A new wearable device that uses artificial intelligence (AI) to help visually impaired individuals navigate their surroundings is showing early signs of promise, outperforming traditional white canes in controlled trials. As reported in Nature Machine Intelligence on April 14, the prototype system uses a camera-equipped pair of glasses, along with AI algorithms, audio feedback, and tactile cues, to help users avoid obstacles and orient themselves in space.
In tests involving 20 participants with visual impairments, the AI-boosted navigation system improved walking distance and reduced navigation time by 25% compared to using a cane alone, when traversing a 25-metre indoor maze. “This system can partially replace the eyes,” said Leilei Gu, co-author of the study and an AI researcher at Shanghai Jiao Tong University.
The core of the system is a small camera mounted on glasses, which captures real-time images of the user’s surroundings. A mini-computer processes the footage using machine-learning algorithms trained to recognize objects such as walls, doors, furniture, and other people. Users receive directional cues through earphones every 250 milliseconds, with beeps in the left or right ear indicating which way to go.
To complement the audio guidance, the researchers developed flexible ‘artificial skin’ patches worn on the wrists and fingers. These patches vibrate when an object is nearby — within a range of 40 to 5 centimeters — and also provide feedback when reaching for objects, signaling the ideal moment to grasp them.
“This paper is about how to make a very, very intelligent stick,” said Botond Roska, director of the Institute of Molecular and Clinical Ophthalmology Basel in Switzerland, who was not involved in the research. He praised the concept but cautioned that it’s “too early to say” how widely the device will be adopted.
While white canes are limited to detecting obstacles roughly one meter ahead, the AI-powered camera system offers a broader and more nuanced view of the environment. “It can only touch the environment; it cannot know what the object is,” Gu said of conventional canes.
In one trial, 12 users tested the system indoors and achieved walking speeds comparable to those using a cane. A second phase saw 8 participants take the technology into the real world — navigating city streets and cluttered conference rooms — with similarly encouraging results.
Eduardo Fernández Jover, a neuroscientist at Miguel Hernández University in Spain, called the system’s dual feedback approach “unique,” noting that combining audio and touch enhances spatial awareness. However, he emphasized the need for broader testing across more diverse user groups.
Looking ahead, the researchers plan to refine the design, making the device more compact and discreet. “Then it’s more invisible,” Gu said, suggesting future iterations could feature cameras embedded in contact lenses to reduce the device’s visibility and improve comfort.
Roska emphasized the importance of co-developing the technology with the blind community and major eye hospitals. “To move forward, they really have to work together,” he said.
While still a prototype, the device represents a promising fusion of wearable tech and AI to expand accessibility, providing a glimpse into how intelligent systems might someday offer enhanced independence for those with visual impairments.
