Intel-NTU Connected Context Computing Center Successfully Revealed State-of-the-Art Technology on Intel Asia Innovation Summit 2014


 
Intel-NTU Connected Context Computing Center successfully revealed diverse state-of-the-art IoT (Internet of Things) technology on Intel Asia Innovation Summit 2014. Through 4-year  collaboration between Ministry of Science and Technology, National Taiwan University and Intel Corporation, the research outcomes of IoT and machine-to-machine (M2M) technology become mature and the preliminary research gradually become influential application to human life.
 
The research demos in this year, our center emphasized more on automobile safety and home care. Moreover, we also expanded the M2M application to IOT of art and education. The research demos are included "Intelligent Transportation System", "Micro Piezoelectric Vibration Energy Harvesters and Its Interfacing Circuits", "IoT for Art" and "RollingLight: Enabling Line-of-Sight Light-to-Camera Communication", etc. 
 
In addition, congradulating to Prof. Hsin-Mu Tsai, Prof. Ching-Ju Lin, Prof. Shao-Yi Chien and Dr. Chia-Han Lee of winning Intel Distinguished Collaborative Research Award in Intel Asia Innovation Summit 2014!

 
Intelligent Transportation System
https://www.youtube.com/embed/kQZ-xNNfdtc
In this project, our goal is to develop a proactive driver assistance system which includes 1) M2M-based neighbor map building, 2) driver behavior modeling and prediction, and 3) design of passive information visualization and proactive warning mechanism. To achieve these, we based on the structure of M2M technology to develop a function named Micro-Navigation to enhance intellihgent transportation system. The GPS system today provides main information of which road o drive or whether the drivers need to turn or not, etc. Furthmore, the Micro-Navigation provide further information to drivers about whether to change the lane, accelerate or reduce speed, or away from the specific dangerous cars. For example, exploiting machine learning and data mining models for ITS data analysis (e.g. red-light runner detection, driven aggressiveness classification, aggressive turn detection) in order to achieve the multiple goals of  safety and oil safeness.
 
 
Distributed Video Cameras
https://www.youtube.com/embed/r_crxtFIxTU
Other than distributed video coding, we will also pay more attention on distributed video analysis. For distributed video coding, our goal is to develop the state-of-the-art and practical DVC, including multi-resolution/multi-view DVC and HEVC/H.265 based DVC. We plan to define an API that can be used to implement video applications on distributed video sensors and develop the associated hardware video analysis engine for distributed video sensors. The sensor configurations can also be selected with the control interface to achieve a good balance between power consumption and recognition performance. Before the end of this project, these techniques will be implemented in an FPGA system as well as chip tape-out. The developed system will finally be deployed for selected real applications.
 
Micro Piezoelectric Vibration Energy Harvesters and Its Interfacing Circuits

We have introduced self-powered synchronous switches in to the interfacing circuit, but find the voltage requirement quite high and the energy loss of the rectifying circuit is considerable in low power occasions. To solve the interface circuit loss problem, we propose a three-phase solution. Firstly, models of discrete components are analyzed. In the same time, experiments using discrete components are performed to discrete realistic occasions. Then, dominant characteristics of the components that determine the loss are specified. In the end of this stage, we should be able to choose appropriate discrete components to proof the best efficiency. Finally, integrated circuit based on CMOS can be designed and conducted. The ultimate device should be able to be packaged in an IC package, i.e., ball grid packaging. We also aim to reduce the resonance frequency value to meet the requirements of wearable energy harvesting devices. Hence we have searched for polymers and other possible structures to eliminate the operating frequency. The device design for wearable energy is our ultimate goal. 
 
IoT for Art

IoT for art demos the possibility of integrate smart devices to perform art activities. Showed two examples as belows. One is the Smiling Buddha, an interactive art installatio, which enables users to exchange their smile to the portraits on the display wall. The other is the Dance Man which enables users to draw their avatars independent and eventually make them to dance together on a combined display.
 
In the display, we encouraged people in all ages to draw pictures on smart devices. Through the M2M information exchanges, all smart devices can be integrated into a combined display. Users can produce mudic and make the pictures dancing with. Those pictures can interaction over the limitation of each screen. The application could accelerate the imagination of M2M.
 
RollingLight - Universal Camera Communications for Single LED

RollingLight - Universal Camera Communications for Single LED, that enables a light to talk to diverse off-the-shelf rolling shutter cameras. To boost the data rate and enhance reliability, RollingLight addresses the following practical challenges. First, its demodulation algorithm allows cameras with heterogeneous sampling rates to accurately decode high-order frequency modulation in realtime. Second, it incorporates a number of designs to resolve the issues caused by inherently unsynchronized lightto-camera channels. The technoligy can be applied in localization to increase the accuracy while driving to enchance automobile safety.