I've been very lucky to have been sent one of Dexter Industries new Thermal Infrared Temperature Sensor.
Whilst Xander over at BotBench has done a more indepth review (here), he was using RobotC and a fairly involved physical construction. I was more interested to see how quickly I could get something up and running in a more basic situcation, typical of what a classroom teacher might experience.
So I took a DomaBot robot, added a few extra pieces to change the Ultrasonic Sensor to point to the side and added the TIR sensor directly below it.
A simple program that drives the robot forward for 30 seconds whilst datalogging both the TIR and Ultrasonic sensor was loaded on.
I then constructed an incredibly complex and involved experimental setup. (ok it was an empty softdrink can, a cold softdrink can, a metal water bottle that had been left out in the sun and a portable freezer pack :D )
And here are the results. Yellow line is the Ultrasonic sensor and the green line is the temperature. You can see on the yellow line, 3 distinct 'dips' as the robot passes the first three objects. The freezer pack was all scrunched up and probably doesn't really give a good surface for the US sensor to detect. Lining up with each of these is a distinct change in temperature (apart from the empty can which expectedly shown no change).
I was really happy with how easy it was to get this up and running. Once the NXT-G blocks were imported, the TIR sensor showed up on the Datalogging selection screen without issue. I orginally tried to log at 25 samples per second but later found out that is was capable of a max of 5 samples per second. Whilst it would have been nice for faster sampling, it was easily compensated by slowing down the robot's speed.
So.... What should I try next?