While there are already apps that guide blind users to the approximate GPS coordinates of a bus stop, these people may unknowingly stand too far away from the actual stop. A new app solves this flaw by getting your smartphone's camera involved.
The artificial intelligence application, called "All_Aboard," was developed by a team of scientists at the Harvard-affiliated Massachusetts Eye and Ear Treatment and Research Center. It can be used with third-party GPS-based navigation applications such as Google Maps.
Users first use other applications to determine the approximate location of bus stops. Then open All_Aboard and hold up your smartphone to let the rear camera "see" the surrounding street.
According to reports, the application uses a deep learning neural network to be trained on about 10,000 bus stop images in the city, and it can intuitively identify the target stop sign if it is within 50 feet (15 meters) away. Once a sign is spotted, the app guides the user with sonar-like beeps that change in pitch and speed as the user gets closer to the bus stop.
In field testing, 24 blind volunteers used both Google Maps and All_Aboard to locate a total of 20 bus stops, 10 of which were in urban areas (Boston) and 10 in suburban areas (Newton, Massachusetts).
When it comes to being "close enough" to these stations, Google Maps only succeeds 52% of the time, while All_Aboard improves the success rate to 93%. Additionally, the average distance between Google Maps' map end points and actual bus stops is 6.62 meters (21.7 feet), while All_Aboard's average distance is just 1.54 meters (5 feet).
"Our research results show that the All_Aboard app can help visually impaired travelers navigate and accurately detect bus stops, thereby greatly reducing their chance of missing the bus due to standing too far away from the bus stop," said Luo Gang, associate professor of the Department of Eye and Ear at Massachusetts. "This study demonstrates that computer vision-based object recognition capabilities can be leveraged in a complementary manner and bring additional benefits to purely map-based macro-navigation services in real-world environments."
So far, All_Aboard has been trained in 10 major cities around the world and is available on iPhone via the AppStore. It was recently introduced in a paper published in the journal Translational Vision Science & Technology, and demonstrated in the video below.