I finally found some time to go through my notes from Ubicomp 2007. Since I already blogged about the first day I’m going to start this overview on Tuesday. This is not a complete overview, but just a list of talks that I found interesting.
The first session had a nice talk titled “My Roomba is Rambo” which studied why people got emotional about their appliances, and why we should care. This is similar to what Philips did with the iCat. Apparantly people seemed to forgive their appliances when they made mistakes, given that they were emotionally attached to them (e.g. helping a Roomba that got stuck).
The next session on location featured an interesting talk by David Dearman on a method to predict location errors. They evaluated their system by letting people locate posters as fast as possible, while varying the location error and using different algorithms to estimate the error, including their own. There were a lot of talks on security, including one in this session on security by spatial reference by Rene Mayrhofer. He made an interesting point, that the methods of security and authentication we use today (e.g. passwords) are inpractical for ubicomp environments.
Shwetak Patel presented his work on Tuesday as well. He received the best paper and best talk award. His idea was very innovative, namely to check for noise on the power lines in a house to detect activity (e.g. opening the microwave would turn on a light which could be detected). The system is quite accurate, although portable devices could be more difficult to support since a training period is required. In the same session there was another security talk on shaking two devices together and thereby generating a unique key for authentication. This illustrates that there were definitely a lot of creative ideas at Ubicomp.
Tuesday evening we had the conference dinner up in the mountains which was quite nice (with an Austrian traditional band that played all kinds of music, including Tom Jones), but it was very cold up there 
Wednesday started with a talk by Tim Kindberg (of Cooltown fame) titled “Merolyn the Phone: a study of Bluetooth naming practices”. He started off with a slide that showed a list of names of detected Bluetooth phones in the conference room. Apparently, the people that featured in his study were more creative than we were (I was guilty as well with the not very original name “Jo’s K750i”). The story behind the name Merolyn the phone was pretty funny as well.
Next was a talk by Yvonne Rogers, of whom I read a very interesting article last year (after it was mentioned on Fabien’s blog). The talk was basically about how Ubicomp technology cannot be evaluated in a lab setting, and needs real-world testing.
Another interesting talk in this session discussed the Whereabouts clock, which reminded me vaguely of the AmbientClock. In the session on privacy Karen P. Tang (if I’m not mistaken) presented privacy controls in IMBuddy, a contextual instant messenger. They allowed people to disclose information at different levels of granularity and get notified when someone queried their presence.
In my opinion, the best presentation was given by Scott Davidoff, who presented speed dating as a method to quickly evaluate different design decisions. His slides are online at Slideshare.net.
The final talk by David Molyneaux showcased an impressive steerable projector system. The innovative part (according to my understanding) was that objects stored and controlled their data (e.g. sensor readings) and metadata (e.g. 3D model) themselves, and decided when to send this to the projector. For example, when two objects with the same appearance are in a room, and one is moving and the other one isn’t (detected with accelerometers), they notify the projector which can then distinguish between them. When an object’s geometry is changed (e.g. when a book is opened), it detects this through sensors and accordingly sends its updated 3D model to the projector.
All in all, the conference was very interesting as was the workshop.
