Interface Design V: Beyond the Desktop

Transcription

1 Interface Design V: Beyond the Desktop Rob Procter Further Reading Dix et al., chapter 4, p and chapter 15. Norman, The Invisible Computer, MIT Press, 1998, chapters 4 and /25/01 CS4: HCI lecture 13 1 Introduction: Interactional Styles and Paradigms Command Line Desktop Speech Virtual Reality Assistant and agent-based Interaction Ubiquitous Computing 11/25/01 CS4: HCI lecture

2 Desktop Metaphor The scalability problem Early GUIs worked: Making everything visible worked: not much to be visible. Learning by doing worked: not much to try out. But now: Desktop RAM has increased fold. Desktop disk space has increased fold. The entire Web is only a click away. It is not possible to make everything visible} All at once = chaos Not all at once = things get lost 11/25/01 CS4: HCI lecture 13 3 Desktop Metaphor Is the desktop metaphor broken? Perhaps the main problem is slavish adherence to the predominant metaphor There are alternatives: Some for standard PCs Lifestreams historical metaphor Adding another dimension Some for smaller devices Newton OS notepad metaphor 11/25/01 CS4: HCI lecture

3 Lifestreams 11/25/01 CS4: HCI lecture 13 5 Adding Another Dimension: Pad++ 11/25/01 CS4: HCI lecture

4 Norman s Diagnosis The difficulty goes deeper WIMP and DM try, and fail, to conceal the sheer complexity and scale of the system beneath them Why is the desktop PC so complex? One device, many tasks One device, many people Each demands the incremental addition of features So far, the solution has been to fit ever more complex help and support services 11/25/01 CS4: HCI lecture 13 7 Alternatives to the Desktop Speech interfaces Return to conversational interfaces Very natural Virtual reality Better state display --- only visible state? Better object manipulation --- data gloves Intelligent interface assistants and agents Conversational Autonomous and adaptive Ubiquitous computing Better object manipulation --- two hands Wearable computers Context sensitive, mixed reality 11/25/01 CS4: HCI lecture

5 Speech Interfaces Distinguish speech recognition from speech synthesis Speech recognition systems already work: Dictation, command and control, inventory Useful whenever user cannot type: No space for keyboard hands busy, eyes busy (e.g: operating, driving) low typing skills motor disability sensory disability Note naturalness of dialogue for feedback 11/25/01 CS4: HCI lecture 13 9 Problems with Speech Interfaces Speech recognition doesn t equate to robust user understanding For language understanding: shared assumptions, deep knowledge required For robust speech recognition: noise must be ignored via selective focus of attention Furthermore: for task success, talk is not enough Some tasks have single, known solutions -- but cannot be described in words: E.g.: tie your shoelace ' Other tasks are open-ended, and have no prior solution: E.g.: check the manuscript for mistakes ' Naturalness of interface is misleading: language understanding isn't mind-reading 11/25/01 CS4: HCI lecture

6 Agents Learning proceeds by: Observing and imitating user Receiving positive and negative feedback from user Receiving explicit instructions It is very difficult to infer user intentions accurately from a finite set of data Sometimes agents help, but are often wrong, which is worse than no help at all What happened to the paper clip? 11/25/01 CS4: HCI lecture Virtual Reality Interfaces Virtual reality, interface as immersion in a computed world, represents natural development of interface as shared object concept It is claimed that people are visual animals For dealing with information overload, more content can be packed onto the desktop by using 3D rendering For training and teleoperation purposes, accurate visual simulations of real or imagined situations can be: very useful cheaper than any alternative Interface to VR world can use vision, touch, speech Avatars -- virtual inhabitants 11/25/01 CS4: HCI lecture Seonaid 6

7 VR Example: A Museum 11/25/01 CS4: HCI lecture Problems with VR We do navigate well in space But not perfectly; much individual difference In desktop VR, the (picture of the) world moves; we don t: We experience movement through prediction and proprioception, as well as through vision In reality, vision and proprioception usually match Problems when: Lack of motion control; receptors non-functional; zero-g Desktop VR susceptible to this mismatch 11/25/01 CS4: HCI lecture

8 Ubiquitous Computing Anything that lets people escape their workstations: Cellphones, PDAs, smart clothing, big screens, handheld and wearable devices, wirelessly networked to the bigger infrastructure. Explores ways of embedding digital systems into real spaces and objects, making them literally tangible Automatic capture of everyday experience capture, integration, access Context-aware computing mostly physical location only Seamless services unobtrusive computing 11/25/01 CS4: HCI lecture Ubiquitous Computing Concepts 11/25/01 CS4: HCI lecture

9 Ubiquitous Computing: Tangible Bits 11/25/01 CS4: HCI lecture Ubiquitous Computing: Tangible Bits A positioning device 11/25/01 CS4: HCI lecture

10 Ubiquitous Computing: Tangible Bits A measuring device 11/25/01 CS4: HCI lecture Ubiquitous Computing: PDA with GPS 11/25/01 CS4: HCI lecture

11 Mixed Realities By tracking a person (& PDA) in a room, the person is also in a VR model of the same room. A map on the PDA shows VR avatars, and audio also links the two rooms. 11/25/01 CS4: HCI lecture Making Do With Small Interfaces The device shrinks; the interface shrinks Replace keyboard, pointer and screen with: a small, bad keyboard, pointer and screen speech recognition and synthesis pen-based input and handwriting or gesture recognition spectacle-mounted screens and speech input If the device itself becomes less complex, good: Compare the Newton high functionality discontinued With the Palm Pilot low functionality selling well 11/25/01 CS4: HCI lecture

12 Making Do With Small Interfaces Don Norman s solution: Information appliances are the true alternative Many, specialised systems are better than few general ones So long as they communicate with each other Devices that are easy to use, not only because they will be inherently simpler, but because they fit the task so well that to learn the task is to learn the appliance. 11/25/01 CS4: HCI lecture Guidelines for Small Devices Use(fulness) immediately apparent Structure interface to task Short cuts and flexibility Minimise memory load Use consistent screen templates Provide a Back function on every screen Selection is better than writing 11/25/01 CS4: HCI lecture

13 Some Examples 11/25/01 CS4: HCI lecture