.: I got an email asking what I knew about gesture recognition for human-computer interfaces in cars.
Unfortunately, I know almost nothing, and only found the briefest mention in a quick search. Any ideas? Let me
know (I'll post what people send).
.: One of the small surprises for me at the recent SAE Digital Car conference was the continued importance of Bluetooth to telematics, as opposed to 802.11b wireless. Extended Systems provides various Bluetooth wireless software development tools, and they recently signed a contract with Johnson Controls for JCI's BlueConnect product. Extended Systems already have a relationship with Motorola as well as other OEMs, including IBM, which resells Extended Systems software as "IBM MobileConnect."
.: I don't quite know what to think about Ricardo Martinez, (former NHSTA administrator who gave a keynote speech reported on earlier) want to be the company that creates the Global Safety Data Vault to manage the data from car black boxes.
.: I went to an IBM class on embedded Java, so I didn't do anything
else--I even missed lunch. Overall, the Digital Car conference was a
success: lots of good new technology and good energy to make driving
safer and better.
.: Most of my day today was on the Digital Car Exhibition
floor. Some brief observations:
.: The keynote lunch speaker was Dr. Ricardo Martinez, a former
NHTSA administrator. He gave an engaging and interesting speech about
safety. He made a strong case for "black-box"-like technologies in
cars, since NHTSA needs lots of accurate data to help make cars
safer. Two things he said that surprised me somewhat: If we could (on
average) increase by 1 second the response time before a rear-end
collision, we could decrease fatalities from these accidents to half
their current level. He also had some good words to say for voice in
the car: I think he believes that we've made (as a society) a choice
for the convenience of using phones (etc.) in cars over safety, and so
it's a matter of decreasing the danger of cell phone use, and voice
could do this.
Tuesday, March 5, 2002
.: I spent my day at the Digital Car Conference with IBM. They're
running a repeating morning session and afternoon workshop on their
software development platform for telematics and embedded systems. IBM
is using their size, longevity, and commitment to standards as key
selling points of their solutions to the automotive manufacturers and
top component suppliers. IBM has relatively recently announced their
Eclipse open source software development environment and the (more
proprietary) "Websphere Studio Device Developer" environment (even one
of the IBMers was willing to say this was a terrible name). Still,
this looks very interesting, and the IBM booth is demonstrating
several devices running on a couple of different car buses all running
their J9 embedded Java virtual machine with remote debugging
capabilities. Cool stuff.
.: W.S. "Ozzie" Osborne, head of IBM's Voice Systems, gave the
keynote lunch speech. He spoke more about "pervasive computing" than
voice per se, and he made some very good points about the importance
of designing speech into an application from the start, rather than
just an add-on to a system. He said the (next) major barrier to
acceptance of voice in cars is the acoustical problem of recognizing
speech in the car environment, but claimed that 1000-word at a time
systems were possible today for cars today, with different recognition
models being tuned for a car at rest, at 10 mph, at 50 mph, etc. In
fact, the engineer on the floor was showing a speech system that
seemed to be performing well even in the very noisy environment of the
show floor--the microphone is just sticking out into the air (no
chamber being used), and it just worked.
Monday, March 4, 2002
.: Digital Car Journal, live again. In honor of SAE's Digital Car
Conference and Exhibition, the Digital Car Journal is live again. I'm
at the conference, and here are some notes and commentary.
.: The Digital Car Conference doesn't start until Tuesday, so today
I attended the SAE technical session on "Human Factors in Driving and
Automotive Telematics". Peter Roessger of CAA gave a presentation on
"Intercultural Differences in the Interaction between Drivers and
Driver Information Systems," reporting on some cross-cultural surveys
that CAA has done. Among the general claims:
- Japanese (in general) don't mind "deeply
structured" menus, like "HiTech" interfaces, want entertainment and
games, accept parallel interfaces (like multiple maps on a display).
- Europeans (in general) want sequential, easy-to-use interfaces that are obvious
and clear. They want "high value," and the look and feel is
important. Use of local language (i.e., not English) is important.
- Americans (in general) want to talk; they are lonely in their
cars. Like Europeans, they tend to be sequential. They want sports.
It might be unfair to stereotype people in this way, but it is the
case that systems developed in a culture for that culture tend to be
more successful than systems developed by people across cultures.
Yauhiro Ihara of Panasonic described an "HMI Design and Evaluation
for the Driver Information System." He contractdicted Roessger a bit,
saying that intuitive and easy-to-use systems were guiding principles
for his design. One of the interesting design ideas Ihara presented
was that of a "virtual radio" for all audio; this consistent interface
for radio, various news feeds and music feeds applied the standard radio
metaphor of a "channel" well. I'm not really doing this justice, I'm
afraid, but it seemed to me that the constraints on human audio
processing (we can generally only listen to one "channel" at a time)
made the ability to move around various virtual channels a powerful
metaphor for streaming audio.
Some mechanical engineers from an accident reconstruction company,
Triodyne, presented some history of how the law has treated driver
reaction to "Sudden Emergencies," which is, apparently, a legal
doctrine that goes back to the 19th century and coach-and-four
days. People make bad decisions when presented with sudden
emergencies, and their legal liability is under consideration. These
guys reported that expectation failure, emotional stress, and
uncertainty of the effects of one's actions, and the actions of
others, play a big part in bad decisions. I was reminded of the Buridan's Principle, "A discrete decision
based upon an input having a continous range of values cannot be made
within a bounded length of time," which applies even under ideal
conditions.
Finally, Vivek Bhise reported on some work investigating the
"IVS-DEMAnD Driver Attention Demand Model," in which the task being
undertaken, the traffic conditions and the age of the driver are major
inputs on stresses to a driver's attention. This was a good piece of
work, and I'll look for more information on it .SAE has a rather
draconian policy of charging $10 for a single reprint of an article
even for attendees.
