Notifications
Clear all

GPS navigation for Visually Impaired

11 Posts
4 Users
6 Likes
739 Views
ss
 ss
(@ss)
Eminent Member
Joined: 1 year ago
Posts: 14
Topic starter  

Hello everyone! I am an absolute beginner and going to build my very first project. I would truly appreciate it if you could please give me some suggestions or any related resources for my project. Through my research, I came up with the following methodology explained below although I did not get any relevant tutorial to support my project.

Well, I wish to make a GPS navigation system on a cane for a blind person. So basically, the user will give the destination's name as the input to the voice recognition module. The system compares the destination's name with the stored location in the database and selects the latitude and longitude of the destination location. The GPS continuously receives the latitude and longitude of the current location and compares it with that of the destination. The system will then send a signal to the vibration motor to provide the user with tactile feedback in the form of vibrations (for navigation directions). The latter will be able to guide the person to where they are supposed to go. 


   
Quote
Will
 Will
(@will)
Famed Member
Joined: 2 years ago
Posts: 2235
 

@maheen

If you're doing a direct comparison between the current and target locations, isn't that going to lead the person into crossing traffic (diagonally and not at crosswalks) and trying to pass through buildings and other obstacles ?

This really seems like a very advanced undertaking for a first time project, are you really sure that you're ready for the challenge ?

If you still feel that you must do it, I would suggest that you do not use the cane to indicate change of direction since they'll be used to the "feel" of the cane and extra weight and the inertial effects will affect the feedback they get from it.

You may be better to consider some form of holder which clips over a belt, for instance. That would provide a secure way to carry the system without interfering with the cane, allow more weight and volume for batteries and circuitry, and also allow for sending voice commands to the user on wired earphone(s).

Experience is what you get when you don't get what you want.


   
ReplyQuote
ss
 ss
(@ss)
Eminent Member
Joined: 1 year ago
Posts: 14
Topic starter  

@will Thanks a ton!

By the way, what do you think if I use Google assistant (Google Map API) integrated with Raspberry Pi for navigation (instead of the idea proposed above)?


   
ReplyQuote
Will
 Will
(@will)
Famed Member
Joined: 2 years ago
Posts: 2235
 

@maheen 

Sorry, I can't advise you on that asI've not used either one before.

Experience is what you get when you don't get what you want.


   
ReplyQuote
ss
 ss
(@ss)
Eminent Member
Joined: 1 year ago
Posts: 14
Topic starter  

@will Thanks for your reply.


   
ReplyQuote
DroneBot Workshop
(@dronebot-workshop)
Noble Member Admin
Joined: 4 years ago
Posts: 1174
 

@will makes an excellent point regarding traffic and buildings.  Other people, parked cars, fire hydrants and lampposts also could be an issue.

One other thing to consider is the accuracy of a standard GPS unit is about 2 meters, which would not be accurate enough to safely navigate someone down a sidewalk without putting them in the middle of the road.

Essentially, your navigation requirements are the same as a robot or self-driving car would have, and the technologies used to meet these requirements (GPS RTK, LIDAR) are both expensive and somewhat bulky.

Plus, unlike a seeing eye dog, a GPS-based system would be useless indoors.

There are existing apps and devices that use GPS to assist visually impaired people, Wikipedia has an article about them.

While I admire your initiative, I also agree with Will that for a beginner that is quite a formidable project to undertake. I'm not a beginner, and I doubt that I could build such a device.

😎

Bill

"Never trust a computer you can’t throw out a window." — Steve Wozniak


   
DaveE reacted
ReplyQuote
ss
 ss
(@ss)
Eminent Member
Joined: 1 year ago
Posts: 14
Topic starter  

Thanks for your input @dronebot-workshop!

During my research, I found the following link for GPS navigation. What are your views? The difficulty level...? Grateful for your answer.

https://www.diva-portal.org/smash/get/diva2:957841/FULLTEXT01.pdf


   
ReplyQuote
ss
 ss
(@ss)
Eminent Member
Joined: 1 year ago
Posts: 14
Topic starter  

What do you guys think of the following?

https://www.hackster.io/rahulkhanna/gps-voice-assistant-for-visually-impaired-people-71caa9#code


   
ReplyQuote
DaveE
(@davee)
Noble Member
Joined: 2 years ago
Posts: 796
 

Hi @maheen,

  As a third commenter, I am sorry to sorry to say that I tend to agree with the comments from Bill (@dronebot-workshop) and Will (@will) that a comprehensive and safe solution based on GPS is a very complex project.

Furthermore, the paper you reference points out  the fundamental problem with the current GPS system, namely that even on their simple test of recording the GPS at the same spot over a number of days, it wandered up to 28 metres from one day to the next. My own experience of GPS watches suggests that even in open rural situations, reflections from hills, trees, buildings, etc. can 'confuse' the position location function. Urban situations with nearby tall buildings, are even worse. As I am fortunate enough to see where I am going, these limitations are only a minor irritation, but for a visually impaired person relying on the system they could be fatal. In summary, GPS can be great to point you in the right general direction, but you must expect errors of at least 30 metres in 'favourable' places and a complete loss of signal/position location in many situations, including urban locations.

Of course there are 'tricks' like differential GPS to reduce the errors in 'open country' situations, but they cannot overcome errors caused by reflections and loss of signal caused by objects close to the 'mobile' user.

Personally, I suspect the system you would like will need to be much more complex than the paper you referred to considered. It could use GPS and magnetometer, but it will also 'observe' the surroundings of the person using a mix of cameras, radar, radio beacons and so on. To me, the problem appears to be at least comparable with fully 'hands-off' automatic car driving. Whilst the speed of someone walking is much slower than a car, the size of the pavement (sidewalk), if it exists, is much smaller, and in the UK (at least), pedestrians can cross roads, etc. at almost any place, so do not follow fixed paths, whilst cars are generally constrained to the road patterns.

---------

Perhaps you could consider turning the question around.

Start by finding out if visually impaired people are using any of the existing GPS direction finding agents like Google or TomTom to find their way around. If so, find out the advantages and problems they face. I guess that some people will tend to remain within a small familiar area. whilst others might be more adventurous. It is possible the needs of these different goups would also differ.  Then try to find a solution for one or more of the problems.

I think your ambition is admirable ... but you are trying to solve problems that are so far proving very challenging to some of the best funded research projects in the world. The trick maybe to adopt the results of some of that research, and adapt it to fit the particular needs of the group of people you have identified.

Good luck and enjoy your research! Dave


   
ss reacted
ReplyQuote
DaveE
(@davee)
Noble Member
Joined: 2 years ago
Posts: 796
 

Hi @maheen,

  My previous reply obviously only applied your GPS paper reference. Whilst I was typing it you added a reference ...  unfortunately it points to the GitHub code ... you have to page up a lot to read the description!

To read the article directly, try:

https://www.hackster.io/rahulkhanna/gps-voice-assistant-for-visually-impaired-people-71caa9

-----------

Unfortunately, the article does not make its achievements clear.

It includes simple one-word speech recognition. I suspect it will need to be trained for each individual user, though that might not be a major problem.

It includes GPS, but only one that outputs coordinates, so obviously that must be sent to Google Maps or similar.

I presume it is using Google Maps to provide directions -- I use that in the car, but I don't how good it is for pedestrians. Paradoxically, the only descrption I noticed was of a map screenshot.

I don't personally use any of the voice assistants like Alexa or Siri, but I would have assumed all of the above functionality is available from a decent smartphone already? Does it achieve anything new?

There is also a mention of ultrasonic hazard detector ... I can see that may be useful, and your previous reference supports this approach.  There is insufficient information to quickly ascertain if this particular project is useful in this respect. I suspect it would make a useful starting point, but maybe benefit from further testing and development.

Best wishes, Dave


   
ss reacted
ReplyQuote
ss
 ss
(@ss)
Eminent Member
Joined: 1 year ago
Posts: 14
Topic starter  

Thank you so much @davee. Truly appreciate your help!


   
ReplyQuote