We created an LED grid that is controlled by live video input.  Lighter values on the video feed illuminate corresponding LEDs, while dark colors turn them off.  We used a LoL shield, which allows for a 1-to-1 relationship between a pin and an LED.  The downside of the LoL shield is that it requires a heroic feat of wiring!  We soldered female headers to the shield and two wires to each LED, roundabout 500 soldering joints.  Then we stuck the wires into their corresponding pins and wrapped the exposed wire with electrical tape.  Finally we glued the LED heads into a foam core grid and covered the grid with two sheets of mylar for diffusion.

Our code uses serial communication between processing and arduino.  Processing reads a video feed and translates that feed into a 14 x 9 grid of pixels.  The RGB value of each pixel is averaged, and based on that average, a 0 or 1 value is sent to arduino.  Arduino cycles through the grid, taking each 0 or 1 from processing and accordingly turning on or off an LED. I worked with Jason Rosen.

OUR CODE:

digidrench is an interactive video installation in which the user controls video playback by filling and draining three tanks. As the water level rises, the video plays forward; as it lowers, the video reverses. The user gets to interact with the very same materials that are used on screen. As the user pours, liquid falls on screen. There is a direct correlation between the user’s action and the video content, allowing for more playful and meaningful engagement with the project.  Collaborators: Jason Rosen, Jee Won Kim, Luca Shapiro, Matt Richardson, Phan Visotyothapival, and Zena Koo.

Our project website is www.digidrench.com.

I observed the self-checkout kiosks at my local CVS.  They are used to facilitate payment without having to go to a manned register.  The first thing I noticed is that of the six kiosks, only three were open.  I’ve actually never seen all six kiosks open at once in all the time that I’ve lived in my neighborhood.  I always assumed it was a theft-deterrent measure, but it’s frustrating as a customer to wait in line while three or four kiosks are inexplicably closed.  If they don’t ever want to open all six, why install them?  I think Norman would take issue with this design.  The designers didn’t consider the emotional state of their customers—people are often short on time when they’re waiting in line, and seeing unused kiosks is irritating to them.

The kiosks have two ways of indicating that they are open for business.  The first is a lighted sign on top that reads, “Express Checkout,” and turns on when the kiosk is available, much like registers in a supermarket.  The second is a message on the screen that tells you if the lane is open or closed.  When I observed the kiosks, one of the lighted signs was broken, so customers got mixed messages about whether or not they could use the lane.

The most successful part about the design is the mix of visual and aural information.  The touch screen display shows options and a woman’s voice prompts you to follow the next step.  She also reads the prices of each scanned item and announces any errors.  Her feedback is very clear for the user.

I think the graphical interface could be greatly improved, however.  Each screen gives you six or seven options, and every option has its own style and graphic.  Blue boxes, grey boxes, red boxes, pictures of extra care cards, and so on.  The screen is so cluttered, you have to stop and read the options carefully.  When it comes time to pay, it asks you to select a payment method on the touch screen.  But the credit card reader also has buttons to select payment methods.  It’s confusing to duplicate the options like that.

The scale also causes problems.  I saw a man buying detergent.  He scanned the item and placed it in his bag.  Then the kiosk voice asked him to, “please remove the scanned item!” over and over again.  Mystified, he gingerly lifted the detergent off the scale and then put it down again.  Somehow that did the trick, and he could continue to check out.  It’s confusing to customers when the same action produces two different results.

I also saw a woman walk up to a store employee and ask him to scan her items for her.  It defeats the purpose of a self-service checkout if you need an employee to operate it.  I wondered why she didn’t just go to a normal register.

The biggest problem with the design is handling error.  The machine freezes, and the light on top flashes.  The voice tells you to wait for assistance, but somehow the message is never conveyed to the store employees.  Customers stand around unsure how to proceed.  Should they go in search of help?  Should they wait?  The feedback from the kiosk and the actual experience are at odds.  Norman would definitely take issue with that part of the design.

A screen shot of our output in cool term:

A video of our graph in processing:

“Attractive Things Work Better,” Ch. 1 Emotional Design by Donald A. Norman

Norman discusses why attractive design might improve usability. The answer lies in our emotional response to design. Norman writes, “Emotions, we now know, change the way the human mind solves problems—the emotional system changes how the cognitive system operates” (1). Anxiety, fear, and other negative emotions cause people to narrow their focus, while positive feelings have the opposite effect—they expand the mind to new, innovative ways of thinking (2, 6).

The potential emotional state of a user is a critical component of design. As Norman writes, “Designers can get away with more if the product is fun and enjoyable. Things intended to be used under stressful situations require a lot more care, with much more attention to detail” (6-7).

Norman doesn’t discuss what qualifies as attractive design. He uses the term as if it is an obvious and objective concept. Attractiveness seems inherently subjective, so I would be interested to know how Norman defines the term.

“The Psychopathology of Everyday Things,” Ch. 1 The Design of Everyday Things by Donald A. Norman

Norman outlines the components of user-friendly design. He stresses “the importance of visibility, appropriate clues, and feedback of one’s actions” (8-9). Straightforward, easy-to-use design is a fairly obvious goal, but its implementation is often far from obvious. I was particularly struck by Norman’s suggestion that “[w]henever the number of possible actions exceeds the number of controls, there is apt to be difficulty” (22). I think many people equate good design with minimalism—the fewer controls the better. But Norman suggests that a dedicated control for each functionality actually simplifies design.

“The principle of feedback” (27) is an especially important idea. The user should know, “what action has taken place, what result has been accomplished” (27). A design fails if a user if left in doubt. This principle ties in nicely with Norman’s discussion of conceptual models. There are three distinct models of an object: the designer’s model, the user’s model, and the system’s model (or the object’s physical interface). Ultimately, the user’s model, the user’s conceptual idea of an object, is paramount. An object may have all the functionality in the world, but if that functionality doesn’t become part of the user’s model, it’s useless. An operational control that doesn’t give feedback to a user is no better than a broken control.

Ambiguous door hardware is a pet peeve of mine, so I was particularly gratified to read Norman’s door discussion. The operation of a door, Norman writes, should be made explicit “by the design, without any need for words or symbols, certainly without any need for trial and error” (3). Preaching to the choir!

In honor of Halloween, I mechanized a doll’s eyes for my physical computing stupid pet trick.  I hooked up two servos to the arduino– one connected to each eye.  I cut through the plastic sheath surrounding each eye socket and connected the servos directly to the weighted ball bearing in the eye.  I controlled the movement with a potentiometer.