Project Gado: An Open-Source Photo Digitization Robot
Project Gado: An Open-Source Photo Digitization Robot
Project Gado is creating a $500, open Arduino-based robotic scanner for digitizing sensitive photographic materials in small archives
Project Gado is creating a $500, open Arduino-based robotic scanner for digitizing sensitive photographic materials in small archives Read more
Images are powerful. They have to capacity to inspire and inform, to enforce stereotypes or refute them. Getting access to good historical images, however, can be tough. Digitizing and sharing images takes time and money, resources which many small archives simply don't have.
This is a problem for teachers, scholars and anyone interested in learning about their past. Without images, teachers can't plan lessons, scholars can't do research, and individuals lose a window into their own history.
This is especially a problem for minorities; most archives with the resources to digitize their visual collections primarily document white American life. Vibrant minority histories around the country are literally going unseen, as historical photographs languish and deteriorate in inaccessible archives.
About the Project
Project Gado seeks to help solve this problem by developing an open-source robotic scanner which small archives can purchase for $500 and use to digitize their photographic collections autonomously.
The Gado 1, completed in September 2010 with support from the Johns Hopkins University Center for Africana Studies, uses suction to lift photos, scan them at full 600dpi resolution, and place them into an output bin. Using advanced sensors, the Gado 1 is able to scan each photo while applying less contact than a human operator.
All aspects of the robot, from the Arduino-based design to custom database and OCR software, is open-source and has been released into the public domain, for others to use and modify.
Gado Needs You!
The Gado 1 is in operation at the archives of the Baltimore Afro American Newspaper, where it has already scanned nearly 1000 historical images. The paper has a collection of over 1.5 million images spanning 115 years, though, so there's still a long way to go!
The Gado 1 is a great proof-of-concept, but the machine itself is tricky to assemble (wood glue and MDF, anyone?) and overly complicated to use. During Phase 2, Project Gado will work with the JHU Center for Social Concern to create a simpler, manufacturable version of the machine (Gado 2) which other archives can purchase in kit form and assemble themselves.
The machine will then be used to digitize as much of the Afro American's collection as possible; the goal is 20,000 images by June 2012!
To make this happen, Gado needs you and your support!
Where Will My Money Go?
I (Tom Smith, see my bio below) have been leading Project Gado as an undergrad at the Johns Hopkins University for the last year. I've now graduated, and am making this my real job!
Project Gado was the recipient of a generous grant from a Baltimore area grantmaker, and will continue to receive partial support from JHU's Center for Africana Studies and Center for Social Concern. This funding is enough to pay my salary for a year, and cover basic equipment costs, ensuring that a robot will be made!
There are other things, however, that would really help the project thrive:
- Promoting the project at Open Source events like Makerfaire
- Hiring a student employee to help develop the machine and accelerate the scanning (scanning > 20,000 images would be great)
- Subsidizing the cost of a Gado 2 kit for archives in Baltimore who can't afford the initial investment
With your support, I hope to make these things possible, and help the project reach its full potential!
I Crave Details!
About the Machine
The Gado 1 is basically composed of a lot of MDF (medium density fiberboard, think sawdust and lots of glue) and parts from Sparkfun.
The machine is a bit like a CNC mill with a flatbed scanner in the middle. It can move back and forth (y axis), side to side (x axis) and up and down (z access). It can also open and close its scanner's lid (i axis) and turn on and off airflow through a suction lifter, which it uses to physically grasp and position photographs.
The scanning sequence starts when the machine travels to the input bin. It lowers its suction lifter, turns on airflow, and lifts a photo. It then travels over to the scanner, drops the photo, and gets out of the way. The scanner lid closes, the scanner digitizes the image, and the lid opens again. The robot comes back, sucks up the image again, and drops it in an output bin. It snaps a photograph of the back of image using a 720P webcam (in case there's useful meta data on there), and then repeats the process. The whole thing takes about 2 minutes.
Why not just use an ADF (Auto Document Feeder), you ask? Historical images are sensitive; many photos in the Afro's collection are over 100 years old, and a lot of images have notes or pieces of newsprint taped on the front or back, standard practice back in the day.
Auto Document Feeders on normal scanners use wheels or rollers to move images; those rollers can tear up the surface of a photo, rip off any attachments, and otherwise cause irreparable damage. Plus, you're always one paper-jam away from destroying a priceless piece of history. No good!
Software and Guts
The brains of the Gado 1 are an Arduino (with Protoshield) and a USB connection to a PC.
The PC runs custom Python software, which communicates with the machine's firmware over a serial port and handles tricky things like automatically triggering the scanner (TWAIN is annoying, but useful) and doing image conversations. The Arduino runs the control code for the machine itself (all custom!). All code is here.
The x and z axes use stepper motors, driven by the excellent EasyDriver board.
The y axis uses a linear actuator driven by a Sparkfun motor driver (the slide assembly is HEAVY).
Suction is provided by a standard vacuum cleaner, which the Arduino controls by way of a Powerswitch Tail to avoid electrocution. The i axis uses a servo and a neat pully system to raise the scanner lid; the machine can work with any off-the-shelf TWAIN scanner.
The Gado 1 achieves its light touch using a mess of different sensors. A photosensor in the suction lifter measures the light level inside the cup; a quick drop means the cup has hit the photo below it, since pressing the cup against the photo blocks incoming light. An infrared proximity sensor just behind the cup takes readings on distance as well, for backup.
Each axis has 2 limit switches, and the y axis has several photointerrupters (again, thanks Sparkfun) to help determine its position.
The Gado 2
The goal for the Gado 2 is to:
- Still cost $500
- Scan an image in 30 seconds @ 300dpi or 45 seconds @ 600dpi
- Be buildable by a normal person
To achieve the last item, we'll need a kit! Instead of using MDF and a table saw, parts will be laser-cut in advance, so they can be screwed together. The custom electronics (currently a mess of wires on a blank PCB + a bunch of solder joints) will be turned into an Arduino shield for easy snap-together assembly, and plug-in connectors will replace the solder.
To speed things up and keep costs down, the Gado 2 will use rotational motion instead of linear motion; the bins and scanner will be arranged in a circle around a central pillar, and the lifting arm will spin around to move from bin to bin, rather than going from side to side as it does now. This will allow the machine to use servos instead of stepper motors, which will be far cheaper and less complicated.
The cost savings on parts will then be put toward manufacturing. Again, I hope to be able to sell the Gado 2 at-cost for $500, which will cover both the parts and the manufacturing. If it comes in under $500, the extra will be put towards continuing the project!
I'm hoping to start immediately! Building the Gado 1 took about 3 months, from design to working prototype. I expect the Gado 2 to take 6 months, with another 6 months for testing and refinement at the Afro.
- (30 days)