Dreamvision [1] was a collaboration in 2001, with VTT Information Technology and Finnish Federation of the Visually Impaired. My role in the project was the electronics and interaction designer, and I worked with an architectural student who co-designed and built the casings. I designed and implemented all electronics and programming in the firmware, as well as protocol design. VTT implemented the DLL-libraries for drivers, and the actual game.

The task required a design and implementation of a user interface device for the blind. The project goal was simply described as: “Create something that you do not recognise, and that you can use with your hands. Preferably something made from different parts that can be assembled during the game.” Major requirements were that it should be wireless, have a usable range of at least three meters and it could be used continuously for at least four hours.

VTT hired script-writers and actors, and supplied their own coders to implement the game mechanics.
Fitting the loose specification-user interface to the game framework started with creating an early sensor-prototype, working according to the specification of a communication protocol designed for the project. This was then used by VTT to create a program that understands the sensor data, and was be used to enable parallel development of the overall system components. VTT was working on the game mechanics and interfacing the data coming from the device, while our team at TUT focused on the development of the hand held user interface.
Once the actual game user interface called the Dreamhandler was ready, it could be used almost immediately with the game core.

As the Dream Vision game[2] was based on the imagination of the user, this created a need for an unidentifiable object for manipulating the game. All feedback from the game came in the form of speech and sound effects from a surround sound system. The intended user group of the game were the legally blind, who generally have no, or very limited, means for receiving visual information, but have well developed touch and aural senses. There was an intention to use a vibration feedback, but this was later discarded.

The game could have used some existing user interface devices, however that was ruled against, since the affordances of those input devices could have been already known to the user, creating expectations for the use. There was a need for something different, which led to the design of the DreamHandler system, which consisted of the device itself, a wireless communication system, a PC, and in the final context, full 5.1 audio system.

The requirements were set by the project team together with end users. The device should be used with either one or both hands, and it should not weigh too much. It should be able to detect most common gestures, such as waving hand from side to side, shaking hand, making circles and such. Measurement of the strength of the grip and vibration feedback were features that could be used, although they were not mandatory.  Major requirements were that it should be wireless, have a usable range of at least three meters and it could be used continuously for at least four hours.

During later stage, more requirements were added; most notable of those was optional use of either a cable or radio for data transfer from the handle to the master. There was also an idea of using a detachable handle, which could be put together as a part of the game. Finally, three different versions were designed; a coat hanger, a small bat, and a small bat with  detachable handle. Of these, the small bat was implemented.

DreamHandler system was to be easy to use. For the PC to interpret signals more easily, the handle should be held in hand the same way all the time. Handle or handles, depending on the design, shouldn’t allow for wrong grip. The handle should be ergonomic enough to allow at least two hours of continuous play. It should also allow natural hand movement, for the natural interfacing to the game.  Since the handle would be wireless, and use batteries, there would have to be way to easily change the batteries. As the handle would be a gaming device, it should also withstand even most rigorous moves and gestures that can happen during a game.


Figure 1. The coat hanger with sensor axes.



Figure 2. The wireless bat-UI

The first concept was the coat hanger version, shown in Figure 1. It was intended to be constructible during the game. The first actual built version was a device resembling small bat, shown in Figure 2.

The implemented prototype was a veined slab, with thick handle on the other end. The whole design was encased in plastic, with the exception of the aluminium handle. The plastic slab part was veined with hot glue to give it an unnatural but lifelike feeling, and then covered with black suede leather. The handle was covered with a thin neoprene sheet, and formed to give correct position for fingers and palm. This also reduced some of the work required for the signal processing, since the handler would always be held the same way.

The small bat had two PCB:s, one for the handle, and one for the slab. The handle contained two acceleration sensors, for horizontal movement and vertical movement. The handle had also a strain gauge sensor for the measuring the strength of the grip. There were three acceleration sensors on the slab. These cover all three dimensions, as with the central unit in the coat hanger design. The slab also contains the RF-module, power electronics, the vibrating-motor and the batteries.

The placement of the horizontal and vertical acceleration sensors allowed for differentiation of small and large swings. This is illustrated in the Figure 3, with the cross as the centre of rotation. The sensors on the handle are not moving, if the swing is small. On the other hand, if the swing is large, both the handle and the slab are moving.

Figure 3. Large and small swing, with cross marking center of the rotation

Figure 4. Dreamhandler,  the wired version.

The wireless version was implemented first, and then later, a second version with a detachable cable was built, shown in Figure 4. This version was constructed out of aluminium and layered wood. The prototypes fulfilled and exceeded all requirements. The final wireless device had about 20 meters of range, and the battery consumption level low enough to keep it running for roughly a week, depending on the usage. It was also notable for being designed and implemented a few years prior to the launch of Wii-console. In the end, the game environment was also built to the premises of Finnish Federation of the Visually Impaired, where it has been used, what I’ve been told, quite a bit.

[1] http://www.tiedetoimittaja.com/english/pages/dreamvision.html (Retrieved 22 November 2015)
[2] Väätänen, A., Strömberg, H., Cluitmans, L., & Urhemaa, T. (2004). Dream Vision: Gesture Recognition User Interface in an Interactive Radio Play. , Intelligent User Interfaces,  Funchal, Madeira, 12-14 January 2004