Master Thesis proposal

Certainly. But then again, any Masters thesis done well will be a
useful object for a future job. The question is, which job? A robotics
based thesis will appeal to those looking to hire robotics experienced
individuals. Using a PLC on some other control application will
probably appeal to 10x as many potential employers. There's more money
in, shall I call it sub-robotics?, than in robotics itself.

If you aren't experienced in robotics, 4 to 6 months of hard and
focused work, will qualify you as a rank beginner. (You could probably
make a kit into a functioning robot, and may or may not add one or two
unique additions showing a slight improvement to the overal body of
the field.) So to an employer, you will look like someone with an
interest, but not someone well versed in robotics. (I say this after
~40 years in electronics, and after 6 years of intensive pushing
myself in robotics, and even though I am now an University instructor
in robotics, I just consider myself becoming somewhat rounded in some
limited areas of the field. Robotics is so interdisciplinary, there is
so much to learn, that it takes a long while to pass all the
preliminaries, and get to new research.)

Your greatest danger is choosing too broadly, and having nothing
unique or new to show at the end of your thesis.

Randy

Hi again, I wanted to amplify on my above post with something written
on SeattleRobotics list by Robert Scheer descibing his first year in
robotics:

"The most difficult and time-consuming part of innovating is in the
electronics and especially in the sensory processing, computational,
control and behavior domains. In my one year in this hobby, I've
already learned how to program in C, designed my first 3 circuit
boards, programmed several AVR microcontroller applications,
confronted the technologies of echolocation, computer vision, PID
control, GPS, odometry, gyros, accelerometers, magnetometers, Kalman
filters, wireless LAN, serial communication.... and many more. I've
set up embedded Linux computers that control and exchange data with a
handful of microcontrollers. I've learned how to share memory between
programs running in parallel, how to run a GUI controlling one
computer on a separate, remote wireless tablet computer, how to create
a boot image in a USB flash memory stick and so forth. I would very
much like to accomplish stereovision obstacle avoidance and EERUF
sonar in the next couple of months. It seems like the challenges of
making robots smart enough and able to sense the world well enough are
still the big challenges. ... There are too many other things getting
in the way. I'm not smart enough. I don't want to cram one more
esoteric subject into my aging noggin, eg digital signal processing
just so that my robot won't hit that tree. Sitting another hour at the
computer will drive me insane."

It has taken Robert a year to get a handle of some of these essoteric
subjects. If you aren't already up on most of those subjects,
"echolocation, computer vision, PID control, GPS, odometry, gyros,
accelerometers, magnetometers, Kalman filters, wireless LAN, serial
communication" you may find 4 to 6 months simply too few to get to the
working robot you had anticipated.

Randy

It's not clear what you have in mind as regards "new robot control",
but a literature search is the first step in any academic research
project. You'll find, for AI and robotics, that Citeseer is a great
resource, and has 100s if not 1000s of papers on robot control
available online ...

http://www.google.com/custom?q=citeseer+robot+control

Click on one, and you'll find links to 50 more. Iterate.

Personally I would not do a thesis on "robot control." Though it looks
fancy on your VC, there are virtually no "robotics" jobs out there, and
the related fields may consider a thesis in robotics to be too limiting.

The same day the OP asked about a master thesis idea, someone else wrote
a job candidancy post looking for industrial control engineers. THAT is
the work you want to go after, because it exists today, there's plenty
of it worldwide, and it's basically a "robot" that has its base bolted
to the floor. Some of the latest factory automation lines even use AI.

-- Gordon