There are various factors which hinder the in-car-computing technology to be a part of the regular feature in the cars of the future, this include:

Price barrier

This more expensive and robust version of in-car computing won't hit the road for a few years. When it does, it's likely to be more expensive than the AutoPC. Analysts believe that the $1,000 to $2,000 price range for such in-car computing is in line with the cost of other options introduced in the past. Anti-lock brakes originally cost about that much but are now standard. Radios, which are now commonplace, cost about $110 when first introduced around 1930. That was a hefty option price when the average car cost $700 and a luxury car tipped the scales at about $2,000. The radio took 20 years to be accepted and reach a price point where it was acceptable to consumers.  With silicon manufacturing capabilities and advances, the pace is so much faster. In six months one can make tremendous steps forward. In two or three years you can get this technology integrated into silicon at very low cost." Despite the fact that concept cars have been built, there are virtually no cars rolling down the streets today with in-car computers. At present, GPS receivers are more common in Japan and Europe than the United States, primarily because of the cost/benefit ratio. It's harder to get lost on the U.S. road system, so there's less need to use GPS to tell where you are. For the more general purpose in-car computers, the problem is the same: the cost is too high. At a meeting of the Intelligent Transportation Society of America (ITS America) in Detroit in May 1998, a representative from Toyota said that what the automotive industry needs is five times the computing power of today¹s desktop units for about $500. Given Moore's law that the number of transistors on a chip doubles every few years, the desired computing muscle isn't that far off. The result will not only be units that are cheaper but that can do a better job on such tasks as speech recognition.

Temperature barrier
Another barrier is that a car is not a desktop. While the temperature inside most homes hovers at around 70 degrees, the temperature inside a car plunges well below 0 and soars above 100. The same wild swings happen with humidity, dust, vibration, and other environmental factors. These variations may occur mainly when a car is parked, but to a driver, a parked car is simply a car that isn't moving. "The product has to work when you fire up the engine. If it doesn't work, then obviously you can't do anything.Having to work at lower temperatures is a problem not only for the electronics but also other system elements. A liquid crystal display, for instance, freezes at very low temperatures and washes out at very high ones. Manufacturers are working to overcome issues like these, and solutions are in sight, but they haven't been implemented yet.

Human Error
There's also the question of safety, driver safety. It does little good to tell a driver how to avoid traffic congestion if doing so causes an accident by distracting the driver. The same problem arises when drivers are listening to e-mail and become too engrossed in it. Many such human factors must be considered. The problem becomes especially complex with some of the more advanced implementations, which might make use of heads-up display technology from fighter planes.

Networking Problems
Another safety concern is a more subtle one. Cars are already crawling with computers in the engine block, the brake system, and elsewhere. There are even networks to connect these devices together. It would be nice to pluck diagnostic information from such computers, but it has to be done safely. Data intended for consumer consumption cannot be allowed to worm its way onto the car information bus. That could lead to such potential catastrophes as air bags deploying or brakes malfunctioning at random. That's why all in-car computing schemes put a gateway or firewall between the consumer system and the in-car operational computers.

Product or Service?
A hidden issue of in-car computing is the interface between the car and the transportation system. In the old days, this interface was confined to where the rubber hit the road. In the future, there will be a great deal of information some of it public, like traffic conditions that drivers will want. At present it looks as though such interface systems will be the result of public-private partnerships "Putting the Smart in Smart Highways". A variation on this infrastructure question is who will pay for the air time. All of the in-car computing schemes make use of wireless data exchange for such features as traffic information, Web connections, and personal phone calls. Depending on how much usage there is, this could add up to some significant minutes per month.

Advertising a solution
One suggestion on how this might be paid for is through the sale of advertising. However, instead of a message played over a radio, the approach might be something like this. Say the car, through sensors, recognizes that it's time to get gas. The car can tell the driver it's low on fuel and point out there's a station up ahead three blocks on the right. The driver gets gas, and the station gets targeted advertising to customers who need its service. In return, the station, or the national gas company, would pay for the customer's wireless air time. There are, of course, other possible methods, including that of having the driver pay.

Time and Technology barrier
However, there's another interface to consider the one between the car and the computer. Today a computer is obsolete in a few years due to technological advances. Over the same time, cars remain largely unchanged. They certainly don't become obsolete, replaced by something that's twice as fast for half the price. One solution to this divergence would be to replace the in-car computer every few years. This implies that in-car computers won't be like traditional car accessories, but then they may not be anyway.

Speech barrier
In theory, speech recognition is easy. Take the output from a human voice, run it through an analyzer, match it against the output from all known words, and magically the corresponding text appears.

In practice things are a bit more difficult. Take the case of a phone conversation. The telephone network limits the speech frequencies transmitted to about three kilohertz. Actual speech, particularly the letters "s" and "c," uses up to eight or nine kilohertz. Those higher frequencies don't make it down the telephone wire. The human mind and ear are good enough to extract information despite this truncation, but for machines, the lost data is vital. Similar problems arise due to slight variations in how individuals talk.

That's one of the reasons why today's best dictation programs require head-mounted microphones and have to be trained to individual users. That's also why speaker-independent systems, such as those used by in-car computers and the like, can only understand a few commands. It's easier to discard the vast majority of audio input as noise and concentrate on just a few words.

The onward march of semiconductors will help, of course. As more computing power becomes available, what today is impossible accurate transcription of telephone conversations, courtroom proceedings, and business meetings will become very possible.

Smart in Smart Highways
Somewhere beyond the current crop of in-car computers lies science-fiction visions of cars that drive themselves. Nowadays dubbed intelligent transportation systems (ITS), these systems were at one time called smart cars and smart highways.

While none of the proposed in-car computers will drive a car, they are capable of intermediate ITS functions. For instance, getting accurate traffic information would enable drivers to avoid congestion. With GPS pinpointing location and a wireless modem to connect an in-car computer to a network, everything would be in place to deliver up-to-the-minute traffic information. Everything, that is, except getting the data to the driver.

Increasingly, state departments of transportation and other government entities are posting traffic information on Web sites. For the average commuter, however, it's impossible, and downright dangerous, to try to scan a computer screen while driving. What's more, simply getting a mass of information about an entire metropolitan area's gridlock woes isn't much use. What's needed is something that filters the information and packages it in a way that is useful for the commuter. Today this doesn't exist. However, just such a service may soon be in place. You're starting to see there will be private companies that will offer that interface.