Presentation Day

Today each of us in the senior seminar class presented our project proposals, and I think all of us have interesting projects that we are passionate about working on. Below is a link to my proposal presentation (just the slides, no recording of me talking), and the research proposal paper for reference. The next step will be to come up with a general idea of a design for the robots and order the parts, so my goal is to do those two things over winter break, so I can start work on the project next semester.

Proposal Presentation

Research Proposal Paper

Presentation Draft

A draft of my proposal presentation can be found here.

I still want to probably add 1 more slide about the design or goals of the robots, add in a couple more pictures, and fill in the references for the final version.

Proposal Time

Sorry for not updating the blog in almost a month. I haven't been updating much because there isn't much to say, just working on the final proposal paper and presentation.

I have the final version of my thesis proposal paper completed, which can be viewed here, and I will put up the presentation when I have it finished (probably next week). The semester is already drawing to a close: there is only 1 week until we do our final proposal presentations. Starting in winter break, I will post about the progress of my project with some more interesting stuff.

More progress and Argonne

First of all, I completed a final draft of a literature review for my senior seminar, which can be viewed here.

Also this week, I presented the research I did over the summer on my REU at the Argonne Symposium for Undergraduates in Science, Engineering, and Mathematics at Argonne National Laboratory in Illinois.

As for project related progress, I have been surveying existing projects, and working to answer the questions I posed last week. Hopefully I will have more to report next week.

Progress

In the last post, I noted that the next steps would be to research existing swarm intelligence projects in order to get an idea of the hardware they used, and what tasks the robots were able to accomplish. I'm still in the process of reviewing these previous projects, but I have developed a list of questions that I need to answer.

1. What kind of micro controller do I want to use?
• I don't have much of a microelectronics background, so I should keep this in mind, although there are people who do who have offered to give advice or to assist occasionally.
• This also slightly depends on what language I would use to program the robots, although learning a new language wouldn't be horrible.
2. What kinds of sensors do I need the robots to have?
• The robots will obviously need some type of "sight", but there are many different ways to do this - range finders, ultrasonic sensors, or possibly using image processing with a camera.
• Will the robots need pressure or sound sensors as well?
3. How many of each of these sensors does each robot need, and where should they be positioned?
• Front, back, side, etc.
4. How will the robots move?
• Do they need multiple motorized wheels, and if so, how many?
5. Is every robot going to be identical, or will some be specialized?
• My first idea is to make them all identical, but later on I may introduce different types of robots.
6. What kinds of tasks will the robots accomplish?
• This is a significant question, and impacts many of the other questions, especially the questions related to sensors.
• Over time, I'd like to get them to accomplish more difficult tasks, but all of the tasks will be similar in some way. For example, they may all involve searching or gathering.

By next week, I hope to have reviewed a sufficient number of previous projects, and have at least a general answer for each of these questions. After I have answered these questions, I should be able to start on a design for a prototype robot, and then start work on my project. (And possibly decide after making the prototype that the design needs to be changed)

Finally, a thesis topic!

First of all, I completed the final draft of a literature review for my senior seminar, which can be viewed here. (No embedded viewer this time, I had too many problems with that)

After considering the topics I proposed in my last post, I've decided on a specific topic for my senior thesis.

I want to create a handful small robots, which can work together to accomplish specific goals. I will be using ideas from swarm intelligence and swarm robotics to create a "swarm" of small independent robots which can cooperate to complete a certain task, without any outside control. There are still many questions to answer, such as what components and sensors will the robots need, what type of micro controller to use, and what task(s) they will accomplish, but I at least now have some idea of what I will be doing (and can research specific materials related to this topic).

Obviously, the first step would be to research how others have designed similar robots, and what tasks they were able to accomplish. I would start out by designing a prototype robot, and see if I can get it to achieve a certain simple task on its own. Later, once I finalize the design, I'll create a couple more identical robots, and then have them communicate with one another to achieve a harder task, without any outside control from me.

Thesis Topic

I've been hard at work trying to find a specific topic for my senior thesis lately, but I realized something. While some of the applications of swarm intelligence seem interesting, they don't seem fun. This is one time where I can choose nearly any topic I want to work on, and I want to really enjoy what I'm doing. Therefore, I decided to change my decision on something I dismissed early on when searching for my project - I actually do want to work on a project that involves some hardware or microelectronics.

I had originally dismissed the idea of doing anything with microelectronics because I'd never done it before. However, I've never done most of the stuff I would be using for my project anyway - part of the project would be to learn these skills. Working with hardware and making something tangible seems much more interesting to me than just running a simulation or applying swarm intelligence ideas to some existing algorithm; it seems like it would be fun.

Based off this new change of plans, I have two main ideas for projects:

1. Create a few small robots, who can work together somehow to accomplish a goal. I would be using some ideas from swarm intelligence and swarm robotics here, and applying them to actual hardware.
2. One of my CS professors got one of these to fly around in my building at Willamette during our weekly CS Tea meetings. I think it would be awesome to get this thing to fly around on its own, and if I get that working, even get it to do something like be able to fly from one point in the building to another, or recognize faces, or practically anything else. This would be moving away from my initial idea of a swarm intelligence type project, but I would still be applying many ideas from Artificial Intelligence.

Both of these projects seem really interesting, and fun, and I think I would be able to accomplish something significant with them. It may not be groundbreaking new research, but it would be a great learning experience, and I might even end up doing something that hasn't been done before, or at least not in the same exact way. If I could find a specific task to give the small robots, I think that might be a really interesting project, and if I am able to accomplish it quickly, I could always get them to do something more complicated. Maybe have them start by finding some object, and then having them all huddle together around it?

All in all, I think that I will have a lot more fun and be a lot more motivated to work on one of these projects than one in the area I was thinking about before. Both of these projects would still involve looking at previous research (and some other sources) and they can still probably be considered research, especially if I do something that hasn't been done in the same way before.

CCSC NW

Note: this post isn't directly related to my senior thesis, but I feel that its is relevant enough to post here.

Last weekend I presented a poster from my summer REU at CCSC NW. I was awarded 2nd place for my poster presentation, and I had a great time talking to some of the professors about my REU and my plans for grad school. One of the things I liked most about the conference was the presentation given by the keynote speaker, Dr. Perry R. Cook. Dr. Cook talked about a lot of cool things that he did at Princeton - applying computer science to music. However, one thing that I really liked about his presentation was the creation of new instruments, using simple microelectronics.

Early on in attempting to find a senior project idea, I quickly dismissed the possibility of doing any kind of microelectronics or hardware. However, working with microelectronics is one thing that I've always wanted to do, and have never had a chance to work with. I think I may have to reconsider my decision to dismiss any type of hardware related project as a possibility for a project.

Project Topic

I'm still trying to find a specific project topic for my senior thesis, but I have been narrowing down the possibilities quite a bit. As I have mentioned earlier, I want to do a project in the field of swarm intelligence. I don't think I have the expertise or time to come up with an algorithm from scratch which would counter existing ones, so I am most likely going to be applying an existing algorithm of swarm intelligence to a particular problem (and possibly making some slight variations to the algorithm afterwards). I also don't have the hardware experience to implement this with robotics, so I will most likely be running a simulation to test the application of this algorithm.

 If I can find a topic that the algorithm has not been applied to yet (in literature) then it could be considered research, otherwise my thesis would be just mainly a software implementation project for the purposes of learning. Therefore, my main goal now is to find an application for the swarm intelligence algorithm.

The applications that would work best are those in which there is a definite measure of performance - where it would be easy to show if, and how well, the swarm intelligence method is working (compared to some other method). I think my best choices would be a simulation of simple (enough) natural system, or a simple task (simple path finding?). I'm going to try and meet with one of my professors at Willamette University this week (one who specializes in Artificial Intelligence and Simulations) to try and further refine my list of possible applications, and hope to have my application picked by next week so that I can actually start work on my project.

Assignment 1 - Bibliography and Annotation

Below is a list of five papers related to Swarm Intelligence, with annotations.

Update:
I was having problems with the embedded document causing the blog to auto scroll to it, so I've removed the embedded document from this post. You can find the document here.

Project Ideas

As the weeks go by, the pressure to find a project topic is building. I haven't found a specific project idea that I'm interested in yet, but I have some general ideas about what I'd like to do.

Generally, I want to do a project in the field of Artificial Intelligence or Machine Learning.

Computer Vision / Image processing might be an interesting topic to look in to, and I've found some interesting papers involving: solving jigsaw puzzles or detecting pedestrians. Some of these papers are really cool, not only in the computation and math behind them, but also in the problems they are solving. It is really interesting when the problem that is being solved is something that a human can relate to - such as solving a jigsaw puzzle.

One of my initial ideas for something that I found interesting (but haven't been able to turn in to a project yet) was the idea of having multiple independent robots which work together to achieve a certain goal, sometimes referred to as Swarm Robotics. For example: this or this. I don't have the background, knowledge, or time to actually implement something like this in hardware, but I could possibly simulate it or do a project specifically related to Swarm Intelligence.

New Blog

The blog has been created! I will use this blog to note and comment on anything related to my senior seminar project at Willamette University.

There isn't much to see yet, but hopefully there will be soon.