Guide dogs play a vital role in helping visually impaired individuals navigate safely, but they’ve never been able to communicate verbally with their owners—until now. Researchers at Binghamton University, part of the State University of New York, have developed a robotic guide dog powered by large language models that can plan optimal routes, guide users safely, and provide real-time spoken feedback throughout the journey.
The study, titled “From Woofs to Words: Towards Intelligent Robotic Guide Dogs with Verbal Communication,” was presented at the 40th AAAI Conference on Artificial Intelligence (AAAI 2026), held in Singapore from January 20 to 27, and is also available on arXiv.
“For this work, we’re showcasing a capability in robotic guide dogs that goes beyond what biological guide dogs can achieve,” said Shiqi Zhang, an associate professor at the Thomas J. Watson College of Engineering and Applied Science’s School of Computing. “While real dogs typically understand only about 20 commands, robotic guide dogs use GPT-4 and voice interaction, enabling them to process far more complex language.”
Smarter Guide Dogs with Two-Way Voice Navigation
Zhang and his team had earlier trained robotic guide dogs to assist visually impaired users by following leash-based cues. The new system advances this approach by enabling two-way spoken interaction between the user and the robot, offering greater control and awareness of the surroundings. It provides route details before the journey—referred to as plan verbalization—and continuous updates during movement, known as scene verbalization.
“This is especially important for blind or visually impaired individuals, as their ability to perceive the environment and surroundings is significantly reduced without sight,” Zhang explained.
To evaluate the system, the researchers enlisted seven legally blind participants to navigate a large, multi-room office space. The robot first asked each user for their destination—such as a conference room—then suggested possible routes along with estimated travel times. After the user chose a route, the robot guided them to the destination while continuously describing the environment and any obstacles (for example, noting “this is a long corridor”) until arrival.
Afterward, participants completed a questionnaire assessing the system’s usefulness, ease of interaction, and overall effectiveness. Most users favored a combined approach that included both route planning explanations and real-time verbal updates. A simulated study further confirmed the effectiveness of this method.
Toward Everyday Use of Robotic Guide Dogs
Looking ahead, the researchers plan to expand user testing, enhance the system’s autonomy, and enable the robots to navigate longer distances in both indoor and outdoor settings.
The aim of this research is to bring robotic guide dogs closer to everyday use, and participants showed strong enthusiasm for that future.
“They were extremely excited about the technology and the robots,” Zhang said. “They asked a lot of questions and clearly recognized its potential, expressing hope to see it become a reality.”
Read the original article on: Tech Xplore
Read more: Tennibot unveils the Partner V2, its newest robotic tennis ball machine