Reflection

I think after completing this module, I can say probably for my whole group, that i've never realised how much HCI is really a philosophy and how it teaches you not to think in a conventional way. We all knew the complexities of the area and quite a bit about the theoretical aspects of HCI picked up from the module we completed last year. Yet I believe that quite a lot of the blank areas that were left void from last year have been filled in. We've learnt how to apply the theory to something in a choosen doman i.e our intelligent car and how to carry this through from initial idea to a final solution. Anyway, I've really enjoyed this module and hopefully I'll be seeing nicely designed software from my fellow students in the future.

(P) Summarising the Project

Here is a brief summary of the life cycle of our design process:

1. Brainstormed initial ideas
2. Refined idea of intelligent car - brainstormed
3. Worked out specific areas we wanted to concentrate on designing
4. Scenario of GPS usage
5. Task analysis on navigation and music playback
   
6. Researched components of our design - GPS and music streaming
7. Researched use of GPS to give driver road information
8. Looked into problems drivers face with visibility and how it could be improved with a HUD
9. Very early prototype demonstrating initial ideas
10. Looked at practical design decisions needed to be considered in order to stream music to a car
11. Looked at detecting eye movement to improve interface but later decided to abandon due to technical limitations
12. Released a questionnaire in order to qualify the assumptions we had made
13. We made prototypes on a white board, using images and animations in order to improve our idea by gaining user feedback
    
14. We asked a domain expert to criticise our idea to further improve our design
15. Refined our idea for hazard avoidance by researching into colour and motion
16. Looked into incorporating emotion as a variable in music playback selection in the car, but abandoned the idea because it was deemed to be too complicated for the end-user
17. Identified the problem of unreliable technology and hence the legal issues surrounding technical failures
18. Prototyped voice recognition by setting scenarios and getting people to talk out loud the commands they would say to achieve given tasks
19. Carried out a prototype to find out whether people react quicker to different stimuli - in our case, we used a drinking game where we found that red tape around a thumb is a better stimuli than a naked one

(P) Reaction game (experiment)

To test the reactions of people (for our hazard avoidance system), what better way to do it than a good old pub drinking game? We have taken a well-known game and adapted it to test people’s reactions using colour for our hazard avoidance system.

The game is simple. There are players (4 in our experiment) who are given 4 cards each that are shuffled up. There’s a total of 16 cards in our game (4 per person) and they are 4 sets of numbers (4 Aces, 4 Kings, 4 Queens and 4 Jacks). The idea of the game is to get a complete set by passing on a card that you don’t want to the player next to you when it is your turn. Once a player has a complete set, they must place their thumb on the edge of the table but continue to pass the cards. It is then the last person to realise that someone has a complete set (by noticing the thumb on the table) who looses.

The way we adapted this was to time how long it took the loosing players to realise there was a player with a complete set. We did the experiment on a “pine” coloured table (so not far off thumb colour). The first 6 runs was a standard game and we didn’t tell them what we were doing. We just told them that it was an experiment and we wanted completely fair results, which they all agreed to.

The results are as follows:

We then took the players away and attached red tape to their thumbs (in a nice safe and easy to remove way) and again, did not tell them what we were doing. We then repeated the experiment and did another 6 runs.

The results are as follows:

Clearly, the red tape helped the players to notice the game was over a lot quicker than without the red tape. When the thumb had no tape on, it blended in very well with the “pine” coloured table, but with the red tape, the thumbs stood out a lot more.

From this experiment, we have concluded that using colour to identify possible hazards will decrease the time it takes the driver to realise that there is a potential hazard on the road and therefore give the driver more time to react safely.

(P) Intelligent Car Article

I came across an article about an intelligent car. This article both envisions levels of intelligence in a car and explains commercial technologies that have been developed with improving the driving experience in mind. The article goes through a typical family journey and highlights key improvements that the intelligent car makes. It links to a great diagram explaining the technologies behind displaying HUDs on the car windscreen (see image).

Many of the features of the car mentioned in the article have been discussed in our design, but some have not. I'll leave you to check out the article yourself if you want the details. Interestingly, it doesn't mention any hazard notification system at all. It only goes so far as explaining a system to automatically adjust car position, e.g. when it is getting too close to a car, or when the car is swerving slightly out of the lane.

(P) Questionnaire Design

In order to write an effective questionnaire a number of points needs to be considered: clarity, leading questions, phrasing, and prestige bias. When designing our questionnaire we were careful to make the process fun and remove repetitiveness in order to get a good response. Yet, there were a few mistakes that we inevitably made that on reflection we would have improved. I believe that overall the clarity of our questionnaire was good although in a few questions we used technical terms i.e. acronyms that could have been avoided. All questions that involved frequency were clear because we used qualified choices that were not open to interpretation. A few of our questions were leading in either the question itself or the answers not allowing a full range of responses. While in terms of phrasing, questions need to take account of the positive and negative connotations of words to prevent a biased response. In our questionnaire I found little use of emotive adjective or verbs and believe the majority of the questions were as neutral as possible.

Finally prestige bias is the tendency for people to answer in a way that makes them feel better. When asked about how long it takes a person to learn an task they are more likely to embellish their answer to make them seem smarter than reality. There is no real way of removing this problem but my placing our questionnaire on the Internet I believe its perceive anonymity helped to reduce this problem.

(P) The Final Prototype...

An interactive flash animation of the final prototype can be viewed here. It's the final result of many weeks of constant tweaking as part of the user centered design.

Please note its 3.5mb in size so give it time to download before you click the buttons. I hope this gives you all a good indication of our project and have fun!

In a few posts time we will give a summary of the design decisions we made during the development of the prototype.

I want one for Christmas!

I can’t seem to believe what I was reading when I was doing my usual BBC News website crawl. Remote controlled insects! At first, I thought they meant robotic insects… but no!

They are on about inserting a chip into the larva of the insects (butterflies, moths etc) and then “re-wiring” the insect so that the chip can essentially remote control the insect and report back vital information. The idea is in the metamorphosis stage (when the caterpillar is in a pupa) will accept the foreign body and rework the internal organs around it.

The pentagon want an army of these insects so they can be used to sniff out explosives and report back sound and video footage of what they see and find.

Is it me, or is this just the most ridiculous idea you have ever heard???? I’m honestly lost for words on what they are suggesting… but if it works, I WANT ONE! Completely unrelated to HCI but what a post its made….

Full story can be found on the BBC website.

(P) The prototyping...

As Mike discussed below, we set up a small prototype, found here (Music Player), where we asked people to say what voice commands they would use to do certain tasks.

The Music Player

The first voice command which we asked our volunteers to use was to change the songs in the play list. The most common answer was ‘next track’ and ‘previous track’. This does seem the most obvious response to use as you are specifying ‘next’, which is the direction you want to go, and ‘track’ which is a typical word for a song.

We then asked our volunteers to change the music playing to a radio station called “Rock FM”. The most common response was ‘Rock FM’ with no prefix of what system would need it. When we asked why they said this, we had a few people explain they would expect favourites that you can programme into the system. Without any suggestion of such features, people are already starting to explore the possibilities of the system. This gave us great feedback for our prototype.

The Satellite Navigation

For the Sat-Nav questions, we had some very mixed replies.

We first asked them to tell us how they would ask the system to go to “Bristol Road, Birmingham”. One person used the phrase ‘Navigate to…’ where another used ‘Directions to…’ and others who simply said ‘Bristol Road, Birmingham’. This could prove difficult to implement a system that could be able to understand all of these responses. From this we concluded that it would be better if the system questioned the driver on where they would like to go. For example, it would say ‘What is your destination?’ This way, the system would know that a place name was going to be said back to it and it can plan the route.

For appending a destination to the end of the current route, almost every body used the word ‘add’ followed by a place name. No one used the word append and no one was unclear if it was replacing the route or adding the route.


From this experience, we were able to see how a system could be implemented using natural language. While understanding every possible command would be incredibly hard, giving the user prompts like ‘What is your destination?’ will narrow down the possible responses that the system would have to deal with, and therefore be able to respond to them better.

(P) Music Control

There was some discussion today concerning how we would change the volume in our car by using our voice. We came to the conclusions that if the volume is really high, our car wouldn't be able to hear our voice and the system could interupt words in the music as commands. For a voice recognition system the later point is not really too much of a problem, assuming our system recognises the drivers voice and they don't listen to recordings of themselves all too often.

While thinking about the problem I remembered quite a nice feature of the mac that might be a good idea for our interface. On the Mac, if you change volume not only do you get a visual display to illustrate the change.



You also get a sound that accompanies the increase in volume, akin to a beep getting louder as you increase the volume. Personally, I think this is quite a good idea because for a first time user it gives a good idea of the correct level they want to achieve instead of saying perhaps volume level 8. If combined with a visual display it allows the user to keep their attention on the road.

(P) Voice Recognition Testing on Prototype

As part of the user centered design process we have created a user evaluation plan which will evaluate a particular sub set of our prototype. The problem with testing our system is that although we have a prototype which explains the effect of invoking a particular feature, it does not allow us to test the voice recognition features, i.e. where drivers can verbally issue commands to the car. The aim of this evaluation was to find out:

1. Whether there are patterns in the structure in verbal commands given by users to solve given tasks
2. Whether there are certain tasks that users wish to perform more often than others
3. Whether interaction by voice recognition is preferred by users over tactile methods
4. Whether there are flaws in our testing methods that may make it unfair

The plan for the test is as follows:

1. Plan a set of scenarios and tasks that users may want to perform in the car
   
2. Show the prototype to the tester and give a general overview of our envisaged car
   
3. Have testers answer verbally what they would say in order to perform each task given the scenarios and record their responses
   

The results will give us an idea of the way humans want to interact with the system without having to know anything about what the system can understand. We want to understand the bigger picture rather than attempt to create the underlying system and understand the technical hurdles that must be overcome. Because of this assumption we do not need to present the results in front of the user; we can just assume that the right results come up. The important thing is that the results of this test should hopefully give us an idea of how people would naturally interact with a system in a way that is new to them; which is exactly what we're interested in.

These results will be discussed in subsequent posts.

Origami

Here's a picture of the thumb interface that ben is talking about.

(P) Task Analysis

Music Player

• Turn on player.
• Change volume.
• Track change
• Turn Off
  

Navigation System

• Setting of destination
• Change of destination
• Turn Off
  

Hazard Avoidance System

• Enable system
• Disable system
• Change notification level

Microsoft's Origami: New, unique and pointless?

Microsoft has released their new ‘Mini PC Project’ called Origami. It is a cross between a laptop and a pocket PC in an attempt to fill a gap in the market. They believe that people want the power of a PC which can be easily carried around, and to a certain point this is true. Looking at the product it doesn’t get my vote.

It is the size of a ‘hard-back’ book or 2 DVD cases and weighs 779g which is just over ¾kg. That is pretty heavy when you want to carry it about on a daily basis. It has a 900 MHz Celeron processor and a 40GB hard drive which is not bad considering its size. The price is a bit steep though, at £699 when it is released. It would be cheaper to buy a laptop which has more processing power, a bigger hard drive and a bigger screen.

In my personal opinion, it seems a bit of a gadget more than a useful every day item. If you want to do word processing on the move, then a laptop would be a better choice rather than a tiny 7” touch type screen PC. If portable gaming is your thing, then the PSP is a far better choice for games and price (only £170). The PSP also supports full feature films which can be purchased or downloaded onto the memory stick as well as playing music and viewing images. It also has wireless for gaming and full web browser capabilities. So.. it basically does everything the new Origami does except the word processing, and the PSP has been out for a couple of years now. If you want to do it all, then I would suggest a cheap laptop for £400 and a PSP off eBay for about £130.

Ok now the HCI stuff. The word processing is done using 2 ¼ circle type keypads at either side of the screen which are pressed using the thumbs. While it sounds like a great idea and does look good, it really is not very comfortable to use. How do I know you ask? I got a book about the size of the Origami and pressed using my thumbs for about 1 minute. After this time, it started to ache so how can you possibly type on it for long periods of time?

Ok what next… carrying it. How is this device going to be portable? If you have a bag for it, like that of a laptop bag, then it’s going to be too small to carry comfortably and look very odd(Imagine carrying a handbag). However, it’s too big to put in your pocket and if your pockets were huge, it’s too heavy.

Finally, the battery life. Bill gates claimed that the device would run an entire days work on one single charge. However, the released version will give about 3 hours battery life which is pitiful. What is the point in having a device that’s mobile and versatile if it can only be used for a few hours before you need a power point? I get more life out of my laptop battery and the PSP has a game time of about 6 hours.

I’m sure my comments are harsh and very critical but when releasing something so unique in the market, it needs to be right. If you release a new unique product that actually doesn’t do much more than current products, then it will not be as popular as first thought.

More information can be found on the BBC website and the Microsoft's UMPC Homepage.

Tactile File Browsing: Crap

I've just been reading articles around the web to do with future envisioning of user interfaces in operating systems as it's an area that I've caught an interest in recently. This is just a report on my findings and a few opinions that I have on a system that tries too hard without looking into real human factors.

People think that the desktop on an OS is 2D. To an extent, they are right, but overlapping windows is an inherent 3D feature. People sometimes call this 2.5D as the windows are essentially totally flat. An article I read explained why 3D should be used for organising things and argued that we are used to organising things in 3D, so that's how it should be done in the computer. The author suggested that because research showed that humans can remember where we put things, we should harness that power into the operating system to organise our files. At this point I started to think to myself - this guy is bonkers! The reason why tree-based navigation and organisation systems work so well is that we can easily conceptualise what's going on. There is just one idea to grasp - that each file and folder has at maximum one parent (aliases and symbolic links are sometimes used, but is not quite considered to be useful to the average user). Once we start trying to remember where we put files in a 3D space, we are essentially reinventing the problems of organisation we have in real life. How many times have you forgotten where you left your keys? Exactly...

I searched further and found that the author plays a part in the development of Tactile 3D (see screenshot), a standalone application for Windows that gives the user a 3D world which has objects representing their files and folders in. I'll let you find out a bit more yourself if you're interested in being appauled, but seriously, I felt like I was concentrating so hard to navigate the world. Immersive interfaces to do everyday tasks are bad. They need too much brain power to operate and therefore should be an interface technique kept to gaming and simulations only. How do the developers hope to increase productivity with this program?

Linked with my post about Spotlight and search technologies, I am starting to realise the usefulness of not having to worry about where the files are physically stored. On the other hand, I still think that the user should still be encouraged to organise their files somehow. I think the hierarchical system of organisation is nearing its death though. We need new ways of organising our files. But how else do we do it?

Office 12 and "Ribbon UI"

In an extention to the previous post, I'm going to talk a little bit about the usability issues that Microsoft found with Office 2003 and below, and how they are trying to fix it for the next version, Office 12. If you weren't at the previous HCI 2 lecture, basically, Russell opened up every single toolbar available in Microsoft Office 2003 (rather tediously using the menus I must say). He then said "how the hell am I supposed to recognise the icons I need when there are so many?". "usability guidelines say we should use recognition rather than recall, so maybe the designers thought icons are good. But there are so many". Basically, he identified the problem with Word having so many features and the user interface designers didn't think enough about how to make the features easier to find.

Has anyone seen the massive floor of Birmingham New Street train station which has a big yellow and white Office 12 teaser saying "We've changed, have you?" (that's probably wrong as it came from my memory). Well, they really are pushing something new with Office 12. They too have recognised the problems mentioned above, and are introducing a new user interface paradigm called "Ribbon UI" (pictured). They have gotten rid of all the traditional menus apart from File completely, and have given users access to features via panels headed by simpler, task-oriented words. The important thing about these panels is that because they are context aware, they make it easy for users to find features they want to use. For example, if I want to edit the look of a chart in Excel 12, I would click on the chart, and a design panel would be selectable. This panel is customised to displaying only design features that can be applied to the selected item (the chart).

Does this solve the problem? I've played with the Office 12 Beta 1 only briefly. Has anyone used it more and would like to share their comments?

More screenshots of Office 12.

Consumers Are Stupid

According to Reuters, 50% of product returns are due to customers not being able to operate it. This surely is one of the many studies that point towards the idea that "consumers are stupid". The article doesn't directly target user interfaces or even HCI to that matter, but we can somewhat generalise. For example, in today's lecture when Russell asked the class how to replace text that had a green squiggly line underneath and proceeded to follow instructions by a fellow student, ending up in a spell checking dialog, another student put his hand up and said "well it's obvious! Stop being stupid" (probably not his exact words but he said something like that). Well, can we assume that people find the same things obvious as you? No! We have to assume people are stupid. If everyone using Microsoft Office were clever, there would be much less need for usability people like our lecturer himself.

Article found via digg.com.

(P) Questionaire - Demographic

The resutls of our questionnaire seem to show that a large percentage of car drivers are male and between the ages of 16 and 25. I'm not sure how representative of the population these results are, but it could be hightlighting the subset that is interesting in our solution.