every three years or so.

All of which means that an adaptive voice recognition system will either need the ability to recognise multiple speakers, or will need a big fat reset button.

Forget drive-by-wire - how about drive-by-thought-alone?

Researchers in Japan have succeeded in controlling a robotic hand by scanning a person’s brain while that person thinks about opening and closing their own hand. This is clearly a great course of research with the potential to transform the lives of amputees. However, it’s likely that those funding the research have much broader applications in mind, for when this blue-sky stuff comes down to earth. After all, one of the organisations behind the experiments is car-maker Honda.

It’s significant that Yukiyasu Kamitani, one of the boffins behind the project, believes that the robot hand could be made to respond faster than a flesh-and-blood limb. "The next step for me is to decode faster, even before the person moves their hand, by reading the brain activity related to intention," he told New Scientist.

As noted in a previous post, there is a roughly one-second lag between a person’s intention and action when steering around a bend, for example. Reaction delays in emergencies – when slamming on the brakes – may be slightly shorter but not by much, particularly given that your brain has to tell your leg to shift itself over to the middle pedal before braking can even start (unless you’re in a Mercedes S-Class, of course, in which case you can leave it up to the automatic braking system, if you're feeling really brave).

What better way to cut reaction times, without wresting control from the driver, than by letting the car’s emergency systems take their cue from the driver’s brain?

Never mind intercepting instructions for the leg to get braking, brain-scans could register a much earlier impulse – the sudden surge of panic a driver feels in the instant that trouble becomes evident. This must register in the brain like a firework going off.

Sadly, brain-machine interfaces are a few years away yet. Honda’s experiments relied on functional magnetic resonance imagining (fMRI) – the kind of huge toroidal scanners used in hospitals. Not only do these scanners weigh upwards of 10 tonnes, they also create magnetic fields of such intensity that they are capable of turning heavy metal objects like fire extinguishers into lethal airborne projectiles.

So the current system would not only be limited to installation in heavy goods vehicles, it might also hasten road accidents – by physically grabbing passing bits of metalwork (ie cars) and mashing them into you. So, some slight refinement will be required before we'll all be driving by the power of thought alone.

Electric dreams

A year ago, the G-Wiz electric Noddy car sold by GoinGreen was a slightly less common sight on British roads than a pink Cadillac. Today, in Central London at least, the little plastic runts seem to be everywhere, silently creeping up on pedestrians with the stealth of ugly cubic cats. The reason is not hard to fathom: London appears to have more parking wardens than it does parking meters, and G-Wiz parks for free. El Presidente Ken Livingstone’s punitive Congestion Charge is a pricey £8 for the privilege of crawling over potholes, but the G-Wiz is exempt. And while the G-Wiz is unquestionably a bit of a slug, the average speed in the capital is still not much above the trotting pace of a three-legged camel. In short, London is about the only place in the world where running an electric runabout actually makes some kind of sense. Squint hard, and it might even make more sense than the other things you might spend eight grand on like, say, a Tube-pass and a second-hand Ford Ka.

The G-Wiz will no longer have this oddball corner of the market to itself, however. A rival has appeared in the shape of the Maranello4. Unlike the G-Wiz it has so-called “suicide” doors, with a hinge at the rear rather than at the front of the aperture. The only other mainstream vehicle on British roads to share this quirk is the £250,000 Rolls-Royce Phantom (RR calls them “coach doors”, not liking the lethal connotations of the normal name. And the Mazda RX-8's rear doors don't really count as suicidal, because you can't open them independently of the front doors).

While the Maranello4 is not as pricey as a fat Roller, it isn't cheap either: it's a full £1,696 more expensive than the G-Wiz at £9,995 on the road. And it’s slower (claiming 45kmh or 28mph top speed versus the G-Wiz’s claimed 40mph). When you’re driving a small, vulnerable vehicle, keeping up with traffic so as not to be overtaken by irate nutters really matters.

The Maranello4’s makers claim that their car has good looks on its side. It’s true that the G-Wiz has the kind of looks that only a blind mother pig could love, and that the Maranello4 looks slightly less like a wheelchair wearing a body-kit, but obviously that’s all just a matter of opinion. And it’s doubtful that either electric vehicle is likely to be bought by anyone who is particularly worried about what other people think.

Auto IT’s prediction is that the G-Wiz will continue to become a more common sight in the city. The Maranello4 needs to be cheaper and quicker to make any kind of impact.

The future of cars: the big-business view

Big-name, big-bucks business consultancy Deloitte has put some of its overpaid MBAs to work on wondering what the future might be like in 2010. But they haven’t exactly strained their precognitive frontal lobes – 2010 is only three and a half years away after all.

“A typical day in 2010 is unlikely to feel much different to today,” the firm's Eye to the Future report says, unsurprisingly. “We will probably not be teleporting breakfast or using quantum computers, nor will we be watching holographic TV or travelling to work in flying cars.”

But it does get a bit more informative as it goes on. The state-of-the-art car in 2010 “is likely to assist the driver by undertaking such tasks as: giving directions, controlling headlamps and windshield wipers, muting audio system when a phone call is received and guiding parking. However, it may also be undertaking more critical activities such as: regulating distance from the car in front, steering the car when it drifts out of lane and moderating speed when approaching an accident black spot.”

All of the above will be familiar stuff to regular Auto IT readers (both of you).

Deloitte does have something to say, though: “[All] this may not necessarily mean that road travel will be safer. Safety innovations can give rise to a false sense of security, and may lead drivers to take greater risks. Furthermore, the growing number of warnings and alerts issued by these systems may well create additional driver distractions.”

The false sense of security argument seems valid – drivers tend to underestimate available grip for braking on a dry road but overestimate grip in the wet and severely overestimate it when snow or ice is around. There is anecdotal evidence to suggest that these dangerous overestimates are made worse when drivers assume that ABS systems will allow them to stop more quickly in difficult conditions. ABS can help but it can't, of course, magic-up grip from a sheet of ice.

In a similar vein, Deloitte predicts that the average driver of 2010 will be a grunting caveman when presented with a paper map, due to the growing reliance on in-car navigation systems.

Auto IT would add that congestion may have worsened by 2010, as satnav systems will disrupt the traffic management measures so carefully put in place by town planners. For example, drivers following road-signs are invariably shunted onto long-way-round bypasses and ring-roads, away from town, village and city centres, whereas TomTom-equipped drivers tend to ignore road-signs and take the direct route even if it is likely to clutter up the high street.

Back to the report: company car drivers can look forward to choosing a vehicle based on its suitability as a mobile office (rather than its ability to be steered with the knees, presumably). “Desired features may include technology that can: read out incoming emails to the driver; allow the driver to dictate responses; permit the driver to set up meetings, update ‘to-do’ lists and write short memos.”

But the authors caution that again, there is a marked downside: “Already studies have shown that the use of hands-free telephones is no safer than using a regular mobile while driving, and concern is likely to grow about the safety of drivers that try to multitask.”

There’s a lot of this kind of material in the report. It provides a great synopsis, sadly little detail, and some helpful pointers to further info at the back. Download a PDF copy of the full thing from the Deloitte site.

Keyless or clueless, part 2

A belated follow-up to the recent post about keyless ignition systems comes courtesy of automotive consultancy SBD. It’s good to see that this team of boffins, which has studied electronic security systems in depth, endorses the Auto IT view.

David Bell, managing director of SBD, said: “The level of encryption in a transponder is so high, that even a thief equipped with the appropriate software would need much longer than [a few minutes to crack BMW's keyless entry code]. The computing power required to break down the protection is way beyond the capacity of a laptop, unless the thieves have previously managed to source confidential security data about the car.”

As Auto IT noted, there is a risk built into any security technology that rests on keeping certain information secret. The privileged few in the know will invariably include at least one corrupt or corruptible individual. Criminal gangs have, for example, been known to infiltrate banks via the front door – getting jobs as cleaners, temporary staff or even front-office staff, often using stolen identities. It’s not exactly ludicrous to suppose that a criminal element might infiltrate a car dealership chain or car factory. Indeed, BMW satnav systems have been stolen from the assembly line and sold online in the past.

Anyway, SBC chief David Bell goes on: “We have undertaken extensive research in this field for our recent report on Key Component Management, which illustrates both the strengths and weaknesses inherent in the systems deployed by manufacturers for storing the sensitive data for the programming of keys. Of greater concern than computer code cracking is the security provided by motor manufacturers and dealers to prevent bogus or criminal access to replacement keys and immobilisers. Obtaining a legitimate vehicle key through relatively simple fraud poses a much greater threat than the concept of thieves applying complex computer technology to do the job.”

So there you have it. It’s not the dodgy computerised keyless entry system you have to worry about – it’s the dodgy moustachioed guy in the dealership that should make you nervous.

Given that you have to hand over your key to get your car serviced, Auto IT wonders if it’s only a matter of time before “skimming” comes to the automotive world. For those not in the know, skimming is the illegal practice of making a duplicate copy of a magnetic-stripe bank card while the owner isn’t looking – typically while a card is being used to pay for a restaurant bill.

Presumably, unscrupulous dealership workers would be well placed to make copies of keyless entry cards. Or, if local copies are not possible due to the technical design, they might make bogus applications for replacement keys on behalf of their legitimate customers. Given that dealers tend to know owners’ address details and thus where cars are kept, those cars could then go missing in the dead of night, weeks or months later...

Bugs on the windscreen

Here’s an interesting snippet from Top Gear magazine’s review of the new Volvo S80 luxo-barge:

“There’s [an] optional driver-prompt: little cameras scan the mirror blind spot and flash an amber light when someone’s abreast of you. Trouble is, at certain speeds in the rain, they gather drops of water which fool the camera and trigger the light. Doubtless a little lip below the lens would cure it. So why not?”

Well, the why not is pretty simple: Volvo engineers probably haven’t thought of it. Even more probably, they hadn’t even registered the problem until they read about it in Top Gear.

The fact is, modern cars are increasingly encrusted with new high-tech systems, and it’s doubtful that the development cycle of the car, long as it is, provides enough time to detect these kinds of wrinkles. When you have to test not only that the car does what it’s supposed to do, but also that it doesn’t do what it’s not supposed to do, testing can take a lifetime.

Cars contain more software, obviously, but they are also becoming more like software: buggy, in need of patches and upgrades to fix the niggling problems caused by unforeseen combinations of events, tiny errors in design and other faults.

Given that more and more cars have touch screens and, in the future, will be internet connected, it won’t be long before drivers will have the option to send bug reports direct to the makers. It’ll be just like using a beta of Internet Explorer 7 – only at high speed, down the motorway. Now won’t that be great!

What do you think you're looking at?

In a previous post, Auto IT alluded to the fact that drivers rarely have a good idea of where they are looking when driving. This may sound daft - it seems self evident that most people are very well aware of where their own eyes are pointing. Alas, this piece of common sense is actually nonsense. Studies that measure eye movements using non-intrusive computerised headgear, such as the system developed by Professor Mike Land at Sussex University, can demonstrate that where we think we look is rarely the same as where we actually look. Just as the eye sees mostly in monochrome but the brain fills in colour with an unconscious combination of memory and guesswork, so the brain fills in our impression of where we look with made up stuff.

Looking back through the literature turns up some interesting gems. In 2001, the prof strapped his gear to the helmet of Tomas Scheckter, then a Formula 3 racing driver, and measured what his eyes got up to during six wet laps of Mallory Park.

On ordinary roads, Land’s research shows that ordinary drivers tend to have consistent patterns of eye motion:

In general, drivers appear to look at the section of road about one second ahead of their current position. This remains fairly constant as speed of driving changes. This suggests a one-second buffer between the visual input to the system [ie the brain] and the motor action - turning the steering wheel appropriately. We also find that when approaching a bend in the road drivers look at or near the tangent point (or reversal point) of the bend – [the moving point on the kerb] where the inside edge of the bend [appears to] change direction. The consistency of this result between subjects suggests that looking at the tangent point when approaching a bend is important for successfully navigating the bend.

Racing drivers (or at least Scheckter) don’t tend to do this. They track the tangent point only on long sweeping corners; on other corners they look beyond the tangent point to assess whether they are on the correct racing line. Also, because braking accurately before a corner is much more important for a racing driver than when trundling home with the shopping, racers tend to flick their eyes from the apex to the outside kerbs as they approach a hairpin to accurately gauge approach speed.

They also tend to turn their head with a great deal of accuracy, to minimise the eye movements needed to track the visual cues that help them stay on the racing line.

Having once served as a test subject for one of Professor Land’s former colleagues, Auto IT can confirm that having your eyeball tracked is a disconcerting experience, particularly when the research assistant is an attractive member of the opposite sex. As mentioned earlier, a lot of eye movements are unconscious and, though we may deny it, eyes have a habit of wandering over curves other than those on the road ahead.

Any colour you like as long as it's quick

An interesting example for students of spin. This story today notes that Volkswagen is “reducing the number of vehicle engine, trim and specification combinations and hence simplifying choice for customers”.

Simplifying choice? Up to a point... It is probably worth casting your mind back a few weeks to the revelation that it takes 54 man-hours to assemble a VW Golf – roughly twice the industry norm for a car in its class. Fewer options will mean less complexity on the production line, which VW evidently hopes will be one way to speed things up.

But this back-to-basics approach comes at a time when customers are becoming quite used to having a wide range of options. As BMW has proven with its ultra-successful Mini, customers like choice – even if it means the option to display their lack of good taste. And buyers are prepared to pay quite a bit over the odds to have a vehicle that they feel expresses their sense of individuality.

Fiat 500 meets the Homer-Simpson-mobile

In a similar vein to the previous post, only more so, Fiat is asking the public for input into its upcoming Cinquecento city car. As Reuters notes, it’s a bit of a PR stunt, as the car is much too close to its 2007 launch for such feedback to affect anything more substantial than the interior fixtures and fittings, paintwork options, and the pieces of plastic screwed to the metalwork.
There’s a silly interactive preview tool for rank amateurs and kids to play around with, but the real beef is at the DesignBoom web site, where a competition invites more thoughtful submissions from budding designers. The competition closes in September, and there are €5,000 prizes up for grabs (plus €3,000 fees for designs that make it to production).
It'll all end in tears. Auto IT can’t help thinking of the time Homer Simpson was asked to put his common man’s touch to work on designing a car...

Download your new Mazda

Why drink stale coffee at a bricks-and-mortar dealer when you can spec your build-to-order Mazda over the web - just like buying from Amazon.com. Well, you can if you happen to live in Japan. And you like being able to order bright red seats. And you can tell the difference between a Scrum Wagon and a wheeled brick.

Keyless or clueless? Mechanical locks still a good idea

First, apologies for the extended break in coverage. That's blogging: sometimes work projects get right in the way. Still, now that the new Audi TT has been unveiled you'll know what Auto IT has been up to (I wish).

Anyway, to business. According to a report in Auto Express, thieves were able to make off with a BMW X5 owned by football star David Beckham, owing to that car's reliance on keyless entry and ignition. The system made the vehicle vulnerable to hackers armed with a laptop and some wireless kit.

The story underscores what's wrong with a lot of the auto makers' adoption of technology: too much blind faith.

The story suggests that the makers thought a 20 minute delay would be sufficient to prevent a brute-force attack, in which every possible electronic combination is exhaustively explored until the lights flash and the doors pop open. But they were clearly wrong.

The story also states that keyless security often relies on "information that only the manufacturer is supposed to know". In the IT industry this tactic is called security through obscurity, and these days it's a tactic employed only by software professionals who are stupid, deluded or con-artists. Secrets tend to leak, and there are better ways to ensure that a system remains closed to all but the most determined prodder.

No security system is foolproof, however, particularly when makers need to provide access to vehicles where owners may have genuinely lost all their keys. The master keys - software or hardware - will be the last line of attack if criminals have no other option. Always assuming, of course, that they don't simply beat up the owner to steal a regular set of keys.