James C. Ferrans

Educating the Next Generation of Mobile Game Developers

Mobile gaming is one of the fastest growing segments in the video game industry. Through a partnership between Motorola Laboratories and GamePipe Laboratory at the University of Southern California, we are exploring Linuxs capabilities for mobile gaming and to provide developers with an alternative to what is predominantly a Java-based medium. Moving beyond Java lets game developers fully leverage the hardware advances and software capabilities of high-end smart phones

Enabling a voice modality in mobile games through VoiceXML

The use of speech recognition in gaming applications is not entirely new. The growth of voice as a part of gaming has exploded largely due to the popularity of online player matchmaking services such as Xbox Live. Yet, the majority of its use is only limited to communications between players to coordinate game play activities through the use of a headset and a microphone. To further explore the possibilities of using speech recognition to affect game play directly, Motorola has partnered with GamePipe Labs at the University of Southern California. This collaboration aims at leveraging the capabilities of VoiceXML (VXML), and use interactive voice dialogues to directly affect game play on mobile phones. This short paper describes the efforts taken under this initiative, and the results of this collaboration.

Bypassing Bluetooth Device Discovery Using a Multimodal User Interface

A fixed Bluetooth transmitter can offer mobile services to people in proximity. However, the Bluetooth device discovery process introduces a significant delay, which impacts usability. A number of solutions have been proposed for bypassing device discovery; in this paper we propose a new technique leveraging multimodal user interfaces. In our approach, someone coming into range of proximity-based services reads a sign listing their names. The person starts a multimodal service browser, and speaks the name of a service. This name, plus the current location (e.g., at the granularity of the current cell tower identifier), is mapped into a Bluetooth address and an optional service identifier. The service browser then connects directly to the transmitter and service. We implemented this approach on handsets connecting into a commercial voice server over a 2G data network. The server-side Bluetooth service directory consisted of 5,727 unique service names. An experiment demonstrated a very good speech recognition rate, and confirmed that the overall time savings were equivalent to other techniques for bypassing Bluetooth device discovery. The multimodal approach also has significant cost and usability advantages.