I haven't blogged in a while, so I thought I'd share the only bit of news I can think of...
My thesis now has a title:
"Using sensor ontologies for real-time reasoning and geotechnical hazard monitoring in a spatial decision support system"
Now that's thesis-y
Subscribe to:
Post Comments (Atom)
4 comments:
That's good and thesis-ish.
Most people get sleepy and/or confused reading the title alone, gotta love it! :)
Mine will be something like
"Soft Fault detection and control of a sensorless dc motor drive in an electric power steerng system". '
Hrmm, still not quite as thesis-y as yours. Heh. :)
Give it to your supervisor, he'll make it thesis-y. That's what happened to mine. My original was something like "A sensor ontology for the purposes of reasoning".
Oh, and sensorless? SENSORLESS? That's not cool man. Our lab is becoming like sensor web HQ...sensors can solve all problems, and here you are taking them away. You suck :)
Yeah, sensors should abound!
"Sensorless" is a bit misleading.. With any control system, you have to have some data fedback (for a closed loop system, that is.)
In this case, there's "electrical" sensors (errm, well, a small resistor for current sensing and measuring the input voltage), but there's no mechnical sensors, ie speed sensor (encoder.) Fear not, there are still sensors! :)
Sure, a control system would be the best if every single possible variable had a sensor (that way you wouldn't have to estimate anything), but it's often impossible or too expensive to do so :) Plus, many sensors like torque sensors or even encoders can decrease the performance of the system somewhat.
What the heck is a "sensor ontology", anyway? Does that mean using a lot of sensors to solve a bigger problem or something?
An ontology is a "specification of a conceptualization" (at least in terms of AI/comp sci...it has a different meaning in philosophy).
Basically I'm creating a standardized description of geotechnical sensors which represents more information than typical metadata (info which describes the data). What sensor webs typically have now is an (x,y,z) coordinate, timestamp, and a unit associated with each value. I want to start representing things like how the sensors are connected to each other, their spatial arrangement, etc. The kind of semantic stuff that allows applications to perform reasoning on the data as opposed to just notifying someone when "x > 4".
Post a Comment