[월:] 2018 3월

2018_ID_Sketches
Week1_Preliminary Sketches
https://pin.it/7xpkhdcjymfo2g

Steve Yohanan and Karon E. MacLean (2008) “The Haptic Creature Project: Social Human-Robot Interaction through Affective Touch”

The communication of emotion plays an important role in social interaction.Social human-robot interaction has focused almost exclusively on the modalities of vision and audition.This paper presents an overview of the Haptic Creature project, where we seek to develop a deeper understanding of affect display through touch in the context of social interaction between human and robot.

The paper aims to create the robotic creature that mimics a small animal. An important aspect of social interaction is affect display. Other than vision and audition, touch also plays a key role. The sense of touch is rather unique: the skin is the largest organ in the human body, the first sense organ to develop, and it plays a major role in early development.

There are many products which supports interaction between human and robotic animal, such as Paro, the Huggable and Aibo. However, they all focus strongly on visual and audio sector, and provides feedbacks within only limited interaction points. The big differentiating aspect of haptic creature project is the level of zoomorphism. It is consciously designed to have very minimalistic appearance. It is composed of five major components: a body, two ear-like appendages, a breathing mechanism, a purring mechanism, and a warming element.

There are three major design considerations for the Haptic Creature. The first is that the interaction centers around the modality of touch. Second is organic feedback, and third is zoomorphism.

The creature could be broken down into five major components: low-level sensing, gesture recognizer, emoter, physical renderer, and low-level actuation. The basic mechanism of the creature is that it senses touch across its entire body, and effectors will be manipulated via servos and motors. Low-level sensing senses information from the real world and collects raw data. Getsure recognizer then constructs an initial model of the physical data to provide a cohesive view. It distinguishes between a hard stroke and a soft pet. Emoter then analyzes the underlying emotional state. Physical renderer is in charge of the higher-order, which translates the results into an orchestrated manipulation of the effectors. Low-level actuation saves the normalized data appropriately for individual hardware devices working closely with the physical renderer.

De Choudhury, M., et. al. (2013) “Moon Phrases: a social media facilitated tool for emotional reflection and wellness

Emotional wellness has huge impact on our lives. However, emotional wellness is one the most important and overlooked parts of overall health. Unhealthy mental processes are often accompanied by negative self-image. But there is no reliable laboratory test for diagnosing most forms of mental illness. Typically, the diagnosis is based on the patient’s self-reported experiences. Therefore, the test result could be biased by the memory.
Moon Phrases is designed to tackle this problem. It tracks the emotion and linguistic expression of individuals based on online social activities, by quantifying his behavior through affective expression and language use within the posts. This approach is relatively accurate in that postings are often made in a naturalistic setting.

There are many supportive literature reviews regarding the connection between social network and individual’s behavioral health. Kawachi explored the role of social ties in the maintenance of psychological wellbeing and treatment of behavioral health concerns. Researchers in psycholinguistics also explored how expressions of language can be used to better understand human intentions, moods and disorders.

The Moon Phrases sought feedback from six fluent social media users. The key question was what are the best social media cues to be visualized, and it was patterns and levels of activities on social media. Also the visualization of either negative and positive linguistic use was critical, since certain forms of languages often entailed mood of the user.

Moon Phrases visualizes that information mainly with colors. The illuminated portion of the moon represent the degree of average of positivity. Orange and darker shades correspond to more than three negative posts. Other than simplified visualization with colors, Moon Phrases also provide more straight-forward information such as lists of total and unique positive and negative words used by the user and language style.

However, Moon Phrases could be controversial since it contains private information regarding health issue. Also, it is little questionable if social network could truly capture one’s behavior, since machine learning has its limitation over people’s idiosyncratic behavior behind the scenes.

Presence and Discernability in Conventional and Non-photorealistic Immersive Augmented Reality

Augmented Reality(AR) has its main purpose to integrate virtual and real imagery, making them visually indistinguishable. Normally, AR has appearance acquisition approach which aims to enhance the photorealism of the virtual objects. However, real-time rendering is time consuming, unaffordable and limited. The alternative could be making the real world seem like virtual world, by using several skills like cartoon-style rendering, painting-look-like rendering, exaggerated color filter, edge detection and so on. The latter approach is called Non-Photorealistic Rendering(NPR).

Traditional photorealistic approach suitably works only in ideal conditions such as having pre-rendered images and videos all set on standard desktop displays. However, when it comes to the head-mounted immersive AR, the issue of discernibility becomes critical in forming a convincing experience. Here, discernibility refers to a user’s ability to correctly distinguish the real world from the virtual one. Not only the discernibility being the huge hurdle, latency between video and tracking data, tracking failures and optical distortions may also distract users from presence within the AR environment. Here, presence means perceptual illusion of non-mediation.

Thus, Steptoe supports NPR approach to immersive AR system, within a low-cost head-mounted video see-through immersive AR system, while the AR-Rift supported by an Oculus Rift Development Kit with an optical motion capture system. The AR-Rift is an immersive head-mounted video see-through AR display which comprises of low-cost, affordable components though it has pretty good specification to support immersive AR experience. The latency of camera images and tracking data are 100ms and 69ms respectively. This brings visual differences which results in objects of real world to be seen as that of the virtual world.

Steptoe introduces three different NPR approaches of conventional, stylized, and virtualized. The conventional mode shows unaltered video feeds and uses standard real-time graphics and algorithms. The stylized mode applies an edge-detection filter to silhouette edges of objects within the full image frame, which includes both video and graphics. The virtualized mode presents an extreme stylization by both silhouetting edges and removing color information. The three models represent different degree of photo-realism. To elaborate the rendering condition, the conventional mode used unmodified video and standard graphical rendering techniques. The stylized mode uses a Sobel edge filter to highlight horizontal and vertical edges. The virtualized mode extends the stylized shades to de-saturate the color of the image using Cg’s saturate function.

The Reactable : Tangible and Teletop Music Performance

Jorda introduces the Reactable, electronic musical instrument with a simple and intuitive interface which turns music into a tangible and visual experience. The Reactable provides fun user experience by allowing people to experiment with sound, change the structure of music freely, which results in making the music very creatively, in an unseen way.

The Reactable is not limited to single user, it aims to give users a shared performance. Users can shift control to the instrument. Visual feedback works as a significant asset for allowing this type of instruments to dynamically communicate the states and the behaviors of their musical processes. This concept could be retrieved to the concept of Tangible User Interfaces (TUIs) combined with a physical artifact. In traditional instrumental playing, every nuance, every small control has to to be addressed physically by the performer. Performing music with computers often tends towards an interactive dialogue between instrument and instrumentalist. The Reactable is built upon a tabletop interface and manipulated by the tangible acrylic pucks on its surface. When any puck is placed on the Reactable’s surface, it illuminates and starts to interact with the other neighboring pucks. These interactions are visible on the surface of table top providing the instant feedback.

Affordance, Conventions, and Design. Donald Norman

Norman suggests three major dimensions of successful design: conceptual models, constraints, and affordances. Norman refers to Gibson’s definition on affordance when explaining basic concept of affordance. Affordances are naturally given relationships. It doesn’t have to be visible nor notable. Then, he elaborates on affordances by giving on example of how people figure out usages of first-seen technologies; Appearance of the device could provide the critical clues required for its proper operation.

Norman brings up a misuse of the term affordance by bringing up the idea of perceived and real affordance. The perceived affordance might not always lead to the right action. For example, if the screen affords touching even thought that touching has no effect on the computer system, this could be considered as a perceived affordance. While real affordance always guides to the desirable action. For example, clickable button doesn’t create affordance of click since you can basically click anywhere else in the screen even with the button. Whereas if you make cursor appear only at spots are to be clicked, this could be considered as the real affordance.

In addition, Norman covers the difference between the affordance and convention. Some designers argue affordance could be made by adding graphical object in the system or technology. However, Norman says affordance cannot be artificially made. Affordance exists independently of what is visible. Rather, those cases of adding graphical objects could be considered as conventions. For instance, change in cursor shape could be explained as a learned convention. It is not that affordance naturally follows, but rather users follow the cultural constraints, the arbitrarily decided action. The changed shape of cursor could be perfectly substituted by any other shape than hand, only if the social consensus is made. This is, in other words, what we call as symbolic communication.

In this article, Norman also covers three behavioral constraints: physical, logical, and cultural. Physical constraints are closely related to real affordances such as not being able to move the cursor outside the screen or locking the mouse button when something is not clickable. Logical constraints use reasoning to determine the alternatives. This is used how the users know to scroll down or they have finished a task. Cultural constraints are conventions shared within society like having the scroll bar at the right side of display. The selection is made arbitrarily, differentiating itself from affordances.

Technology Affordances, Gaver

Gaver thinks affordance provides link between perception and action. To elaborate, it focuses on the interaction between technologies and the people expected to use them. Thus, affordance does fundamental role in perception. However, the concept of affordance varies in different perspectives. Gaver gives Norman’s example of doorhandle to illustrate basic concept of affordance. ; Thin vertical doorhandles afford pulling, while flat horizontal plates afford pushing. This kind of action occurs due to human motor system. However, even if the doorhandle affords pulling, it doesn’t necessarily mean the door is unlocked. This example of doorhandle shows several important aspects of affordances. Affordances are independent of perception. They exist whether perceived or not and whether there is perceptual information provided or not. From the example of the doorhandle, if door is open –which means affordance actually exists—we call it perceptible affordance. On the other hand, if the doorhandle is locked, it means it affords non-existing affordance and we call it false affordance. There is also a case called hidden affordance, which indicates there is real existing affordance but no perceptible hint exists. From this point of view, we may tell good design is to hint users with perceptible affordances.

Gaver also talks about the effect of cultural factor on affordance. Gaver consider culture and experience as helping rather than key factor of perception of affordances. To elaborate on this statement, Gaver brings up the example of button design.; The button appears to be raised from the background, to have depth. This button supposed to support two functions, clickable and user-tailorable. However, the depth of button only affords pressing. While some people interpret this miscommunication as absence of tailoring culture, Gaver explain this as limitation the design has.; Button with depth do not afford much except pressing.

For more complicated technologies, affordance works as a sequence. Here, not every affordance might be perceptible through passive observation. Instead once first step has carried out, perceptible clues would follow thereafter. In short, one affordance leads to another. Perceptible affordances might not be limited to visual cues but vary various media such as sound.