I recently spent nearly two weeks using a rental car - a 2013 Ford Explorer - on a family vacation. We drove from New York to Chicago, around Chicagoland and Northern Illinois, and then back to New York, with daily trips in the car. One of the highlights was taking advantage of the My Touch infotainment system in the vehicle. My car at home is a decade-plus old Subaru, so "hi-tech" for me means a cassette deck. Call it a busman's holiday, but I was excited to use an up-to-date interactive system that's in so many vehicles (and is somewhat notorious). My first and lasting impression is that the UI is more suited to a tablet than an automobile. The screen was reasonably responsive, and there was some pairing with physical (albeit digitized) controls. Giving Ford the benefit of the doubt, I'm sure they tested it in a moving vehicle, but I found the the typography too fine, the layout cramped, and enough nomenclature and flow quirks to trip me up to the point of defeat. I couldn't escape the feeling that this worked well in a simulator and on someone's tablet, but hurtling at 70MPH down the highway is an entirely different experience.
It seems to me that a serious usability test for vehicle manufacturers building these systems is how quickly a car rental customer could figure out basic tasks. Like climate control, the stereo, and then synching a mobile phone, getting directions, and on to more complex tasks. Like a lot of rental car drivers, I get into a car in the airport parking lot and drive off without stopping for couple of minutes to figure out controls beyond the mirrors and seat. Frequent rental car customers might actually have an advantage, since they could be exposed to a multitude of systems, but I don't know how much thought people give to how things work. Perhaps a "simple mode" for frequently-accessed items, and — this is a minor annoyance — some way of resetting everything so that a new renter doesn't have to encounter a previous user's radio stations, trips, etc.
Not nearly as ubiquitous as the airticket redesign projects (1, 2, 3, 4, etc.), but perhaps much more meaningful in potentially saving lives — and as a design exercise, significantly more challenging — is the vehicle touch screen redesign exercise. This one caught my eye as being one of the better ones of the breed: it showed an appreciation for the context by attempting to use multi-touch to overcome the lack of physical affordances in a screen; another one addresses Tesla's attempt to overcome the problem (by going all-in with a giant screen and no physical controls) with an attempt at splitting input and output into two different screens. Actually, the Ford Explorer I rented had some elements of this, although it seemed they were unable to restrain themselves. In the end, I don't think a heavy reliance on a touchscreen is a good idea - reserve them for what computers do best: complex, nested, changing features, and use physical controls (with digital refinements) for the top 3-5 user needs.
News reports tell me that Apple and Google/Android are getting into the car UI business, which may be welcome news. Its difficult to say if touchscreen GUIs are too far outside the expertise of carmakers to do well, or its an issue of large companies not being able to do small things well. While an outsider perspective might work well, the decades of deep expertise carmakers have is extremely valuable and can't be discounted.