Project People Paper
English Japanese


This research proposes a novel way to allow non-expert users to create smart surroundings. Non-smart everyday objects such as furniture and appliances found in homes and offices can be converted to smart ones by attaching computers, sensors, and devices. In this way, non-smart components that form non-smart objects are made smart in advance.For our first prototype, we have developed u-Texture, a self-organizable universal panel that works as a building block. The u-Texture can change its own behavior autonomously through recognition of its location, its inclination, and surrounding environment by assembling these factors physically.


This research focuses on block materials that have uniform shapes to allow users to create various objects such as furniture, floors in homes, sidewalks outdoors, homes, or buildings by connecting or assembling them. There are many kinds of block materials such as bricks, panels, and tiles in our surroundings. As our approaches in developing the technology for non-expert users, we have been developing smart objects' block materials in advance, not by converting existing non-smart objects to smart objects. Required operations for users are just to assemble the block materials in shapes suitable for what users wish to do. The advantage of this system is that customization or calibration, which require computing skills, are unnecessary.

Assembled u-Textures create smart surroundings that have shapes such as shelves, tables, and walls, which correspond to assembled shapes. Figure 1 shows example of u-Textures and how to create smart surroundings and context-aware applications by assembling them.

Figure 1

In this case, u-Textures are used as several smart surroundings for discussing and creating a logo design for our project. First, each u-Texture is used as a drawing tool individually. Afterwards, users can connect their u-Textures each other horizontally and exchange and merge their drawing data among connected u-Textures by drag-and-drop operations. Finally, when the u-Textures are assembled horizontally and set in vertical, users can look four candidates of drawing data on each u-Texture at the same time and expand one of them on four u-Textures. Once assembled in shapes, u-Textures can autonomously provide suitable actions to the shapes by recognizing the shapes through exchanges of information on their connections and inclinations among each u-Texture. Without knowing anything about electrical connections, users can assemble u-Textures into smart objects practically and operate them to correspond to their assembled shapes.


To explain our purpose in this research, the main concepts of the u-Texture are considered as follows:

Easy Assembling A user can assemble u-Textures easily without knowing the configuration with computers, sensors, and networks inside each u-Texture. No specific tools are necessary for assembling u-Textures.

Self-Recognition As a user assembles u-Textures, the assembled u-Textures exchange information such as whether connected or not, connection directions, IDs of adjoining u-Textures, and the u-Textures' inclination. With that information, each u-Texture recognizes its assembled shape as well as its location and inclination in the assembled shape.

Behaviors to the Shape Available applications corresponding to the shape of the assembled u-Textures will be extracted automatically among multiple applications pre-installed in each u-Texture. When there are several candidate applications to the shape, with minimum selection of the user, each u-Texture behaves autonomously and runs the same application together.

Assembling u-Textures

Figure 2 shows a user assembling u-Textures and establishing a shelf-shaped smart object: from top; A user assembles u-Textures. The u-Textures indicate application candidates corresponding to the assembled shapes. A user selects a desired application. The u-Textures cooperate and run the selected application.

Figure 2

System Design

The prototype of u-Texture is 320 mm square, 48 mm thick, and weights 4300 g. Every u-Texture is designed to the same specification to enable users to assemble smart objects with any u-Textures. It costs more to equip full functions in all the u-Textures than to install them with limited functions according to assembled systems. Since one of our purposes for developing this prototype is to confirm various potential advantages that u-Textures can provide, we designed sensors and networks redundantly. The u-Texture should have a structural function possible to be assembled and an electrical function possible to be connected electrically. In the current version, a blockable prop, u-Joint (Figure 3, left), supports to assemble u-Textures.

Figure 3

System Architecture

Sensing Capability Figure 4 shows a block diagram of the u-Texture and the u-Joint.

Figure 4

Each side of the u-Texture and the u-Joint is equipped with an RS-232C interface as a connection sensor examines whether the u-Texture is connected to other u-Textures or not, and its direction if it is connected. The 3D sensor is used to obtain inclination data. The sensor is created with two dual axis accelerometers (ADXL202, Analog Devices). Infrared sensors are located on each side of the u-Texture as a proximity sensor, and are used to detect nearby or incoming remote u-Textures. All sensors outputs mentioned above are processed by a microcontroller (H8/3664F 16MHx, Hitachi). The microcontroller dispatches the network ID of the u-Texture and the network IDs of the connected u-Textures located on its sides. The microcontroller can also simultaneously receive the same data from connected u-Textures, and then send the data to the main computer to have the data processed.

Computing and Networking Capability The main computer consists of the Mobile Pentium processor, a speaker, an IEEE 802.11b wireless network and display (SXGA 14.1type). All are parts adapted from a laptop computer (CF-Y2DM1AXR, Panasonic). A tough panel (AST-140, DMC) enables direct inputs. An RFID tag reader/writer (13.56MHz, Sofel) is embedded and antennas are built-in around the display. Ethernet interfaces are also implemented on each side of u-Texture. The basic idea behind a wired network for exchanging data among the assembled u-Textures internally is to create an extemporaneous bridge network with broadcast support and loop back prevention. This is accomplished by binding each Ethernet interface into one pseudo network device, and binding an active network ID to the device. With additional help from a loop back prevention protocol, the packet storming within the network can be reduced. The detailed design is described in our other paper.

u-Joint The u-Joint provides power to connected u-Textures. The RS-232C and Ethernet interfaces are built into the u-Joint to transmit and receive data between u-Textures.

Structural Interface The u-Texture is defined with directions of up and down, and left and right physically as shown in Figure 3. u-Joint is also defined up and down. To simplify the automatic structure recognition by the software, there are three assembly regulations as follows. (1) u-Texture should be assembled to others horizontally or vertically via u-Joints. (2) The surfaces of the u-Textures should be connected with the same directions when assembling them horizontally. (3) The top of the u-Texture should be connected to the bottom of the other u-Texture and the right side to the left side. The structure of the u-Texture has an original connecting mechanism to make users unconsciously adhere to the regulations.


AwareShelf The AwareShelf can be created on a shelf-shaped u-Textures as shown in Figure 2. When a user puts a real object such as a camera, a book, or a key on a u-Texture, it enables to browse information of the real object on the display on another u-Texture. The u-Textures have to be connected vertically to the u-Texture on that a thing is placed. For example, a user can check an electrical manual of a camera, or duration of times that a key was put on the AwareShelf before. In this application, real objects are attached unique RFID tags and u-Texture recognizes the tag IDs. When the AwareShelf connects to an external network, a user can look for real objects. Currently, the advanced application has been implemented so that a user can find the AwareShelf on which the real object was placed by using a mobile phone. In this application, the AwareShelf is used as a system for storing and recognizing objects(Figure 5 ).

Figure 5

CollaborationTable The Collaboration Table is a system that supports cooperative work with several participants by connecting u-Textures horizontally as shown in Figure 6. Users can exchange and merge drawing data among connected u-Textures by drag-and-drop operations. One advantage of this application is that it is possible to expand users' creative ideas and collaborations in cooperative works. Another advantage is that it is possible to use a u-Texture both privately and publicly by connecting a u-Texture or setting it closer to another u-Texture.

Figure 6

ProjectionWall The ProjectionWall magnifies a connected u-Texture's small display onto a big one as shown in Figure 7 . It is effective for displaying a large picture that is too small to be shown on only one u-Texture. Data can be handled interactively by users with touch panels. Although the similar interactive large display can be created with projectors, users have to establish or attach computers, sensors connected to the projector and to calibrate them. Converting coordinates for magnifying data is implemented by automatic cooperation among u-Textures, which can be one of the advantages for application developers and non-expert users in the respect that it is not necessary to be aware of changing coordinates.

Figure 7


This work is part of the Ubila Project, and done in cooperation with Uchida Yoko Corporated

copyright (C) UbiLab, Keio University All Rights Reserved.
contact: uservice at