It'll still be a whole semester till my thesis but I am starting to think about its subject. The institute of CS and the institute of mathematics both offer various projects but they're all not to my liking. Most of them are a safe (and boring) bet: Implement a QT User Interface, a MATLAB toolbox for Image Processing, et cetera... The good news is that we may suggest a topic of our own. I'd like something more challenging and am willing to risk my bachelor in exchange for a cool project. Keep in mind that I am still a grasshopper though, so ideally it would tease the explorer in me while still allowing me to tackle the whole thing. So what is cool? Frankly, I don't know, that's why I'm posting. Cool could be something that spans across a lot of different areas. I've had few exposure to hardware problems/designs so a problem involving a bit of that among other things might do the trick. Cool could be one of those projects that once completed behaves in completely new, not pre-defined ways (modelling just a framework of rules, like AIs or physics). Cool could be something with pulsing LEDs, LCDs or any other kind of show effect. Cool could be a theoretical work about how to improve a certain unimportant, long forgotten but still interesting algorithm. Cool could be something that I could put to real use after all is said and done.

.. Ideas, anyone? I know we've got CS here, so what was your Bachelor Thesis?

I'm prepared to attack unknown territories but for what it's worth here's the stuff I know about. ( [] := Know about it, but don't want to touch it again if possible. )
Knowledge:
- C, C++, Java(Enterprise), Assembler, [Basic, Pascal], Design Patterns
- Database Systems (Relational Theory, SQL, Oracle, JDBC)
- Formal Languages and Methods(State Machines, Turing Machines,...)
- Mathematics(Differential and Integral, Vectors & Matrices, Logic, Linear Algebra, Numerical Computations, Statistics...)
- Computer Architecture
- Datastructures and Algorithms (Theory and common examples)
- Operating System Concepts
Basic knowledge:
- Programmable Logic Devices (Xilinx)
- Computer Graphics (Open GL)

None of us can guess what excites you. That's really for you to decide. You've mentioned a variety of things, some of which are eye candy, others of which are anything but eye candy. A better place to start would be to hunt around SourceForge or whatever for a project that catches your interest and propose to improve it in some way.

For example, you might take one of the empeg projects out there and add new features to it. Maybe you could do something with the microphone input on the empeg, such as an automatic parametric equalizer setting gizmo. Or, if you're less inclined to do audio processing, you could write the glue to make an empeg synchronize with iTunes the same way as an iPod does. Lots of options, and that's only in one small space of ideas.

I would strongly Suggest a 6.5 Inch color touch screen kit that just happens to be a direct bolt on replacment for the VFD on an Empeg

None of us can guess what excites you. That's really for you to decide. You've mentioned a variety of things, some of which are eye candy, others of which are anything but eye candy.

All of the things I mentioned, eye candy or not, were in fact examples(a small subset) of things that excite me. What I need is a formal frame, a project title under which I can do things that excite me. I did search the web and the usual suspects but didn't find the spark I was looking for yet. Since this BBS has proven countless times that it is a pool of clever heads, of which many have hands-on experience with projects/theses, I figured I'd ask what you guys have come across. It seems like you have a pretty decent idea about all the perennial favourites like the genetic algorithms used to breed pretty pictures. Definitely a cooler programming project than anything we did up to now. What else do your students like?

I'm a CS, and in lieu of a bachelor thesis, we did a senior programming project involving neural networks to analyze stock market trends. We did well on it, but not well enough that we were able to put it to use to make a fortune.

My Masters project, however, was more interesting (to me, anyway) and involved creating, imagine this, a "Smart Music Player." I don't think you have to go far to think of what inspired me to pitch that to my team. Unfortunately, I was one of two software developers on the project, so while I think we did a good job of the planning, requirements, analysis, and early design, we never got to do any implementation (which we weren't necessarily expected to, since the scope of our project was so wide.)

Not sure if the format of this project is anything like what you'll be expected to do for your thesis, but if you want something to get your mind going, take a look.

http://www.tonyc.org/files/SmartMusicPlayer.pdf

Honestly, different students like different things. You need to somehow narrow down your options. Do you want to do a graphics project? An AI project? A new feature for the empeg crowd to drool over? Do you want to add some feature to an e-mail or calendar client?

I could rattle off project ideas for pages and pages, but you first need to narrow down your interests. You also need to say something about how much time and energy you've got to spend on the problem. For some people, hacking Gnome Evolution to have decent shared calendar support is a weekend hack. For others, it's months of work.

How about writing a program to do point-to-point routing using the TIGER line data sets (Census bureau maps of U.S. roads)-- quickly and with little memory usage. If someone could give me an algorithm like that maybe I can even integrate it into GPSApp-Roadmap.

Sure, that'd be cool. I'm looking forward to driving up freeway onramps, or getting onto bridges that cross over but don't connect to my path. Always wanted to be able to do that.

Where's my Sky Car?

Point taken- but it would still be a fun exercise and in my little area I doubt I would never have to worry about one-way streets and overpasses. A little bit of clever code could detect freeway onramps anyways. I've also heard that the next TIGER data set would actually include intersection information at least and maybe even one-way information. Of course they may not be doing that until the next Census in 2010 I don't know..

...take a look.

That was a most interesting read, thanks Tony.

... TIGER data set...

I thought about empeg related projects for quite some time since it would have killed two birds with one stone. One thing that caught my attention and was within my grasshopper abilities was to use the MIC input for some limited software voice recognition. Nothing fancy, it would have been a template matching based, speaker dependent and isolated word recognition approach, which probably wouldn't ever have worked inside a running car and would have eaten up a lot of CPU. I abandoned the idea when Patrick suggested a far superior approach involving dedicated hardware.

You need to somehow narrow down your options.

Fair enough. I always want to do everything at the same time. No calendar, AI sounds interesting... but give me a few days and I'll give it some serious thought.

... how much time and energy you've got to spend on the problem.

The upper limit, which I'll fully invest without hesitation, is:
1.5 semesters (4 month, 3 evenings(1800-0100) per week plus one full day(1200-0300) per week)
1 summer (3 months of which half can be spent full time(1400-0500) on the project including weekends plus the evenings(1900-0100) of one month). It's a single person project, I tend to work slow but exact (over-engineering) and I need to thoroughly document the whole thing.

Assuming 4 weeks per month we get 1.5*4^2*(3*7+15)+6*7*15+4*7*6=1662 man-hours minus bath breaks minus time to prepare(not eat) food.

Sorry for the awkward way to express this but I don't know about a standardized way to calculate man hours (though I'm sure there is one somewhere).

You need a girlfriend, mate.

Sorry for the awkward way to express this but I don't know about a standardized way to calculate man hours (though I'm sure there is one somewhere).

Hehe. Let me introduce you to the most important skill of project management that they _never_ teach at Universities (and we wonder why so many business projects are late) - Take that number and multiply it by 0.8. Then, when scheduling tasks, calulate the number of man hours you think they'll all take and double that number.

Hmm, okay. My guess is that the 20% are lost due to unproductiveness and the 50% lost on top of that are due to an underestimation of the problem or overestimation of ones/the teams capabilities. About right?
So theoretically I could go for a project that's rated at 1662*0.8*0.5=664.8 man hours.

Take that number and multiply it by 0.8. Then, when scheduling tasks, calulate the number of man hours you think they'll all take and double that number.

Amen.

You need a girlfriend, mate.

Girlfriend? Do they come in through the firewall or by other means?

EDIT: I've seen one of those girlfriends in my snort log. Not with me, buddy.

Assuming 4 weeks per month

Maybe it doesn't matter all that much... but a month is really four and a third weeks. (52 weeks in a year divided by 12 months = 4.33 weeks/month.)

That gives you another 8% to work with.

tanstaafl.