Is your car too smart? May 18, 2012Posted by Cameron Shelley in : STV202 , comments closed
Robert Charette at IEEE Spectrum reports on complaints that computerized amenities in modern cars are too confusing for many drivers. Car producers are packing more and more computerized features into the cabins of cars for operators to interact with. Unfortunately, the operators sometimes have difficulty figuring out how to work their new toys. The MyFord Touch system, for example, has attracted complaints that it is too complex for people to operate.
To deal with the problem, Ford is trying to persuade dealers to become more expert in the operations of these new, in-car electronics. With greater training, dealers can then help to train buyers in the ins and outs of their new computerized gear.
Charette puts the problem down to poor design:
Of course, it might help if car designers spent a little more time with their human factors counterparts to make the operations of the electronics more transparent and easy to use. There has been several occasions where I would have been more than pleased to explain in detail to the designers of several of the electronic systems on my Toyota Sienna how they got it dreadfully wrong. Needing a couple of hundred page manual to explain how to use my car’s electronics is a symptom of the problem.
It would not be the first time that poor design sabotaged a new, in-car system. Early versions of BMW’s iDrive system were notoriously difficult to deal with, requiring drivers to use a joystick to navigate a hierarchical menu system to operate the simplest functions, e.g,. the radio, all while driving at speed. Usability has always been an issue for the software industry, which has a tendency to present functions in a way that prioritizes their abstract relationships rather than their practical uses. Now that cars are ever more computerized, these problems with software design have become problems for automotive design.
The response of having dealers train users is reminiscent of the problem of training new drivers when cars themselves were a new invention. In his book, “User unfriendly“, Joseph Corn documents how the first automobiles confounded their users. Cars were radically different than the familiar horse and buggy, and early operators experienced many problems with driving and just keeping their machines working. Auto manufacturers responded by training dealers enough to talk up the features of their cars, and by issuing manuals that purported to explain their purchases to new car owners. Neither effort was much of a success. The situation was amended only when new technologies made cars reliable enough to work without so much fuss by the drivers.
It may be that this will have to happen with in-car electronics too. That is, instead of training drivers to adapt to their gear, automakers will have to design the gear so that drivers can operate it without so much preparation.
Crowdsourcing education February 7, 2011Posted by Cameron Shelley in : STV302 , comments closed
One interesting thing that blog entries can do is to juxtapose two unrelated items to see how they might connect. Today, I have found a couple of things that make an interesting pair. First, there is a New York Times article that discusses the possibility of online courses with no human instructors. So far, no institution has created a course without a human being running the show. Why not? I suppose no one has yet figured out how to do it, although it seems quite plausible.
This is where the second article comes in. This Technology Review blog discusses an experiment where a program managed a group of humans in the construction of an encyclopedia entry, using the crowdsourcing tool Mechanical Turk. Appropriately, the project was called “My boss is a robot.” The experiment was successful enough that its creators are looking for other projects to manage with a similar approach.
(Image courtesy of Mikael Nordin via Wikimedia Commons.)
So, I would suggest that the robot boss be put in charge of organizing university courses. It would be interesting to see what courses are most amenable to this approach. I suspect that many introductory classes, e.g., Calculus 101 or Introductory Psychology, might lend themselves to almost total automation. More specialized courses would be tougher to automate entirely as they involve more specialized knowledge of a field and often need to be conducted in a more flexible and responsive manner. Of course, I could be wrong.
Would such a system appeal to universities? Maybe. Of course, many universities employ inexpensive, human sessional instructors to deliver the low-level mega-courses, so development costs for the robot boss would constitute a substantial barrier to entrance in the marketplace. However, Bill Gates is throwing a fair amount of money into this forum, and he has very deep pockets and was a university drop-out. So, your professor may be a robot sooner than you think!
And I, for one, welcome our robot overlords to uWaterloo!
Smart cars February 4, 2011Posted by Cameron Shelley in : Uncategorized , comments closed
No doubt you know that cars are getting smarter. That is, cars are becoming more computerized each year. There are various reasons why, including:
- Cars with funky hybrid or electric drive trains need computers simply to function. Their complexity rules out a purely mechanical approach to design.
- Cars are becoming extensions of the livingroom or work desk. So, they need electronic entertainment systems to keep occupants amused and connectivity options to keep them productive.
- Cars are becoming computer network objects. Cars on the road can exchange signals and analyze road conditions via networks so that they can avoid accidents and plan routes efficiently.
- Cars with computers onboard risk rapid obsolescence. Therefore, the onboard computers need to be highly generalized and upgradeable. Essentially, wouldn’t it be nice to have an App Store for you wheels? Then, you would not get stuck with last year’s software under the hood.
Naturally, along with the advantages that computers bring come a train of problems. These include privacy, since networked objects tend to leak information like crazy (if past experience is any guide). Your car vendor, the app vendors, and probably many others may be able to track your whereabouts. And what about the police?
(KITT image courtesy of Magnus Manske via Wikimedia Commons.)
Then there are the security issues. We have already mentioned car hacking. Connecting your car to the Internet will mean that people can break into it without even being physically present.
Your car will also become more subject to intellectual property laws. If an app you are running turns out to illicitly contain patented code, it could be disabled remotely, as happened to owners of the EchoStar DVR player.
I suppose there are more esoteric concerns too. As cars get “platooned” into chains on the road, the occupants get “platooned” onto the ‘net. The feeling of autonomy that once was a major attraction of the personal automobile is lessened, in exchange for other benefits. Of course, with a built-in entertainment system, passive car passengers can play Grand Theft Auto as their ride conveys them effortlessly and heedlessly across the blacktop to their next appointment.
Complexity vs. consumerism: The Chevy Volt November 8, 2010Posted by Cameron Shelley in : STV202, STV302 , comments closed
In an earlier post, I noted the argument made by Matthew Crawford that modern design is tending more and more towards consumerism. That is, designs have a tendency to become more complex and less accessible over time. Crawford contrasts early motorbikes with modern ones as an example: Early bikes were open to intervention and modification by their owners. They positively invited exploration. Modern bikes, by contrast, are highly computerized and actively resist intervention and modification by their owners. Crawford deplores this trend as a promotion of consumerism, that is, modern gear is designed to appear like a black box that is merely to be consumed, used up, and then thrown out when it has become boring or broken down. The opportunity for designers to educate users about their gear via its design is deliberately passed over.
Of course, another explanation for the increasing complexity of our gear is that the computerization that makes for more complex designs makes for more efficient designs too. So, designers computerize things like motor bikes to make them work better (a point that Crawford seems to concede) and the trade-off is that their users are excluded from tinkering with them.
(Image courtesy of Mariordo via Wikimedia Commons.)
Wired.com notes the presence of this trade-off in the Chevy Volt. In case you had not heard, the Volt is GM’s entry into the electric car market, with an all-electric range of some 40 miles (ca. 65 kms). The Volt is highly computerized, sporting over 100 electronic controllers, a unique IP address for each vehicle, and an astonishing 10 million lines of code!
For comparison, the new Boeing 787, which is widely considered to be the most electronic airliner ever, has around 8 million lines of code. And that includes the complex avionics and navigation systems. The new F-35 Joint Strike Fighter? Around 6 million.
I can personally image a program (or suite of programs) containing hundreds or even thousands of lines of code, but 10 million just boggles the mind! How could anyone really understand what is going on in the digital depths of the Volt?!
As the Wired.com article points out, this level of complexity will surely inhibit owners from messing with the Volt’s inner workings:
We’re not software engineers here at Autopia, but with all those lines of software code, anybody looking to tweak a Volt may have quite a puzzle on their hands. Sure the days of a new intake manifold and a four barrel carb are long gone, but now it looks like the modern version of ‘chipping’ a car is far from adequate for the new cars on the block.
So, is the Volt the ultimate (to date) consumer gadget, the latest way Detroit has found to turn its clients into dependent ignoramuses? Or is it part of a necessary progression of technology in which consumer education must be traded off for gains in efficiency and performance?