Topics for thesis

[kissandra79]

I am in my final semester of my masters program. As part of the course work, I should present an elaborate thesis at the end of this semester. My majors are robotics and software development. I already know that I want to do my thesis on a topic related to robotics. But I am so very utterly confused about what aspect of robotics i want to work on. I have spent the last few weeks trying to decide on a topic. I visit various websites looking for ideas....and I also browse various tech websites looking for some inspiration. I've been doing this for the past three weeks and have so far been unsuccessful in coming up with a decent thesis topic. Can someone please give me any ideas?. Or at least point me in the direction of some robotic inspiration? Not having a topic and not having spent the past three weeks usefully, I feel a little defeated and restless.. .

My skills are c,c++,python,Linux,embedded programming, mnicrocontrollers. I've dabbled with an arduino before and also have some experience with real time operating systems...

The truth is..even though I have a few ideas of my own; I want to do something that will look good oin my resume. Because after my masters, I want to work and I feel that I have a better chance of getting a job if my thesis was based on some real world consumer or industrial aspects of robotics...

Thanks. Any ideas...however wild and outlandish will be appreciated .

[corrado33]

Shouldn't a thesis be an original project?

It doesn't have to work, as long as you tried something extraordinary.  

Or even perfected a non-original technique.

[Admin]

[rant]
A thesis isn't about building a robot. It's about doing science, of which a robot may or may not assist you with.

Unfortunately, the robotics field is infested with 'I made a cool robot' papers - written by people who only want to be 'published' for resume purposes. They don't actually answer any important research question, or even say how what they learned with their robot can be applied to any other robot.

It's understandable to feel this way, as it's important for our careers. However, it does a huge disservice to science. I often have to waste weeks reading through dozens of junk papers just to find a few golden gems of knowledge
[/rant]

Keeping that in mind, your paper should be about answering some important question that will be really useful to others making robots. It's likely to not result in a cool robot.

For example, one thing I'd really like to know is the range of a wireless transceiver with respect to water salinity, frequency, and baud rate. Knowing this will be very useful for underwater robots, but I haven't seen much of any information. This requires you to get a whole bunch of transceivers of various capabilities, then go test each in various lakes/beaches. Useful for robotics - but no cool robot will ever be made. But that's science!
(btw, I've done some rough tests myself on this topic)

Or another example . . . What is the optimal spring/dampener ratio for a legged robot of mass X moving at velocity Y? You'll have to make a robot leg, then try out a ton of different weights, springs, dampeners, and jumping speeds to see how fast and energy efficient it is (using current and force sensors). Then you make fancy 3D graphs . . .

To me, the question of energy efficiency of various robot designs is rarely answered. If you focus your thesis on something like which design is more efficient, or how to optimize efficiency for a particular form of robot locomotion, that information will be both unique and useful. For example, which is more energy efficient: helicopter, or flapping wings? For all cases, or just some?

[Billy]

But I am so very utterly confused about what aspect of robotics i want to work on.

Do something that is good for the world.

• Allow the blind to navigate without a cane?
• Prevent senior people from falling(or detect when one has)?
• Detect a sleeping driver and park the car?
• A cell phone that stops drunk dialing?
• ....

You're looking for a solution to a problem you haven't defined. Here is what you should do:

• Find a problem you want to solve. A problem that really needs solving. A problem that has real world consequences.
• Think of 10 ways to solve the problem.
• Choose the one that most interests you.

If it turns out to be robotics, then great, but if you solve it without robotics, that's still OK. If it's truely good, you've done good in the world. That is more than most of us can say.

[corrado33]

[rant]
A thesis isn't about building a robot. It's about doing science, of which a robot may or may not assist you with.

Unfortunately, the robotics field is infested with 'I made a cool robot' papers - written by people who only want to be 'published' for resume purposes. They don't actually answer any important research question, or even say how what they learned with their robot can be applied to any other robot.

It's understandable to feel this way, as it's important for our careers. However, it does a huge disservice to science. I often have to waste weeks reading through dozens of junk papers just to find a few golden gems of knowledge
[/rant]

....Other stuff...

I agree with you admin.  Research isn't always about something COOL and exciting.  In fact, research is USUALLY dull and boring.    The thing is since most research is so specialized now-a-day, not many people get to work with a finished (fun) product.  

For example, I have a friend working in antimicrobial coatings on fake hips etc.  Does she ever get to see an actual fake hip?  Heck no, well, probably but not very often.  She stares at culture plates all day.  It's boring, but important.  

Same with this kinda stuff.  Sure, studying efficiency in a jumping or walking leg would be boring, as you'd be looking at the same leg over and over and over and over and over and over and over (get my point yet?) and over and over again.  But, the point of research is so that someone can use it in the future.  Maybe an engineer designing an automated assembly line would use your work to decide which parts to use or what material.  

Another example, I did my undergrad thesis on a PART of a potentially antimicrobial molecule that could inhibit an enzyme only present in microbes.  I did ONE tiny part.  I understood it, and when I graduated my work was used to further someone else's work.  I never got to see the whole picture, but I understood my part, and I did it well.  Doing scientific research is like being on the worlds biggest (and smartest) team.  Everyone does their part.  

If you want to DESIGN robots for companies, you'd probably need to be a robotics engineer or something like that. (I'm speculating here)

Since your strong suits are programming, you might want to think about thinking up a new localization algorithm or something like that.  

[Admin]

corrado33, I completely agree.

Since your strong suits are programming, you might want to think about thinking up a new localization algorithm or something like that.  

Perhaps one that'll work on microcontrollers

A birds' brain isn't much bigger than an ATmega640, no?

[kissandra79]

Thank you everyone for some interesting ideas. Even though I am still not sure what I will do, the posts on this thread has given me a new approach to thinking about what I want to do. I was always thinking about a finished product like a robot or some kind of fully functioning software. But after reading admin's post, I realize it is not necessary as long as something new and original is done or perfecting something that is non-original like one of the post's mentioned. I'm reading a few papers on motion planning to see if there is anything I can improve on. In any case, I will let you guys know my topic once I settle on one : ) Thanks once again -)

[Admin]

more advice:
http://www.societyofrobots.com/robotforum/index.php?topic=14241.0

[Fr0stAngel]

kissandra79
wow...i had no idea more people would be in the same dilemma as mine...research thesis for Masters  
I'm currently doing my Masters in Mechatronics, and while its not completely Robotics, i'll share of what i've learned so far,

- you must have some friends working in industrial environments, try asking them what sort of machinery (related to robotics ofc) is there, maybe that will give you an idea to optimize or improve an already working process or machinery
- any safety system robots could run better than humans...
- a recent trend in robotics is robot swarms, or multi-agent robot systems, which cooperate to perform different tasks..you could take a look at that, although it involves the domain of artificial intelligence.
- try to think of your budget and resources before you start ( in my case, university funding is not enough for something elaborate, so i had to turn down a couple of ideas due to lack of resources =/)
- check if you have someone in faculty with the expertise in the domain you are going to work in, that would be a huge help.

actually i'm just getting used to being all nerdy etc...i was just a mediocre student at Bachelors level, but truthfully, SOR inspired me while i was working on the final project. it was then i decided to go for Masters, because i wanted to learn all that i had missed, and so much more,. I wish i could personally meet Admin and thank him, but for now thank you and *Jazak-Allah-u-Khair* ( may God give you better reward). I hope ill use my knowledge one day to help future generations too.

[Admin]

In terms of funding . . . find a company that has a small research department, and ask them if you can do a research project for them (for free) in exchange for some funding to pay for equipment (or access to their equipment) that you need. They might have a neat project for you to do that could make them money while giving you a chance to do good research: it's a win-win for everyone.