Equipped with a camera and with the help of machine learning, a smart cane can warn blind people of obstacles and indicate the right direction.
The traditional white cane for blind and visually impaired people has been around for over 100 years. But while the world has changed quite a bit since it was invented in England in 1921, the functionality of the blind cane has remained virtually identical to this day.
A team of researchers at ETH Zurich is about to change this by developing an intelligent cane for the blind called NextGuide. Thanks to an integrated camera, it detects the surroundings and shows the way with a tactile pointer.
Pointing in the right direction
Using various on-device machine learning algorithms, the cane can estimate a path free of obstacles, and assess many other data points, about a hundred times per second.. The scientists integrated a camera, which makes it possible to recognize and communicate dangerous situations, read out text, and find objects in a field of view of around 160° in front of the white cane.
Haptic feedback for different obstacles
The visually impaired person receives information on where to turn to avoid obstacles with haptic feedback from the cane in the form of a tactile pointer and vibration. The smart cane is able to send different vibration signals, for example, if the person is standing in front of a door, a pedestrian crossing or a staircase. Optionally it is also possible to turn on sound output. The cane itself is made of carbon. According to the inventors, the cane is suitable for navigation inside and outside the home.
Inspiration from a former classmate
A blind former classmate inspired ETH student Alexander Bayer to work on a new smart cane. The Student Project House at ETH Zurich promotes a thriving creative community at ETH. Students like Alexander are given the opportunity to think outside the curriculum and set their own path, to gather, experiment, test concepts, supercharge group projects and build tomorrow's ideas today.