论文成果 / Publications

We present a novel rotation based method for detecting where a mobile device is worn on a user's body that utilizes the fusion of the data from accelerometer and gyroscope. Detecting the position of a mobile device could improve the performance of on-body sensor based human activity recognition and the adaptability of many mobile applications. In our method, the radius and angular velocity for a position is calculated based on the data read from the sensors integrated in a mobile device. We have evaluated our method with an experiment to detect four commonly used positions: breast pocket, trouser pocket, hip pocket and hand.

Mobile phone with high accessibility and usability is regarded as the ideal interface for the users to monitor and control the approaching smart home environment. Moreover, networking technologies and protocols have been advanced enough to support a universal monitoring and controlling interface on smart phones. This paper presents HouseGenie, an interactive, direct manipulation application on mobile, which supports a range of basic home monitoring and controlling functionalities as a replacement of individual remotes of smart home appliances. HouseGenie also addresses several common requirements that may be behind the vision, such as scenario, short-delay alarm, area restriction and so on. We demonstrate that HouseGenie not only provides intuitive presentations and interactions for smart home management, but also improves user experience comparing to present solutions.

We created PicoPet, a digital pet game based on mobile handheld projectors. The player can project the pet into physical environments, and the pet behaves and evolves differently according to the physical surroundings. PicoPet creates a new form of gaming experience that is directly blended into the physical world, thus could become incorporated into the player's daily life as well as reflecting their lifestyle. Multiple pets projected by multiple players can also interact with each other, potentially triggering social interactions between players. In this paper, we present the design and implementation of PicoPet, as well as directions for future explorations.

RFID-based indoor location systems have been proved to be both accurate and cost-effective. However, current implementations mainly use active tags which suffer the issues of batteries replacement, installation, maintenance and per-unit cost. Besides, as the number of users grows, how to guarantee the stability and high speed transmission of the location information can be challenging. To address these challenges, 1) we propose a passive RFID-based system for localizing multi-users, and 2) detecting human motion from various types of embedded sensors to be supplemented. In addition, we also implement a reliable transmission protocol to guarantee the location data transition between RF nodes based on dynamic PRI. According to the performance analysis, the tracking accuracy of our system is well assured. Its quick responsiveness and good scalability, as well as low cost on energy and infrastructure, make this system a more cost-effective and easy-to-deploy solution for stable indoor positioning.

Location-Based Service (LBS) is becoming a ubiquitous technology for mobile devices. In this work, we propose a signal-fusion architecture called XINS to perform effective indoor positioning and navigation. XINS uses signals from inertial navigation units as well as WiFi and floor-map constraints to detect turns, estimate travel distances, and predict locations. XINS employs non-intrusive calibration procedures to significantly reduce errors, and fuses signals synergistically to improve computational efficiency, enhance location-prediction accuracy, and conserve power.

The real-time interactive virtual classroom with tele-education experience is an important approach in distance learning. However, most current systems fail to meet the new challenges raised by the development of the serviceoriented architecture. First, the learning systems should be able to facilitate easier integration of increasingly dedicated services, such as language services on the Internet. Second, the learning systems must open their internal interfaces as web services to other systems, so as to enable deeper integration of these systems and easier deployment. Third, the systems are expected to provide flexible interfaces to support mobile device interaction. To address these issues, we build a prototype system, called Open Smart Classroom, by upgrading the original Smart Classroom into a service-oriented open system. With the help of Language Grid services, two Open Smart Classrooms deployed in Tsinghua University and Kyoto University are connected and experimental co-classes have been successfully held. The results of the user study show that integrating Smart Classroom and language services is an interesting and promising approach to building future multicultural distant learning systems.

We present uTable, a very large horizontal interactive surface which accommodates up to ten people sitting around and interacting in parallel. We identify the key aspects for building such large interactive tabletops and discuss the pros and cons of potential techniques. After several rounds of trials, we finally chose tiled rear projection for building the very large surface and DI solution for detecting touch inputs. We also present a set of techniques to narrow the interior bezels, uniform the color/brightness of the surface and handle multiple streams of inputs. Finally uTable achieves a good overall performance in terms of display effect and input capability.

We present Air Finger, a novel technique that enables controlling CD ratio by finger height above the touch sur-face for multi-scale navigation tasks. Extending previous research on virtual touch, Air Finger divides the space above surface into two layers and associates the high, medium and low CD ratios to the touch surface, the lower air and the higher air respectively. Users can fluidly switch between the three navigation scales by lifting and pressing the finger. Air Finger enables multi-scale navigation control using one hand.