Robotic guide dog uses AI to talk with visually impaired users during navigation

AI-Powered Guide Dogs Transform Mobility for the Blind

Researchers at Binghamton University have developed a robotic guide dog that goes beyond traditional service animals by adding a capability no canine could match: conversation. Led by associate professor Shiqi Zhang, the team equipped a Unitree Go2 robotic dog with large language models, specifically GPT-4, to create an assistive technology that can discuss routes, describe surroundings, and respond to complex voice commands 1

1

4

4

.

The innovation addresses a critical gap in accessibility. Traditional seeing-eye dogs require extensive training, with only 50-60% graduating from programs, and cost between $20,000-50,000. As a result, only 2-5% of the blind community can access these animals 1

1

. The robotic alternative promises to expand access while offering enhanced situational awareness through verbal communication.

How the Guidance System Works

The AI robotic guide dog system operates through what researchers call "plan verbalization" and "scene verbalization." Before departure, the robot asks users where they want to go, then presents multiple route options with estimated travel times. During the journey, it provides real-time spoken feedback about the environment, such as "this is a long corridor" or "you're passing by the main lobby, which is an open area with seating and information desks" 1

1

2

2

.

Shiqi Zhang emphasizes the advantage over biological guide dogs: "Real dogs can understand around 20 commands at best. But for robotic guide dogs, you can just put GPT-4 with voice commands. Then it has very strong language capabilities" 1

1

4

4

.

The system relies on link grounding, matching spoken words to real objects and locations within a mapped environment. When a user mentions thirst, the robot can connect that request to a fountain or vending machine, keeping conversation tied to actionable navigation 2

2

.

Testing with Legally Blind Participants

Seven legally blind participants navigated a large, multi-room office environment to test the system. Following their experience, they completed questionnaires rating helpfulness, ease of communication, and usefulness on a five-point scale. The combined approach of route planning and navigation with real-time narration scored 4.83 for usefulness and 4.50 for ease of communication 2

2

3

3

.

Participants preferred the combination of verbal and physical guidance through leash pressure rather than being pulled along without explanation. However, the safety rating slipped to 3.83, which researchers attribute to unfamiliarity with walking alongside a robot rather than the traditional human-canine bonding experience 1

1

2

2

.

Despite trust concerns about safety, participants showed enthusiasm for the technology. "They were super excited about the technology, about the robots," Zhang said. "They asked many questions. They really see the potential for the technology and hope to see this working" 1

1

4

4

.

What Comes Next for This Alternative to Service Animals

The research team plans to conduct more studies with longer indoor distances and outdoor testing. Increasing system autonomy remains a priority, as current demonstrations used a "Wizard of Oz" setup where a hidden expert controlled robot movement during walks to reduce risk while preserving the conversation test 2

2

.

In additional testing, GPT-4 successfully navigated the dog through 77 different scenarios using natural language commands 1

1

. The team presented their paper "From Woofs to Words: Towards Intelligent Robotic Guide Dogs with Verbal Communication" at the 40th Annual AAAI Conference on Artificial Intelligence in January 4

4

.

For the blind community to assist visually impaired users effectively, robotic guide dogs must prove they can handle complex environments beyond controlled office settings. The lower safety rating signals that building daily trust will matter more than demonstration success. Yet user autonomy through conversation represents a meaningful step forward, turning navigation into a shared dialogue rather than simple commands. If researchers can solve autonomy challenges and outdoor navigation, AI-powered guide dogs could become a practical mobility aid that explains itself while moving safely through the world.