Education RSS Feed
In Your Classroom: "Luis von Ahn & Human Computation"
by In his demonstration of CAPTCHAs, those ubiquitous pattern recognition challenges we face when we need to authenticate ourselves as humans on the internet, Luis von Ahn describes challenges computers have with images. He talks specifically about two very tough problems, optical character recognition (OCR) and labeling images with words. By presenting humans with activities such as solving CAPTCHAs to create a new yahoo account or image labeling games such as ESP, von Ahn is orchestrating a grand teaching experiment -- teaching computers about image recognition through many millions of hours of online work and play.
In my class, we are working toward the solution of an image analysis problem to help our northern New Hampshire community deal with snow. Called Project Argus, a series of students have been working during the past three years to build a network of small outdoor computers that will sense snow and ice conditions on remote roads and warn our local highway department. With better information about remote road conditions, our local highway department can make better decisions regarding resource allocation and optimize road clearing efforts.
What I will present here is a description of where my students started and how far they have come during the past three years of this extra-curricular activity. Supported by the Lemelson-MIT Inventeam program, my students have spent many hours in independent study classes, after school invention club meetings, and their own homes tinkering with hardware and software to collect remote environmental data. This blog entry is less of a "how to" and more of a "look what they are doing". I do this for pragmatic reasons. They have not actually solved the problem yet. But they have tinkered with some pretty cool hardware and software that might have application in your classroom today or in the future.
Where Did We Begin?
Matt, Tom, and Travis were seniors three years ago. Having had them in three previous science courses, I knew their computer and problem solving skills. I asked them if they wanted to help a local teacher solve a computer network problem. Sure, they said. The problem was that the teacher wanted to monitor his school's greenhouses remotely so he and his students could ensure that the plants inside were doing well, especially over weekends and school vacations. Matt, Tom, and Travis set to work on this and decided to use a device called a "slug", a device that Matt had already started to tinker with on his own time. "A slug?", I asked. "Yeah", said Matt, "a slug is a $70 linux computer that we could use to control webcams and sensors. It has an ethernet port so we can tie it into the network and log into it from outside the school!"
It sounded good to me. Matt, Tom, and Travis' skills surpassed mine so I was excited to learn from them as their solution to the problem unfolded.
What Did They Do?
What started as a greenhouse monitoring project quickly evolved into an outdoor environmental monitoring project. Years before, my students had done some physics and engineering research for our town. When our town's manager heard about our current outdoor computer project, he had an idea for us. After several meetings with leaders in town government and the highway department, the students identified the issue the town wanted to address with the Argus system. The issue was snow and ice conditions on roads. The town was interested in the student's system because it would allow them to check on the condition of the roads at key intersections before heading out in the trucks.
By the summer of 2006, Matt, Tom, and Travis assembled a weatherproof computer, webcam, and sensor package that was placed on the roof of the school. This test of the system would show that all of the component systems worked together -- linux based computer, webcam, and sensors -- and that they could use the computer as a webserver so they could create webpages of data for the town.
A Multi-Year Project
Matt, Tom, and Travis graduated in 2006 but they are still working as alumni members of the team. They work as technical consultants to the team. Matt continues to provide us with linux and programming assistance. Tom created a website for the project which serves to keep alumni members of the team connected. It has helped to ensure that all members of the team share a vision for the direction the project is going. The website can be found here.
What Tom, Matt, and Travis started during the 2005-2006 school year has continued through the present school year. Last year, team members Anson, Ben, and Greg joined the effort. Their goal was to continue talking with the town and create a prototype system that could get us closer to the actual task of monitoring a road. In order to achieve the goal of monitoring critical road intersections, the system needed to operate by its own power and communicate via wireless protocols. Whereas the first year of the project produced a system that could operate outdoors, provide webcam images, and sensor data, the 2nd year of the project produced a system that could be powered by a solar panel and utilized short range wireless communication.
This year, Anson is joined by Alex, Tom, Chris, and Ben. The current team of Argus students are exploring "long range" wireless and higher resolution image techniques. As you can see from the movie sequence below, our current webcam produces only small images. If we are to help the town discern between wet, icy, and snowy conditions, we will need better images. By adding better imaging techniques and long range wireless, we should be able to set up an Argus system 1-2 miles away from the school and provide the town with rich data as a test case.
After three years of work completed by nine students during their independent study classes, after school and weekend time, we are now very close to the system first brainstormed in 2006. With each year's work, we have added more components and increased the level of sophistication of the components of the system. Because the project goals have remained the same and alumni students come back to help current students (and me!), we do not start at square one each time. We can move forward slowly toward achieving some pretty grand goals!
Image Analysis?
I started this blog entry by linking this project with Luis von Ahn's work. In actuality, we have yet to fully assemble the hardware that will allow us to do the kind of detailed image analysis for the town that will yield a desired outcome -- whether the road is wet, icy, or snowy. That, however, is the lofty goal we are working toward. As a high school project, this is not something you can give to a student for homework and expect it completed with all the i's dotted and t's crossed on Monday morning. It is however, a great high school project that can be passed on from group to group. Tom, Matt and Travis built the first working outdoor system. Anson, Ben, and Greg took it from there and added short range wireless communication and solar power. This year, Anson, Alex, Tom, Chris, and Ben are re-building the base computer to add long range wireless and better imaging hardware/software to the system. Each year builds upon the previous, slowing assembling a very cool project that will help our town deal with the winter conditions that we face six months of the year!
Resources
I provide here a list of useful links if you are interested in working on a project similar to Argus.
Hardware We Use
Tiny computers that currently run or can run linux:
NSLU2
Alix 1c
Vernier Labquest
Webcams:
Logitech Zoom
Wireless ethernet bridge:
Linksys wireless game adapter
Temperature, Humidity, and Pressure sensors:
iButton Link 1-wire sensors
1-wire pressure sensor
Solar panel
Unisolar solar shingle
Charge controller
Built from a schematic from CirKits
Software
Linux:
NSLU2-Linux (we currently use unslung)
Ubuntu
Debian
Webserving:
Apache via unslung linux (see NSLU2 link above)
1-Wire sensors:
One Wire Weather
Image acquisition:
gphoto
Image analysis:
ImageMagick
ImageJ
Many Thanks and Other Credits
The students and I are not alone in our work on this project! I can not complete this entry without thanking the following supporters for helping my students and me on this project:
• Town of Littleton Highway Department
• Littleton High School Faculty, Staff, and Students
• Hugh Galen Vocational Center Faculty
• Lemelson-MIT Inventeams
• Tufts Center for Engineering Educational Outreach
• John Wheeler, Dan Holmquist, and Dave Vernier of Vernier Software and Technology
• Cisco Systems
• Ed and Nancy Roberts
• Dick Ratico of Solarwind Electric
• Forrest Cook of Cirkits
• Tony Ffrench
• Steve Halasz
• Jon Aurnou
• Elizabeth Turtle
• Sarah Turtle
• John Turtle
• Ralph Lorenz
Additional WIRED Science Video Segments
Don't forget to check out our Video Section for other segments from WIRED Science that you can use in your classroom.
Tags:
0 Comments
+ Add Comment
Post your comment