Only the voices of soldiers are selected and delivered

[Senior Researcher Daok Kwon, Power Support System Research Team 2, Defense Agency for Technology and Quality] The U.S. Army Combat Capabilities Development Command Army Research Laboratory (ARL), under the Army Futures Command, has collaborated with the University of Southern California’s Institute for Creative Technologies to develop a new interactive interface system called JUDI (Joint Understanding and Dialogue Interface). The U.S. Army stated that JUDI’s speech recognition algorithms enable two-way interactive communication between soldiers and autonomous systems.

Devices like Amazon’s Alexa rely on network connections through cloud services equipped with large-scale data sets, which allow them to learn tasks and perform them for users.

In contrast, JUDI’s dialogue processing method is based on statistical classification to interpret intent from soldiers’ language. Additionally, JUDI’s speech recognition device utilizes a specially developed voice model designed to overcome the acoustic characteristics of noisy and echoing battlefield environments. The JUDI system supports collaborative decision-making between soldiers and autonomous systems by leveraging multiple contextual sources, including soldiers’ conversations and the cognitive systems of robots such as visual, auditory, or radar mapping.

This is significant because it enables newly introduced robots on the battlefield to engage in more diverse and natural two-way interactions with soldiers in real combat situations, which are environments with a high level of ‘noise.’

Since the JUDI system can listen to verbal commands and identify the words with the highest overlap to apply to the instructions, it is expected to reduce the burden on soldiers of memorizing command lists to interact with robots.

In future battlefields composed of manned-unmanned combined systems, it is no exaggeration to say that victory or defeat depends on how quickly and accurately communication and interaction occur among battlefield components such as combatants and unmanned robots. The selection of artificial intelligence as a key research area in initiatives like the U.S. Army modernization strategy, which emphasizes enhancing soldier lethality and promoting unmanned systems, supports this prediction.

However, unlike relatively refined operational environments such as offices or homes where commercial AI assistants operate, building a robust AI system that can overcome various noises in harsh battlefield conditions and respond only to commands from combatants within an established command structure is expected to be technically very challenging.

The ARL has indicated that the future integration of JUDI into various software domains will likely involve the development and integration of diverse core technologies in human-robot interaction, ranging from AI-based speech recognition technology to emotion and intent recognition through brain-computer interaction. Ultimately, it is analyzed that this will be closely linked with human enhancement technologies and have significant potential for military application.

This content was produced with the assistance of AI translation services.