A Different Focus

Here’s an exerpt from one of my textbooks that I feel exemplfies what I love about engineering:

“… be open-minded about the design problems posed. Don’t approach design problems with the attitude of trying to find “the right answer,” as there is none. Rather, be daring! Try something radical. Then test it with analysis. When you find something that doesn’t work, don’t be dissapointed; instead, realize that you have learned something about the problem you didn’t know before” (Norton, “Machine Design: An Integrated Approach”).

For the caliber of “nerd” that the author no doubt represents, you have to admire his enthusiasm, even in the face of dissapointment. While I love a good abstract challenge, the excitement of someone who has been pushing to solve problems that most people have never even thought of is so helpful to hear about during my own problems.

Well, my anniversary with Ashley went very well. We took a trip to Balboa Island between Los Angeles and San Diego for the day, then came back for champagne and a nice dinner. Spending time with her has been wonderful, and excitement is building as we continue planning for next year (she even went wedding dress shopping, which is a strange thing to think about).

After a tough semester startup due to demanding courses and a short summer break, things are starting to wind down towards “auto-pilot” mode. This is my favorite part of school because by now I have a good idea of what the future subject matter will relate to, and I can revert my focus towards friends and family. I have been spending a lot of time lately in and out of class getting to know some more engineering students, and I feel like we work really well together so far. We have a few projects that have been off to a good start, so we’ll see how they go. Outside of class, I’m taking some time today and tomorrow to head to the Stone Brewery in Escondido with two of my friends; a trip we’ve been looking forward to for a while. Planning and preperation for excusions like this, no matter how small, alleviates any frustration of homework by helping me focus on a reachable, tangible goal.

In terms of class work, I have at least one presentation every week this semester, but with our PDR for robotics coming up soon, the practice will be tremendously beneficial. As part of the vision team, I’ve been spending time with my group reviewing last years vision algorithm, researching and testing a potential new algorithm that uses dynamic color detection so we don’t have to manually reset our robot’s color recognition for each testing/competition surface, and finally experimenting with our new stereoscopic camera, which we hope to utilize for range detection of obstacles. In our PDR, we will discuss the previous algorithm, and compare with these potential improvements.

Now, when I signed up for robotics, I was planning to do something more mechanical than this, but I was placed into the vision team because, well, no one really wanted to be there. To be honest, I was unsure because I have minimal experience with video editing, so I had no clue how an IGV’s  (Intellegen Ground Vehicle) programming could analyze a visual input from a camera. Fortunately, with the help of our project leader Rome and some studying on the side, I can explain our algorithm as such: the software takes the image input and, using a color scale reference, removes everything in the image that is not white (we use two painted white lines on grass as guidelines for the path that the robot is not supposed to cross). The program then sets everything removed to black and the white lines to a bright white as a binary input that the computer can more easily analyze. We do some cleaning up of rogue white pixels that don’t represent the actual lines (including any barrel and sawhorse obstacles, which are handled by our laser range finder), as well as a perspective correction to account for the angled camera placement (I don’t have pictures available on this computer for now). From here, we seperate the image into a left half and a right half, which are analyzed seperately. A Hough transform is performed, which is a feature extraction technique to provide us with a line function that best represents our boundary lines mathematically. In order to assist the robot in the event that only one boundary line is visible at a time, the program generates a “ghost line” that is a slight offset mirror of the actual line for each side, providing the IGV with two opposing lines to follow at all times. From this data, we generate a histogram that gives the robot a proper heading in degress from the current orientation. All of this is calculated by the robot at a frequency of roughly 20 times a second.

The dynamic color detection is our primary focus this year, in which we generate a program that will detect the color range of a certain, predetermined section of the image, and uses that as a template for color removal. So far, we have found a program that accomplishes this to a degree, but it when it selects all areas of the same color, it does not remove any obstacles of differing color within that area, such as an orange cone in the middle of a grass field. Further, depending on the time of day and the IGV’s orientation, the robot may be afraid of it’s own shadow, which is not what we want. Currently we are working on tweaking the algorithm to fix these issues.

The stereoscopic camera responsibility has been handed primarily to our graduate student partner, but I have been helping out due to fasination with the technology. A stereoscopic camera is an assembly of two cameras that focus on the same image with a slight offset, mimicking a pair of human eyes, in order to allow range detection through trigonometric calculation. While not new technology, very few teams present at the IGV competition last year used stereoscopic vision for navigation, so we have a potentially strong source of design credit. Experience with the stereoscopic camera thus far has been optical testing for color, brightness, and range parameters, adjusting calibration settings as we go along. While I don’t forsee us relying solely on the stereoscopic camera this year (we would like to simplify our current setup of a single camcorder and laser range finder), I believe our practice this year will pave that path for next year.

The IGV competition and the group members that I work with have been a trememdous experience that helps me get a grasp for some potential opprotunities in the engineering world. While I’m working with a problem that is new to me, I am happy to report that I’m picking things up quickly, and I hope to be the leader of vision for next semester. In the meanwhile, I have some beer in Escondido waiting for me.

Advertisements

~ by MichaelStaudenmeir on September 24, 2010.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

 
%d bloggers like this: