computing
-
KAIST Joins IBM Q Network to Accelerate Quantum Computing Research and Foster Quantum Industry
KAIST has joined the IBM Q Network, a community of Fortune 500 companies, academic institutions, startups, and research labs working with IBM to advance quantum computing for business and science.
As the IBM Q Network’s first academic partner in Korea, KAIST will use IBM's advanced quantum computing systems to carry out research projects that advance quantum information science and explore early applications. KAIST will also utilize IBM Quantum resources for talent training and education in preparation for building a quantum workforce for the quantum computing era that will bring huge changes to science and business. By joining the network, KAIST will take a leading role in fostering the ecosystem of quantum computing in Korea, which is expected to be a necessary enabler to realize the Fourth Industrial Revolution.
Professor June-Koo Rhee who also serves as Director of the KAIST Information Technology Research Center (ITRC) of Quantum Computing for AI has led the agreement on KAIST’s joining the IBM Q Network. Director Rhee described quantum computing as "a new technology that can calculate mathematical challenges at very high speed and low power” and also as “one that will change the future.”
Director Rhee said, “Korea started investment in quantum computing relatively late, and thus requires to take bold steps with innovative R&D strategies to pave the roadmap for the next technological leap in the field”. With KAIST joining the IBM Q Network, “Korea will be better equipped to establish a quantum industry, an important foundation for securing national competitiveness,” he added.
The KAIST ITRC of Quantum Computing for AI has been using the publicly available IBM Quantum Experience delivered over the IBM Cloud for research, development and training of quantum algorithms such as quantum artificial intelligence, quantum chemical calculation, and quantum computing education.
KAIST will have access to the most advanced IBM Quantum systems to explore practical research and experiments such as diagnosis of diseases based on quantum artificial intelligence, quantum computational chemistry, and quantum machine learning technology. In addition, knowledge exchanges and sharing with overseas universities and companies under the IBM Q Network will help KAIST strengthen the global presence of Korean technology in quantum computing.
About IBM Quantum
IBM Quantum is an industry-first initiative to build quantum systems for business and science applications. For more information about IBM's quantum computing efforts, please visit www.ibm.com/ibmq.
For more information about the IBM Q Network, as well as a full list of all partners, members, and hubs, visit https://www.research.ibm.com/ibm-q/network/
©Thumbnail Image: IBM.
(END)
2020.09.29 View 7688 -
Quantum Classifiers with Tailored Quantum Kernel
Quantum information scientists have introduced a new method for machine learning classifications in quantum computing. The non-linear quantum kernels in a quantum binary classifier provide new insights for improving the accuracy of quantum machine learning, deemed able to outperform the current AI technology.
The research team led by Professor June-Koo Kevin Rhee from the School of Electrical Engineering, proposed a quantum classifier based on quantum state fidelity by using a different initial state and replacing the Hadamard classification with a swap test. Unlike the conventional approach, this method is expected to significantly enhance the classification tasks when the training dataset is small, by exploiting the quantum advantage in finding non-linear features in a large feature space.
Quantum machine learning holds promise as one of the imperative applications for quantum computing. In machine learning, one fundamental problem for a wide range of applications is classification, a task needed for recognizing patterns in labeled training data in order to assign a label to new, previously unseen data; and the kernel method has been an invaluable classification tool for identifying non-linear relationships in complex data.
More recently, the kernel method has been introduced in quantum machine learning with great success. The ability of quantum computers to efficiently access and manipulate data in the quantum feature space can open opportunities for quantum techniques to enhance various existing machine learning methods.
The idea of the classification algorithm with a nonlinear kernel is that given a quantum test state, the protocol calculates the weighted power sum of the fidelities of quantum data in quantum parallel via a swap-test circuit followed by two single-qubit measurements (see Figure 1). This requires only a small number of quantum data operations regardless of the size of data. The novelty of this approach lies in the fact that labeled training data can be densely packed into a quantum state and then compared to the test data.
The KAIST team, in collaboration with researchers from the University of KwaZulu-Natal (UKZN) in South Africa and Data Cybernetics in Germany, has further advanced the rapidly evolving field of quantum machine learning by introducing quantum classifiers with tailored quantum kernels.This study was reported at npj Quantum Information in May.
The input data is either represented by classical data via a quantum feature map or intrinsic quantum data, and the classification is based on the kernel function that measures the closeness of the test data to training data.
Dr. Daniel Park at KAIST, one of the lead authors of this research, said that the quantum kernel can be tailored systematically to an arbitrary power sum, which makes it an excellent candidate for real-world applications.
Professor Rhee said that quantum forking, a technique that was invented by the team previously, makes it possible to start the protocol from scratch, even when all the labeled training data and the test data are independently encoded in separate qubits.
Professor Francesco Petruccione from UKZN explained, “The state fidelity of two quantum states includes the imaginary parts of the probability amplitudes, which enables use of the full quantum feature space.”
To demonstrate the usefulness of the classification protocol, Carsten Blank from Data Cybernetics implemented the classifier and compared classical simulations using the five-qubit IBM quantum computer that is freely available to public users via cloud service. “This is a promising sign that the field is progressing,” Blank noted.
Link to download the full-text paper:
https://www.nature.com/articles/s41534-020-0272-6
-Profile
Professor June-Koo Kevin Rhee
rhee.jk@kaist.ac.kr
Professor, School of Electrical Engineering
Director, ITRC of Quantum Computing for AIKAIST
Daniel Kyungdeock Parkkpark10@kaist.ac.krResearch Assistant ProfessorSchool of Electrical EngineeringKAIST
2020.07.07 View 10417 -
A Deep-Learned E-Skin Decodes Complex Human Motion
A deep-learning powered single-strained electronic skin sensor can capture human motion from a distance. The single strain sensor placed on the wrist decodes complex five-finger motions in real time with a virtual 3D hand that mirrors the original motions. The deep neural network boosted by rapid situation learning (RSL) ensures stable operation regardless of its position on the surface of the skin.
Conventional approaches require many sensor networks that cover the entire curvilinear surfaces of the target area. Unlike conventional wafer-based fabrication, this laser fabrication provides a new sensing paradigm for motion tracking.
The research team, led by Professor Sungho Jo from the School of Computing, collaborated with Professor Seunghwan Ko from Seoul National University to design this new measuring system that extracts signals corresponding to multiple finger motions by generating cracks in metal nanoparticle films using laser technology. The sensor patch was then attached to a user’s wrist to detect the movement of the fingers.
The concept of this research started from the idea that pinpointing a single area would be more efficient for identifying movements than affixing sensors to every joint and muscle. To make this targeting strategy work, it needs to accurately capture the signals from different areas at the point where they all converge, and then decoupling the information entangled in the converged signals. To maximize users’ usability and mobility, the research team used a single-channeled sensor to generate the signals corresponding to complex hand motions.
The rapid situation learning (RSL) system collects data from arbitrary parts on the wrist and automatically trains the model in a real-time demonstration with a virtual 3D hand that mirrors the original motions. To enhance the sensitivity of the sensor, researchers used laser-induced nanoscale cracking.
This sensory system can track the motion of the entire body with a small sensory network and facilitate the indirect remote measurement of human motions, which is applicable for wearable VR/AR systems.
The research team said they focused on two tasks while developing the sensor. First, they analyzed the sensor signal patterns into a latent space encapsulating temporal sensor behavior and then they mapped the latent vectors to finger motion metric spaces.
Professor Jo said, “Our system is expandable to other body parts. We already confirmed that the sensor is also capable of extracting gait motions from a pelvis. This technology is expected to provide a turning point in health-monitoring, motion tracking, and soft robotics.”
This study was featured in Nature Communications.
Publication:
Kim, K. K., et al. (2020) A deep-learned skin sensor decoding the epicentral human motions. Nature Communications. 11. 2149. https://doi.org/10.1038/s41467-020-16040-y29
Link to download the full-text paper:
https://www.nature.com/articles/s41467-020-16040-y.pdf
Profile: Professor Sungho Jo
shjo@kaist.ac.kr
http://nmail.kaist.ac.kr
Neuro-Machine Augmented Intelligence Lab
School of Computing
College of Engineering
KAIST
2020.06.10 View 10834 -
Professor Dongsu Han Named Program Chair for ACM CoNEXT 2020
Professor Dongsu Han from the School of Electrical Engineering has been appointed as the program chair for the 16th Association for Computing Machinery’s International Conference on emerging Networking EXperiments and Technologies (ACM CoNEXT 2020). Professor Han is the first program chair to be appointed from an Asian institution.
ACM CoNEXT is hosted by ACM SIGCOMM, ACM's Special Interest Group on Data Communications, which specializes in the field of communication and computer networks.
Professor Han will serve as program co-chair along with Professor Anja Feldmann from the Max Planck Institute for Informatics. Together, they have appointed 40 world-leading researchers as program committee members for this conference, including Professor Song Min Kim from KAIST School of Electrical Engineering.
Paper submissions for the conference can be made by the end of June, and the event itself is to take place from the 1st to 4th of December.
Conference Website: https://conferences2.sigcomm.org/co-next/2020/#!/home
(END)
2020.06.02 View 8436 -
AI to Determine When to Intervene with Your Driving
(Professor Uichin Lee (left) and PhD candidate Auk Kim)
Can your AI agent judge when to talk to you while you are driving? According to a KAIST research team, their in-vehicle conservation service technology will judge when it is appropriate to contact you to ensure your safety.
Professor Uichin Lee from the Department of Industrial and Systems Engineering at KAIST and his research team have developed AI technology that automatically detects safe moments for AI agents to provide conversation services to drivers.
Their research focuses on solving the potential problems of distraction created by in-vehicle conversation services. If an AI agent talks to a driver at an inopportune moment, such as while making a turn, a car accident will be more likely to occur.
In-vehicle conversation services need to be convenient as well as safe. However, the cognitive burden of multitasking negatively influences the quality of the service. Users tend to be more distracted during certain traffic conditions. To address this long-standing challenge of the in-vehicle conversation services, the team introduced a composite cognitive model that considers both safe driving and auditory-verbal service performance and used a machine-learning model for all collected data.
The combination of these individual measures is able to determine the appropriate moments for conversation and most appropriate types of conversational services. For instance, in the case of delivering simple-context information, such as a weather forecast, driver safety alone would be the most appropriate consideration. Meanwhile, when delivering information that requires a driver response, such as a “Yes” or “No,” the combination of driver safety and auditory-verbal performance should be considered.
The research team developed a prototype of an in-vehicle conversation service based on a navigation app that can be used in real driving environments. The app was also connected to the vehicle to collect in-vehicle OBD-II/CAN data, such as the steering wheel angle and brake pedal position, and mobility and environmental data such as the distance between successive cars and traffic flow.
Using pseudo-conversation services, the research team collected a real-world driving dataset consisting of 1,388 interactions and sensor data from 29 drivers who interacted with AI conversational agents. Machine learning analysis based on the dataset demonstrated that the opportune moments for driver interruption could be correctly inferred with 87% accuracy.
The safety enhancement technology developed by the team is expected to minimize driver distractions caused by in-vehicle conversation services. This technology can be directly applied to current in-vehicle systems that provide conversation services. It can also be extended and applied to the real-time detection of driver distraction problems caused by the use of a smartphone while driving.
Professor Lee said, “In the near future, cars will proactively deliver various in-vehicle conversation services. This technology will certainly help vehicles interact with their drivers safely as it can fairly accurately determine when to provide conversation services using only basic sensor data generated by cars.”
The researchers presented their findings at the ACM International Joint Conference on Pervasive and Ubiquitous Computing (Ubicomp’19) in London, UK. This research was supported in part by Hyundai NGV and by the Next-Generation Information Computing Development Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT.
(Figure: Visual description of safe enhancement technology for in-vehicle conversation services)
2019.11.13 View 14822 -
Image Analysis to Automatically Quantify Gender Bias in Movies
Many commercial films worldwide continue to express womanhood in a stereotypical manner, a recent study using image analysis showed. A KAIST research team developed a novel image analysis method for automatically quantifying the degree of gender bias in films.
The ‘Bechdel Test’ has been the most representative and general method of evaluating gender bias in films. This test indicates the degree of gender bias in a film by measuring how active the presence of women is in a film. A film passes the Bechdel Test if the film (1) has at least two female characters, (2) who talk to each other, and (3) their conversation is not related to the male characters.
However, the Bechdel Test has fundamental limitations regarding the accuracy and practicality of the evaluation. Firstly, the Bechdel Test requires considerable human resources, as it is performed subjectively by a person. More importantly, the Bechdel Test analyzes only a single aspect of the film, the dialogues between characters in the script, and provides only a dichotomous result of passing the test, neglecting the fact that a film is a visual art form reflecting multi-layered and complicated gender bias phenomena. It is also difficult to fully represent today’s various discourse on gender bias, which is much more diverse than in 1985 when the Bechdel Test was first presented.
Inspired by these limitations, a KAIST research team led by Professor Byungjoo Lee from the Graduate School of Culture Technology proposed an advanced system that uses computer vision technology to automatically analyzes the visual information of each frame of the film. This allows the system to more accurately and practically evaluate the degree to which female and male characters are discriminatingly depicted in a film in quantitative terms, and further enables the revealing of gender bias that conventional analysis methods could not yet detect.
Professor Lee and his researchers Ji Yoon Jang and Sangyoon Lee analyzed 40 films from Hollywood and South Korea released between 2017 and 2018. They downsampled the films from 24 to 3 frames per second, and used Microsoft’s Face API facial recognition technology and object detection technology YOLO9000 to verify the details of the characters and their surrounding objects in the scenes.
Using the new system, the team computed eight quantitative indices that describe the representation of a particular gender in the films. They are: emotional diversity, spatial staticity, spatial occupancy, temporal occupancy, mean age, intellectual image, emphasis on appearance, and type and frequency of surrounding objects.
Figure 1. System Diagram
Figure 2. 40 Hollywood and Korean Films Analyzed in the Study
According to the emotional diversity index, the depicted women were found to be more prone to expressing passive emotions, such as sadness, fear, and surprise. In contrast, male characters in the same films were more likely to demonstrate active emotions, such as anger and hatred.
Figure 3. Difference in Emotional Diversity between Female and Male Characters
The type and frequency of surrounding objects index revealed that female characters and automobiles were tracked together only 55.7 % as much as that of male characters, while they were more likely to appear with furniture and in a household, with 123.9% probability.
In cases of temporal occupancy and mean age, female characters appeared less frequently in films than males at the rate of 56%, and were on average younger in 79.1% of the cases. These two indices were especially conspicuous in Korean films.
Professor Lee said, “Our research confirmed that many commercial films depict women from a stereotypical perspective. I hope this result promotes public awareness of the importance of taking prudence when filmmakers create characters in films.”
This study was supported by KAIST College of Liberal Arts and Convergence Science as part of the Venture Research Program for Master’s and PhD Students, and will be presented at the 22nd ACM Conference on Computer-Supported Cooperative Work and Social Computing (CSCW) on November 11 to be held in Austin, Texas.
Publication:
Ji Yoon Jang, Sangyoon Lee, and Byungjoo Lee. 2019. Quantification of Gender Representation Bias in Commercial Films based on Image Analysis. In Proceedings of the 22nd ACM Conference on Computer-Supported Cooperative Work and Social Computing (CSCW). ACM, New York, NY, USA, Article 198, 29 pages. https://doi.org/10.1145/3359300
Link to download the full-text paper:
https://files.cargocollective.com/611692/cscw198-jangA--1-.pdf
Profile: Prof. Byungjoo Lee, MD, PhD
byungjoo.lee@kaist.ac.kr
http://kiml.org/
Assistant Professor
Graduate School of Culture Technology (CT)
Korea Advanced Institute of Science and Technology (KAIST)
https://www.kaist.ac.kr Daejeon 34141, Korea
Profile: Ji Yoon Jang, M.S.
yoone3422@kaist.ac.kr
Interactive Media Lab
Graduate School of Culture Technology (CT)
Korea Advanced Institute of Science and Technology (KAIST)
https://www.kaist.ac.kr Daejeon 34141, Korea
Profile: Sangyoon Lee, M.S. Candidate
sl2820@kaist.ac.kr
Interactive Media Lab
Graduate School of Culture Technology (CT)
Korea Advanced Institute of Science and Technology (KAIST)
https://www.kaist.ac.kr Daejeon 34141, Korea
(END)
2019.10.17 View 22696 -
Object Identification and Interaction with a Smartphone Knock
(Professor Lee (far right) demonstrate 'Knocker' with his students.)
A KAIST team has featured a new technology, “Knocker”, which identifies objects and executes actions just by knocking on it with the smartphone. Software powered by machine learning of sounds, vibrations, and other reactions will perform the users’ directions.
What separates Knocker from existing technology is the sensor fusion of sound and motion. Previously, object identification used either computer vision technology with cameras or hardware such as RFID (Radio Frequency Identification) tags. These solutions all have their limitations. For computer vision technology, users need to take pictures of every item. Even worse, the technology will not work well in poor lighting situations. Using hardware leads to additional costs and labor burdens.
Knocker, on the other hand, can identify objects even in dark environments only with a smartphone, without requiring any specialized hardware or using a camera. Knocker utilizes the smartphone’s built-in sensors such as a microphone, an accelerometer, and a gyroscope to capture a unique set of responses generated when a smartphone is knocked against an object. Machine learning is used to analyze these responses and classify and identify objects.
The research team under Professor Sung-Ju Lee from the School of Computing confirmed the applicability of Knocker technology using 23 everyday objects such as books, laptop computers, water bottles, and bicycles. In noisy environments such as a busy café or on the side of a road, it achieved 83% identification accuracy. In a quiet indoor environment, the accuracy rose to 98%.
The team believes Knocker will open a new paradigm of object interaction. For instance, by knocking on an empty water bottle, a smartphone can automatically order new water bottles from a merchant app. When integrated with IoT devices, knocking on a bed’s headboard before going to sleep could turn off the lights and set an alarm. The team suggested and implemented 15 application cases in the paper, presented during the 2019 ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp 2019) held in London last month.
Professor Sung-Ju Lee said, “This new technology does not require any specialized sensor or hardware. It simply uses the built-in sensors on smartphones and takes advantage of the power of machine learning. It’s a software solution that everyday smartphone users could immediately benefit from.” He continued, “This technology enables users to conveniently interact with their favorite objects.”
The research was supported in part by the Next-Generation Information Computing Development Program through the National Research Foundation of Korea funded by the Ministry of Science and ICT and an Institute for Information & Communications Technology Promotion (IITP) grant funded by the Ministry of Science and ICT.
Figure: An example knock on a bottle. Knocker identifies the object by analyzing a unique set of responses from the knock, and automatically launches a proper application or service.
2019.10.02 View 25379 -
Flexible User Interface Distribution for Ubiquitous Multi-Device Interaction
< Research Group of Professor Insik Shin (center) >
KAIST researchers have developed mobile software platform technology that allows a mobile application (app) to be executed simultaneously and more dynamically on multiple smart devices. Its high flexibility and broad applicability can help accelerate a shift from the current single-device paradigm to a multiple one, which enables users to utilize mobile apps in ways previously unthinkable.
Recent trends in mobile and IoT technologies in this era of 5G high-speed wireless communication have been hallmarked by the emergence of new display hardware and smart devices such as dual screens, foldable screens, smart watches, smart TVs, and smart cars. However, the current mobile app ecosystem is still confined to the conventional single-device paradigm in which users can employ only one screen on one device at a time. Due to this limitation, the real potential of multi-device environments has not been fully explored.
A KAIST research team led by Professor Insik Shin from the School of Computing, in collaboration with Professor Steve Ko’s group from the State University of New York at Buffalo, has developed mobile software platform technology named FLUID that can flexibly distribute the user interfaces (UIs) of an app to a number of other devices in real time without needing any modifications. The proposed technology provides single-device virtualization, and ensures that the interactions between the distributed UI elements across multiple devices remain intact.
This flexible multimodal interaction can be realized in diverse ubiquitous user experiences (UX), such as using live video steaming and chatting apps including YouTube, LiveMe, and AfreecaTV. FLUID can ensure that the video is not obscured by the chat window by distributing and displaying them separately on different devices respectively, which lets users enjoy the chat function while watching the video at the same time.
In addition, the UI for the destination input on a navigation app can be migrated into the passenger’s device with the help of FLUID, so that the destination can be easily and safely entered by the passenger while the driver is at the wheel.
FLUID can also support 5G multi-view apps – the latest service that allows sports or games to be viewed from various angles on a single device. With FLUID, the user can watch the event simultaneously from different viewpoints on multiple devices without switching between viewpoints on a single screen.
PhD candidate Sangeun Oh, who is the first author, and his team implemented the prototype of FLUID on the leading open-source mobile operating system, Android, and confirmed that it can successfully deliver the new UX to 20 existing legacy apps.
“This new technology can be applied to next-generation products from South Korean companies such as LG’s dual screen phone and Samsung’s foldable phone and is expected to embolden their competitiveness by giving them a head-start in the global market.” said Professor Shin.
This study will be presented at the 25th Annual International Conference on Mobile Computing and Networking (ACM MobiCom 2019) October 21 through 25 in Los Cabos, Mexico. The research was supported by the National Science Foundation (NSF) (CNS-1350883 (CAREER) and CNS-1618531).
Figure 1. Live video streaming and chatting app scenario
Figure 2. Navigation app scenario
Figure 3. 5G multi-view app scenario
Publication: Sangeun Oh, Ahyeon Kim, Sunjae Lee, Kilho Lee, Dae R. Jeong, Steven Y. Ko, and Insik Shin. 2019. FLUID: Flexible User Interface Distribution for Ubiquitous Multi-device Interaction. To be published in Proceedings of the 25th Annual International Conference on Mobile Computing and Networking (ACM MobiCom 2019). ACM, New York, NY, USA. Article Number and DOI Name TBD.
Video Material:
https://youtu.be/lGO4GwH4enA
Profile: Prof. Insik Shin, MS, PhD
ishin@kaist.ac.kr
https://cps.kaist.ac.kr/~ishin
Professor
Cyber-Physical Systems (CPS) Lab
School of Computing
Korea Advanced Institute of Science and Technology (KAIST)
http://kaist.ac.kr Daejeon 34141, Korea
Profile: Sangeun Oh, PhD Candidate
ohsang1213@kaist.ac.kr
https://cps.kaist.ac.kr/
PhD Candidate
Cyber-Physical Systems (CPS) Lab
School of Computing
Korea Advanced Institute of Science and Technology (KAIST)
http://kaist.ac.kr Daejeon 34141, Korea
Profile: Prof. Steve Ko, PhD
stevko@buffalo.edu
https://nsr.cse.buffalo.edu/?page_id=272
Associate Professor
Networked Systems Research Group
Department of Computer Science and Engineering
State University of New York at Buffalo
http://www.buffalo.edu/ Buffalo 14260, USA
(END)
2019.07.20 View 37677 -
Play Games With No Latency
One of the most challenging issues for game players looks to be resolved soon with the introduction of a zero-latency gaming environment. A KAIST team developed a technology that helps game players maintain zero-latency performance. The new technology transforms the shapes of game design according to the amount of latency.
Latency in human-computer interactions is often caused by various factors related to the environment and performance of the devices, networks, and data processing. The term ‘lag’ is used to refer to any latency during gaming which impacts the user’s performance.
Professor Byungjoo Lee at the Graduate School of Culture Technology in collaboration with Aalto University in Finland presented a mathematical model for predicting players' behavior by understanding the effects of latency on players. This cognitive model is capable of predicting the success rate of a user when there is latency in a 'moving target selection' task which requires button input in a time constrained situation.
The model predicts the players’ task success rate when latency is added to the gaming environment. Using these predicted success rates, the design elements of the game are geometrically modified to help players maintain similar success rates as they would achieve in a zero-latency environment. In fact, this research succeeded in modifying the pillar heights of the Flappy Bird game, allowing the players to maintain their gaming performance regardless of the added latency.
Professor Lee said, "This technique is unique in the sense that it does not interfere with a player's gaming flow, unlike traditional methods which manipulate the game clock by the amount of latency. This study can be extended to various games such as reducing the size of obstacles in the latent computing environment.”
This research, in collaboration with Dr. Sunjun Kim from Aalto University and led by PhD candidate Injung Lee, was presented during the 2019 CHI Conference on Human Factors in Computing Systems last month in Glasgow in the UK.
This research was supported by the National Research Foundation of Korea (NRF) (2017R1C1B2002101, 2018R1A5A7025409), and the Aalto University Seed Funding Granted to the GamerLab respectively.
Figure 1. Overview of Geometric Compensation
Publication:
Injung Lee, Sunjun Kim, and Byungjoo Lee. 2019. Geometrically Compensating Effect of End-to-End Latency in Moving-Target Selection Games. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems (CHI’19) . ACM, New York, NY, USA, Article 560, 12 pages. https://doi.org/10.1145/3290605.3300790
Video Material:
https://youtu.be/TTi7dipAKJs
Profile: Prof. Byungjoo Lee, MD, PhD
byungjoo.lee@kaist.ac.kr
http://kiml.org/
Assistant Professor
Graduate School of Culture Technology (CT)
Korea Advanced Institute of Science and Technology (KAIST)
http://kaist.ac.kr Daejeon 34141, Korea
Profile: Injung Lee, PhD Candidate
edndn@kaist.ac.kr
PhD Candidate
Interactive Media Lab
Graduate School of Culture Technology (CT)
Korea Advanced Institute of Science and Technology (KAIST)
http://kaist.ac.kr
Daejeon 34141, Korea
Profile: Postdoc. Sunjun Kim, MD, PhD
kuaa.net@gmail.com
Postdoctoral Researcher
User Interfaces Group
Aalto University
https://www.aalto.fi Espoo 02150, Finland
(END)
2019.06.11 View 44818 -
Sound-based Touch Input Technology for Smart Tables and Mirrors
(from left: MS candidate Anish Byanjankar, Research Assistant Professor Hyosu Kim and Professor Insik Shin)
Time passes so quickly, especially in the morning. Your hands are so busy brushing your teeth and checking the weather on your smartphone. You might wish that your mirror could turn into a touch screen and free up your hands. That wish can be achieved very soon. A KAIST team has developed a smartphone-based touch sound localization technology to facilitate ubiquitous interactions, turning objects like furniture and mirrors into touch input tools.
This technology analyzes touch sounds generated from a user’s touch on a surface and identifies the location of the touch input. For instance, users can turn surrounding tables or walls into virtual keyboards and write lengthy e-mails much more conveniently by using only the built-in microphone on their smartphones or tablets. Moreover, family members can enjoy a virtual chessboard or enjoy board games on their dining tables.
Additionally, traditional smart devices such as smart TVs or mirrors, which only provide simple screen display functions, can play a smarter role by adding touch input function support (see the image below).
Figure 1.Examples of using touch input technology: By using only smartphone, you can use surrounding objects as a touch screen anytime and anywhere.
The most important aspect of enabling the sound-based touch input method is to identify the location of touch inputs in a precise manner (within about 1cm error). However, it is challenging to meet these requirements, mainly because this technology can be used in diverse and dynamically changing environments. Users may use objects like desks, walls, or mirrors as touch input tools and the surrounding environments (e.g. location of nearby objects or ambient noise level) can be varied. These environmental changes can affect the characteristics of touch sounds.
To address this challenge, Professor Insik Shin from the School of Computing and his team focused on analyzing the fundamental properties of touch sounds, especially how they are transmitted through solid surfaces.
On solid surfaces, sound experiences a dispersion phenomenon that makes different frequency components travel at different speeds. Based on this phenomenon, the team observed that the arrival time difference (TDoA) between frequency components increases in proportion to the sound transmission distance, and this linear relationship is not affected by the variations of surround environments.
Based on these observations, Research Assistant Professor Hyosu Kim proposed a novel sound-based touch input technology that records touch sounds transmitted through solid surfaces, then conducts a simple calibration process to identify the relationship between TDoA and the sound transmission distance, finally achieving accurate touch input localization.
The accuracy of the proposed system was then measured. The average localization error was lower than about 0.4 cm on a 17-inch touch screen. Particularly, it provided a measurement error of less than 1cm, even with a variety of objects such as wooden desks, glass mirrors, and acrylic boards and when the position of nearby objects and noise levels changed dynamically. Experiments with practical users have also shown positive responses to all measurement factors, including user experience and accuracy.
Professor Shin said, “This is novel touch interface technology that allows a touch input system just by installing three to four microphones, so it can easily turn nearby objects into touch screens.”
The proposed system was presented at ACM SenSys, a top-tier conference in the field of mobile computing and sensing, and was selected as a best paper runner-up in November 2018.
(The demonstration video of the sound-based touch input technology)
2018.12.26 View 7615 -
It's Time to 3D Sketch with Air Scaffolding
People often use their hands when describing an object, while pens are great tools for describing objects in detail. Taking this idea, a KAIST team introduced a new 3D sketching workflow, combining the strengths of hand and pen input. This technique will ease the way for ideation in three dimensions, leading to efficient product design in terms of time and cost.
For a designer's drawing to become a product in reality, one has to transform a designer's 2D drawing into a 3D shape; however, it is difficult to infer accurate 3D shapes that match the original intention from an inaccurate 2D drawing made by hand. When creating a 3D shape from a planar 2D drawing, unobtainable information is required. On the other hand, loss of depth information occurs when a 3D shape is expressed as a 2D drawing using perspective drawing techniques.
To fill in these “missing links” during the conversion, "3D sketching" techniques have been actively studied. Their main purpose is to help designers naturally provide missing 3D shape information in a 2D drawing. For example, if a designer draws two symmetric curves from a single point of view or draws the same curves from different points of view, the geometric clues that are left in this process are collected and mathematically interpreted to define the proper 3D curve. As a result, designers can use 3D sketching to directly draw a 3D shape as if using pen and paper.
Among 3D sketching tools, sketching with hand motions, in VR environments in particular, has drawn attention because it is easy and quick. But the biggest limitation is that they cannot articulate the design solely using rough hand motions, hence they are difficult to be applied to product designs. Moreover, users may feel tired after raising their hands in the air during the entire drawing process.
Using hand motions but to elaborate designs, Professor Seok-Hyung Bae and his team from the Department of Industrial Design integrated hand motions and pen-based sketching, allocating roles according to their strengths. This new technique is called Agile 3D Sketching with Air Scaffolding. Designers use their hand motions in the air to create rough 3D shapes which will be used as scaffolds, and then they can add details with pen-based 3D sketching on a tablet (Figure 1).
Figure 1. In the agile 3D sketching workflow with air scaffolding, the user (a) makes unconstrained hand movements in the air to quickly generate rough shapes to be used as scaffolds, (b) uses the scaffolds as references and draws finer details with them, (c) produces a high-fidelity 3D concept sketch of a steering wheel in an iterative and progressive manner.
The team came up with an algorithm to identify descriptive hand motions from transitory hand motions and extract only the intended shapes from unconstrained hand motions, based on air scaffolds from the identified motions.
Through user tests, the team identified that this technique is easy to learn and use, and demonstrates good applicability. Most importantly, the users can reduce time, yet enhance the accuracy of defining the proportion and scale of products.
Eventually, this tool will be able to be applied to various fields including the automobile industry, home appliances, animations and the movie making industry, and robotics. It also can be linked to smart production technology, such as 3D printing, to make manufacturing process faster and more flexible.
PhD candidate Yongkwan Kim, who led the research project, said, “I believe the system will enhance product quality and work efficiency because designers can express their 3D ideas quickly yet accurately without using complex 3D CAD modeling software. I will make it into a product that every designer wants to use in various fields.”
“There have been many attempts to encourage creative activities in various fields by using advanced computer technology. Based on in-depth understanding of designers, we will take the lead in innovating the design process by applying cutting-edge technology,” Professor Bae added.
Professor Bae and his team from the Department of Industrial Design has been delving into developing better 3D sketching tools. They started with a 3D curve sketching system for professional designers called ILoveSketch and moved on to SketchingWithHands for designing a handheld product with first-person hand postures captured by a hand-tracking sensor. They then took their project to the next level and introduced Agile 3D Sketching with Air Scaffolding, a new 3D sketching workflow combining hand motion and pen drawing which was chosen as one of the CHI (Conference on Human Factors in Computing Systems) 2018 Best Papers by the Association for Computing Machinery.
- Click the link to watch video clip of SketchingWithHands
2018.07.25 View 8809 -
A New Theory Improves Button Designs
Pressing a button appears effortless. People easily dismisses how challenging it is. Researchers at KAIST and Aalto University in Finland, created detailed simulations of button-pressing with the goal of producing human-like presses.
The researchers argue that the key capability of the brain is a probabilistic model. The brain learns a model that allows it to predict a suitable motor command for a button. If a press fails, it can pick a very good alternative and try it out. "Without this ability, we would have to learn to use every button like it was new," tells Professor Byungjoo Lee from the Graduate School of Culture Technology at KAIST.
After successfully activating the button, the brain can tune the motor command to be more precise, use less energy and to avoid stress or pain. "These factors together, with practice, produce the fast, minimum-effort, elegant touch people are able to perform."
The brain uses probabilistic models also to extract information optimally from the sensations that arise when the finger moves and its tip touches the button. It "enriches" the ephemeral sensations optimally based on prior experience to estimate the time the button was impacted. For example, tactile sensation from the tip of the finger a better predictor for button activation than proprioception (angle position) and visual feedback.
Best performance is achieved when all sensations are considered together. To adapt, the brain must fuse their information using prior experiences. Professor Lee explains, "We believe that the brain picks up these skills over repeated button pressings that start already as a child. What appears easy for us now has been acquired over years."
The research was triggered by admiration of our remarkable capability to adapt button-pressing. Professor Antti Oulasvirta at Aalto University said, "We push a button on a remote controller differently than a piano key. The press of a skilled user is surprisingly elegant when looked at terms of timing, reliability, and energy use. We successfully press buttons without ever knowing the inner workings of a button. It is essentially a black box to our motor system. On the other hand, we also fail to activate buttons, and some buttons are known to be worse than others."
Previous research has shown that touch buttons are worse than push-buttons, but there has not been adequate theoretical explanation.
"In the past, there has been very little attention to buttons, although we use them all the time" says Dr. Sunjun Kim from Aalto University. The new theory and simulations can be used to design better buttons.
"One exciting implication of the theory is that activating the button at the moment when the sensation is strongest will help users better rhythm their keypresses."
To test this hypothesis, the researchers created a new method for changing the way buttons are activated. The technique is called Impact Activation. Instead of activating the button at first contact, it activates it when the button cap or finger hits the floor with maximum impact.
The technique was 94% better in rapid tapping than the regular activation method for a push-button (Cherry MX switch) and 37% than a regular touchscreen button using a capacitive touch sensor. The technique can be easily deployed in touchscreens. However, regular physical keyboards do not offer the required sensing capability, although special products exist (e.g., the Wooting keyboard) on which it can be implemented.
The simulations shed new light on what happens during a button press. One problem the brain must overcome is that muscles do not activate as perfectly as we will, but every press is slightly different. Moreover, a button press is very fast, occurring within 100 milliseconds, and is too fast for correcting movement. The key to understanding button-pressing is therefore to understand how the brain adapts based on the limited sensations that are the residue of the brief press event.
The researchers also used the simulation to explain differences among physical and touchscreen-based button types. Both physical and touch buttons provide clear tactile signals from the impact of the tip with the button floor. However, with the physical button this signal is more pronounced and longer.
"Where the two button types also differ is the starting height of the finger, and this makes a difference," explains Professor Lee. "When we pull up the finger from the touchscreen, it will end up at different height every time. Its down-press cannot be as accurately controlled in time as with a push-button where the finger can rest on top of the key cap."
Three scientific articles, "Neuromechanics of a Button Press", "Impact activation improves rapid button pressing", and "Moving target selection: A cue integration model", will be presented at the CHI Conference on Human Factors in Computing Systems in Montréal, Canada, in April 2018.
2018.03.22 View 6902