Science & Health

Latest Stories

Advancing Accessibility: AI-Driven Robot Reads Braille Twice the Speed of Humans

Written by: Archie Williamson | Posted: 19-02-2024

Advancing Accessibility: AI-Driven Robot Reads Braille Twice the Speed of Humans

This is an AI-generated image created with Midjourney by Molly-Anna MaQuirl 

 

Researchers at the University of Cambridge have announced they have made a robotic sensor that incorporates artificial intelligence techniques to read braille at around double the speed of most human readers.

The robotic sensor can read braille with 87% accuracy at a record reading speed of 315 words per minute. The sensor uses AI algorithms and ‘high sensitivity’ to interpret braille with good levels of accuracy by mimicking human-like reading behaviour. This revolutionary breakthrough could bring new opportunities for assisting those with vision impairments. 

The robot was not originally designed to assist those with a disability; rather, it was developed to work as an AI translator for tools such as screen readers. However, this ‘high sensitivity’ breakthrough could lead to other practical applications, too, such as developing robotic hands and prosthetics.

This development has proven that sensors can reproduce human tactile skills in a landmark example of AI applications. This will likely be just one example of a huge number of potential uses to come to light in the near future.

It seems to be the case that every day in the world of AI news there is a new development and a new application. People across the professional spectrum are starting to understand the real-life applications of AI and how it can drive life forward, increasing the quality of life for a variety of people, including by being used as an accessibility aid.

The AI-powered “Fingertip”

This discovery is part of the research team’s ongoing pursuit to recreate the ‘sensitivity of touch’.

The researchers have had the difficult task of replicating what the human fingertip can do. Human fingertips are so sensitive that they can detect even minuscule changes and show how much force to use when carrying or grasping things. This is how we are able to do things like pick up a pencil with the right force without snapping it.

The challenge has come from reproducing this sensitivity in a way that is energy efficient. Professor Fumiya Iida has led the research, helping develop ways for the sort of ‘fingertip’ tasks to be carried out using cameras.

The softness of human fingertips is one of the reasons we’re able to grip things with the right amount of pressure,” said Parth Potdar, a researcher within Cambridge’s Department of Engineering. “For robotics, softness is a useful characteristic, but you also need lots of sensor information, and it’s tricky to have both at once, especially when dealing with flexible or deformable surfaces.”

Using an over-the-counter sensor, the team was able to produce an algorithm that took the images and ‘deblurred’ them before the sensor tried to recognise the letters. The algorithm was trained using a set of braille images but with a blur added in editing software afterwards. Once the algorithm was trained, a computer vision model detected each character.

As this technology gets better at understanding real-life ‘blurred’ imagery, as well as artificially processed ones, it is very possible that the algorithm may allow tools to perform significantly better in the future and with even more speed and impressive accuracy.

Explore More News

Advancing Accessibility: AI-Driven Robot Reads Braille Twice the Speed of Humans

Written by: Archie Williamson | Posted: 19-02-2024

Advancing Accessibility: AI-Driven Robot Reads Braille Twice the Speed of Humans

This is an AI-generated image created with Midjourney by Molly-Anna MaQuirl 

 

Researchers at the University of Cambridge have announced they have made a robotic sensor that incorporates artificial intelligence techniques to read braille at around double the speed of most human readers.

The robotic sensor can read braille with 87% accuracy at a record reading speed of 315 words per minute. The sensor uses AI algorithms and ‘high sensitivity’ to interpret braille with good levels of accuracy by mimicking human-like reading behaviour. This revolutionary breakthrough could bring new opportunities for assisting those with vision impairments. 

The robot was not originally designed to assist those with a disability; rather, it was developed to work as an AI translator for tools such as screen readers. However, this ‘high sensitivity’ breakthrough could lead to other practical applications, too, such as developing robotic hands and prosthetics.

This development has proven that sensors can reproduce human tactile skills in a landmark example of AI applications. This will likely be just one example of a huge number of potential uses to come to light in the near future.

It seems to be the case that every day in the world of AI news there is a new development and a new application. People across the professional spectrum are starting to understand the real-life applications of AI and how it can drive life forward, increasing the quality of life for a variety of people, including by being used as an accessibility aid.

The AI-powered “Fingertip”

This discovery is part of the research team’s ongoing pursuit to recreate the ‘sensitivity of touch’.

The researchers have had the difficult task of replicating what the human fingertip can do. Human fingertips are so sensitive that they can detect even minuscule changes and show how much force to use when carrying or grasping things. This is how we are able to do things like pick up a pencil with the right force without snapping it.

The challenge has come from reproducing this sensitivity in a way that is energy efficient. Professor Fumiya Iida has led the research, helping develop ways for the sort of ‘fingertip’ tasks to be carried out using cameras.

The softness of human fingertips is one of the reasons we’re able to grip things with the right amount of pressure,” said Parth Potdar, a researcher within Cambridge’s Department of Engineering. “For robotics, softness is a useful characteristic, but you also need lots of sensor information, and it’s tricky to have both at once, especially when dealing with flexible or deformable surfaces.”

Using an over-the-counter sensor, the team was able to produce an algorithm that took the images and ‘deblurred’ them before the sensor tried to recognise the letters. The algorithm was trained using a set of braille images but with a blur added in editing software afterwards. Once the algorithm was trained, a computer vision model detected each character.

As this technology gets better at understanding real-life ‘blurred’ imagery, as well as artificially processed ones, it is very possible that the algorithm may allow tools to perform significantly better in the future and with even more speed and impressive accuracy.