Toddler's Twittering

A prototype device, Twoddler, allows toddlers to tweet. The device looks a lot like a baby's activity board, except that it has presumably user-inserted pictures on it. If the child plays with a picture long enough, the board sends a predetermined tweet to that person. Twoddler also allows users to press buttons and turn knobs on their Twoddler and make other Twoddler light up and play sounds.

Mostly, this seems like a useless toy. I doubt the child is going to have any idea that they are communicating when they play with the pictures. Possibly, the Twoddler to Twoddler communication could be popular among young children, but I'm not sure that they would realize that their actions influenced their friends board, especially if they are not close together.

Moving Keyboard

One of the problems with using only multitouch interfaces is that typing is difficult. Since multitouch screens don't have physical buttons, the user can't feel if they are on the key. Shawn O'Neil found a Microsoft concept that positions the keyboard under your fingers regardless of where they are on the screen.

This could be a very useful concept. If done right, the keys will always be the right size for your fingers and exactly where you expect. For example, when a little kid is typing the keys will be smaller than when an adult is typing. This could also potentially remove the problem of having to switch between the mouse and the keyboard. At least for me, every time I have to switch, I have to reposition my hands. With the moving keyboard, I could simply move my hand to click and then immediately start typing in the new position.

Lasar Harp

Using a modified Wiimote and some music editing software, it is possible to make a laser harp similar to those found in science museums. Stephen Hobley has designed and built a very impressive laser harp:

Using a Wiimote retrofitted with a laser filter and the WiimoteLib, Hobley was able to communicate with a music editing program. By sliding ones hand up or down a beam of light, the sound of the note changes. This looks like it could be a lot of fun, but using it could be difficult because there is no tactile feedback (as noted on his page).

Uncertainty in Human Decision-Making

In the talk by Dr. Konrad Kording, The Influence of Uncertainty on Motor Learning and Its Neural Representation, Dr. Kording explained the baisen probability model of human decision-making. The basien probability is the product of the probability of an outcome based on observed data and the probability of an outcome based on previous experience.

To test the validity of the model, Dr. Kording created an experiment where the subject had to drag their finger across a board. However, the participants could not see their finger, rather they saw a projection that did not reflect reality. During most of the test, the participant received no feedback, at about halfway across they were shown a cursor that could be displaced by some (usually small) distance from where their finger actually was. At the end, they were also showed a cursor. By measuring how the participants changed their motion, the actual displacement and the predicted displacement could be compared. The model fits the data quite well, indicating that humans us baisen probability to decide how to best respond to uncertain situations.

Robotic Hand

While humanoid robotics are not currently developed enough for general human interaction, there is much research directed at creating a robot that can safely interact with humans. An important sub-system of the robot will be the end-effector, or hand. Ishikawa Komuro Lab has developed a high-speed robotic hand capable of manipulating human objects. Some of it's more impressive capabilities are throwing and catching a cell phone and dribbling a ball.

The system consists of a high-speed camera, a parallel photo processing array and the three-fingered end-effector. Using novel control algorithms, the robot is able to throw a ball, tie a knot and catch a cell phone. This end-effector is probably too fast to be used around humans, but for human-robot interaction to become a reality, a similar system will have to be developed.

Laugh Floor at Walt Disney World

Similar to my last post, this post will also be about computer-generated figures interacting with the humans. Unlike the last post, this really works. Disney has several human-cartoon interaction shows in their parks.

At Walt Disney World they have a show called Monsters Inc. Laugh Floor where characters from Monster's Inc. come out and crack jokes with the audience. Instead of just a pre-recorded show, the cartoon characters actually interact with the audience. For example, a two-header monster comes out and asks guests where they are from. Based on the response of the guest, they react accordingly. Its June and a guest says they are from Arizona, the monster might insist that it is freezing cold in Arizona and start shivering. The Laugh Floor works because of the corny humor and the wow factor of cartoon characters interacting with you in real time.

Milo

A really cool application of Project Natal is that very natural interaction

could occur between humans and computer generated characters.

Milo is one such character. He is a young British boy who lives in

your television. Using Project Natal's interface, he can talk with

you and take you on his adventures. For example, in the

YouTube video, Milo and Clara, the human, were talking. It sounded

a bit scripted, but his inflection and movements were fairly realistic.

Clara discovers that Milo has not started his school project (how

she knew he had one was never explained), so she offers to help.

Milo decides that he wants to catch and draw fish for his project,

so they head over to the local fish pond. Milo tosses Clara some

glasses through the TV, which Clara pretends to catch. According

to the narrator, everyone who has tried Milo out goes to catch the

glasses. After Clara looks in the pond and splashes around for

a bit, Milo wants some help drawing a picture of a fish. Using a

conveniently located paper and marker to draw a fish, Clara holds

it up to the TV. Project Natal recognizes that it is paper and scans

it. Meanwhile, Milo reaches out and "takes" the drawing from Clara.

While I strongly doubt that Milo is currently functional enough to be

released to the general public, as the video implies, some of how

Milo interacts is possible. I also don't know what function a boy inside

your TV would serve, but its really cool nonetheless.

iPhone App Design

A company, Create with Context, has developed a presentation about iPhone app design based on research about how people really interact with their phone. In particular, they looked at an older demographic - 30 to 55 year-olds. From their research, they came up with 8 design criteria*:

1. Take advantage of learned behaviors - Put things where they usually are (i.e. the search box at the top) and make them work the same across applications. For example, make a plus button in the top right always be add.
2. Avoid interaction inconsistencies - This is quite similar to #1 but focuses more on the user action. For example, to edit, the user usually just clicks on what they want to edit. It is a bad design decision to make them go hunting for an edit button which they must click to edit.
3. Provide clear conceptual link across widgets - Put things that go together next to each other and separate things that don't go together. For example, put the "Go Search" button next to the search bar.
4. Put space between action widgets - Leave room between buttons. Remember that fingers are larger than mice and somewhat imprecise.
5. Plan for accidental overswiping - Similar to #4, users are imprecise.
6. Don't rely exclusively on multi-touch - It can be difficult to perform certain actions at times. For example, if the user is holding something in their other hand, they need to be able to perform the action one-handed.
7. Provide visual feedback for taps -Make it clear that the user hit the button by having some sort of action quickly.
8. Provide interaction affordances -Make it clear what the user is supposed to do. For example, if they can scroll up and down, have cut off pictures or text at the top and/or bottom

* Orange, italic Georgia font is taken directly from the presentation; black Arial font is my comments.

While these criteria were designed for iPhone apps, they are applicable to all touchable interfaces. The eight guidelines can be summarized as follows: Make the application interface look and act like other interfaces, leave enough space around buttons and provide feedback.

Touchless Interface

A senior design group from Northeastern designed a touchless computer interface. You can see a video of it in action here. Using electric fields, the device is able to detect the x, y and z locations of objects over top of it. The user then waves their hand over the device to control the computer. For example, the group designed an application that allows the user to draw on the screen. The x and y coordinate determine the position on the screen and the z coordinate determines the color. The color control is very crude; it only blue-ish to green-ish colors.

Another application the group developed is 3-D object rotation. The control they showed worked pretty much just like a mouse would - move right to rotate counter-clockwise, left to rotate clockwise. It did not allow the user to rotate the object in a particularly intuitive manner. I am not sure if it is just because it was demo app or if it because the device cannot pick up that sort of motion. I suspect its some of both.

Virtual Reality with a Wiimote

Johnny Chung Lee from Carnegie Mellon University has devised a set-up to create a virtual 3-D set-up using nothing more than a television, laptop, Wiimote and sensor bar. His description includes a demonstration starting at 2:30.

Instead of the user holding the Wiimote with the sensor bar attached to the TV, the user wears the sensor bar, and places the Wiimote by the TV. Because wearing a sensor bar looks pretty silly, Mr. Lee bought safety glasses with LED lights and replaced the LED lights with inferred lights. Once everything is hooked up correctly, the user sees a 3-D environment with targets of different sizes and locations. As they move their head around, it truly seems as if the targets are hovering in 3-D space. One problem with the system is that only one person can experience it at a time. Also, the size of the experience is limited to the size of the TV. Overall, the system provides a decent, but cheap, 3-dimensional immersion experience.

Interface or Intelligence

Although many of the points I bring up in this post are applicable to any interface, I will focus on online shopping sites. In particular, I will compare Amazon and Target searching for a decent, cheap electric kettle.

The initial effort should focus primarily on the interface and making it easy to navigate. If the public can't find what they are looking for, they won't use site and then even the best "smarts" won't be any good. However, the one "smart" that should be well developed from the beginning is a good search engine. If I don't know exactly what I want is called (for example, I think an electric kettle is called an electric teapot), the search should offer suggestions.

Once a website is well developed, then there should be a balance between smarts and interface. Again, if the interface is lacking, costumers will be annoyed and avoid the site. One of the current, popular bad designs (in my opinion, at least), is menus that pop-up when your mouse passes over them and mostly fill the screen. They block whatever you were looking at and rarely contain detailed enough groupings to be useful. Unfortunately, Target has chosen to use these on their website.

Another very important interface consideration is what the user needs to click to see the detailed product information. Clicking either the picture or the text will bring up the detailed description on most shopping sites, including Amazon and Target.

Not too surprisingly, both Amazon and Target have similar information - features, reviews, additional product information (size, ASIN #, ...). All of this information is considered standard and customers expect the interface to have them. A designer should make sure the interface contains at least the minimum expected information and in any easy to see interface. Any intelligence (for example, the "Customers Who Viewed This Item Also Viewed" section on Amazon) are extras and should only be worked on once a good interface is developed. However, if two interfaces are roughly equal, the customer will probably choose (at least to figure out what they want) the smarter site.

In summary, the initial design effort should be a good interface with a helpful search. Once that is developed, then most of the development should be in developing intelligence.

Wii Fit - How not to design an interface

My roommate got a Wii Fit last week, so of course we have been playing it. Unfortunately, the interface was not well designed. The main problem is that it takes way too many "A" clicks to start playing a game. For example, to play the ski jump game, you must:

1. Select the ski jump game from the balance menu (this is fine and expected)
2. Choose your level/confirm that you want to play the game (this is also reasonable)
3. Click "A" to clear the message about not jumping on the balance board (I suppose they should have this)
4. Click "A" to say that you are ready (not sure what the purpose of this is)
5. Click "A" again to actually start
6. Put the WiiMote down
7. Play the game, once
8. Pick the WiiMote back up
9. See your score and push "A" to continue (this is reasonable)
10. See your ranking and push "A" to continue (also reasonable)
11. Choose whether you want to replay the game or quit (fine and to be expected)

So, in order to start the game you must click "A" five times! However, only three of the clicks actually do anything useful as far as the user is concerned. For the complete game cycle, you must push "A" eight times! And there is no way to skip any of them. It would have been nice to be able to skip directly from step 8 to step 11 once the game finishes (by pushing a button(s)), especially if you know you did horribly and all you want to to is replay the game.

To be fair, if you choose to replay the game, they start you at step 4. Also, it is really only the balance and aerobic games that are this bad. Nevertheless, Nintendo could have done a much better job designing the interface.

My First Post

This is a blog for one of my college classes: CSE 60416 - System Interface Design. The main purpose of the blog is to practice writing (particularly technical writing). I will acomplish this by looking for cool user interfaces and describing them.

In the class, we will be using, almong other things, two Microsoft Surfaces. They are kind of like giant iPhones.