The problem

There are currently about 285 million people that are visually impaired on the world, 39 million of them are blind. Despite the affectance rate is slightly reducing since the 90s, it stills represent a major health problem that affects more in developing countries.

People with serious visual impairment need to be taught about how to accomplish daily task and travel safely and independently during the daily activities. They usually rely on the use of the white cane to detect obstacles, with a small number employing guide dogs. Local and central governments have also making navigation easier for visually impaired people by adapting public transportation and street furniture to be easier to use by them.

Danger situations in outdoor life

One of the most common fears of visually impaired and partially sighted people is having accident on outdoor activities. Some common causes of accidents are falling from stairs, falling into holes or pits, bumping into objects and collisions with vehicles. Road crossing represents a high risk task due to the different types of crossing, lack of audible semaphores and pedestrian visibility. However, these dangers doesn’t translate directly into heightened risk, since blind and partially sighted people regulate their behaviours in order to avoid these dangers. Some people tend to suppress their pedestrian activity by going out less often or using taxis, which does have an impact on their physical activity and social integration.

Formal road crossings may be controlled (when they have semaphores with fixed time windows or on-demand pushbuttons to request the right of way) or uncontrolled when only black and white stripes are painted on the roadway. Controlled crossings rely on beeping sounds to indicate blind and partially sighted people that they may cross: but, for uncontrolled road crossings, the detection of a stopped vehicle is essential for providing them the peace of mind to initiate the road crossing.

Modern electric and hybrid vehicles

The usage of electric motors as car engines presents the advantage of a quieter functioning mode compared to internal combustion engines, in addition to be more environmentally friendly. The dominant car noise at low speeds is the engine sound, but with electric and hybrid cars it is almost non-existent. That presents a problem to errant pedestrians and visually impaired people.

This problems has been spotted by various blind associations that advocate to include mandatory noise generator on electric and hybrid vehicles that should be activated when the car is in movement at slow speeds. As a consequence, several laws by US government and EU government have been approved, but they won’t be mandatory until 2019. However, these noise generation devices can be deactivated on some situations, and, until cars with these devices appears, there will be a significant number of electric and hybrid vehicles in circulation without these sound generators.

The approach

The proposed device aims to detect near electric drive vehicles through its electric motor switching noise. A visually impaired person may use the device to check if it is completely safe to cross a uncontrolled road crossing.

This approach has been used previously by M. Takagi et al at University of Tokyo using the microphone contained in the cell phone to alert pedestrian about near vehicles through wearable devices (Google Glass) using sound fingerprinting to identify vehicles. They have validated the use of sound to detect them earlier than an untrained pedestrian achieving a good accuracy for a initial work.

That same sound detection technique can be used to help visually impaired pedestrians being more safe when crossing the streets. However, some changes to the system can be applied in order to be usable by them, and possibly increase accuracy and range.

Implement the system in a dedicated device: a smartphone is a fast and good development platform, but it lacks connectivity options and the user interface is not very adequate for visually impaired people, so it may not be the optimal platform for a navigation aid.

Use of better microphones: developing a dedicated device allows the selection of a microphone with better fit for the application than the one included in the phone. Multiple microphone may be used to gather more data about the environment.

The resulting device may be used in the future to analyse more information about the environment that may be useful for navigation of blind and partially sighted people.

Architecture

The expected device architecture is the following:

Challenges

The aim of this project is merely exploratory. The field of electronic helpers for visually impaired people is not developed since it requires a wide range of technologies to be mature in order to enable the apparition of devices usable in real life that effectively enhance the navigation in daily tasks in a significant way. Below, there are some challenges that this proof of concept should solve in order to be of some use to visually impaired people.

Range

On the paper published by M. Tagaki and his team, they achieve the detection of electric vehicles farther than the average pedestrian could hear them. However, blind and partially sighted people are trained to use the hearing to help their navigation, so it is plausible to think they will be able to hear a electric vehicle earlier than an untrained person. For this device to be effective, it should be able to detect approaching vehicles before a trained individual.

Acoustic fingerprinting

The vehicle detection based on acoustic fingerprinting imposes that the analysis software should have been trained to detect the acoustic features of the vehicles. Since there is a lot of different car makers and models, the classifiers should be tuned with a wide representative sample of current electric and hybrid vehicles. In addition, new car models appear every year, so the detection device should be easily upgradeable with new analysis parameters in order to maintain the detection accuracy.

User interface

Creating an user interface to present data to a visually impaired person is more difficult than a traditional one, not only due to the lack of visual representation, but also they are used to use the other senses in a different way to supplement the reduced sight.

End user acceptance

Last, but no less important, there is the issue of the acceptance of navigation aids by the users. While the use of the white cane is widely accepted, some users do not like that other people notice it. For that reason, the implications of how the device is used and how it is perceived should be taken in strong consideration on the design stage.

Acknowledgements

I’d like to thank S. Real for the insight and comments on the subject and its vision on navigation aids for visually impaired people.