Blind and Visually Impaired Community
↑
10
↓
Self-driving human? (self.Blind)
submitted by spaceship4parakeet
I’ve been using a cane, GPS, and Soundscape as well as a sonar device and apps like Envision AI and Seeing AI, but can’t help but feel that with self-driving vehicle technology, there should be a way for an app or hardware to tell me everything around me using a camera and AI and not just GPS.
If a self-driving car can decide when to cross an intersection or slow down for a pedestrian, couldn’t there be something that tells me if bikes or runners are approaching quietly on the path that I’m about to turn on to? Couldn’t it tell me that there is a homeless man sleeping on the footpath ahead of me before I poke him with my cane? (Or worse was when I nearly tripped on a sand-colored nudist sleeping on the beach!) Perhaps it would warn me that there are three dogs playing at the shoreline that I can’t hear over the waves, but now one is making a b-line toward me. Perhaps it would tell me that I have a clear shot across the subway platform to the staircase 50 meters to my right and that the sign next to the stairs is for 1st Street. (GPS doesn’t do so well in buildings or underground.) I wouldn’t mind having to use a glasses-mounted camera and a computer unit larger than a phone for the needed processing power. Is there anything like that in the works?
If a self-driving car can decide when to cross an intersection or slow down for a pedestrian, couldn’t there be something that tells me if bikes or runners are approaching quietly on the path that I’m about to turn on to? Couldn’t it tell me that there is a homeless man sleeping on the footpath ahead of me before I poke him with my cane? (Or worse was when I nearly tripped on a sand-colored nudist sleeping on the beach!) Perhaps it would warn me that there are three dogs playing at the shoreline that I can’t hear over the waves, but now one is making a b-line toward me. Perhaps it would tell me that I have a clear shot across the subway platform to the staircase 50 meters to my right and that the sign next to the stairs is for 1st Street. (GPS doesn’t do so well in buildings or underground.) I wouldn’t mind having to use a glasses-mounted camera and a computer unit larger than a phone for the needed processing power. Is there anything like that in the works?
Rethunker 7 points 11m ago
My company is working on guidance and navigation. It's a huge, difficult, and costly problem to solve, but we're hammering away at it step by step. Though I can and will end up writing several replies over the next day or so that would likely amount to an essay, I want to ask a few quick questions first.
1. Would you be willing to get involved in testing?
2. Would you be available for a Zoom call to talk through this?
3. How much time could you devote to working on this?
And to follow up a few of your comments briefly before I need to run off for errands and meetings:
* Liability insurance is an issue with safety-related tech. My company has liability insurance just to cover one app, and even then the insurance agent has helped me by pointing out what I should and shouldn't say the app can do.
* Very, very few people are well suited to working on these problems. You've met your share of sighted people who make weird assumptions about blind people, and sighted people just like the ones you know are developing tech.
* Lots of money is required to prototype what can seem to be the simplest, smallest component of a system such as you're describing.
* Machine learning has issues, and these issues are very problematic for safety-related systems.
* Coordinating all the different features you mention is itself a very hard problem. That takes some explaining I'll have to save for later.
* Investors may support the idea of creating a guidance system, but they're not necessarily keen on putting money into it. The BVI community is comparatively small, and the ranks of BVI folks don't include many with deep pockets. Investors tend to put their money in companies that grow fast by selling stuff to people with money ready to spend. That's a gross oversimplification, but I can go into it more later.
* It'll be expensive. No two ways about that.
* Microsoft, Google, and Apple are unlikely to develop this tech, although they have the means to develop much of it. That takes some explaining becomes it's not necessarily obvious.
* When someone with the right background talks about this kind of thing with enthusiasm, it's hard to do so concisely. If you have 30 seconds to 3 minutes to convince someone that a need exists, that you have a solution, that there's money in developing that solution, and that investing money in the development is a good idea, well, that's hard to pull off. I've been hacking away at that problem for years, with partial success.
* The necessary hardware isn't quite there yet. My company has early access to some of the necessary hardware, but using early stage hardware also means dealing with early stage features. You spend more time struggling with someone else's prototype than you do building your own prototype on top of it.
So that's my answer for now, but now I have to get back to a full day of working on exactly these things.
1. Would you be willing to get involved in testing?
2. Would you be available for a Zoom call to talk through this?
3. How much time could you devote to working on this?
And to follow up a few of your comments briefly before I need to run off for errands and meetings:
* Liability insurance is an issue with safety-related tech. My company has liability insurance just to cover one app, and even then the insurance agent has helped me by pointing out what I should and shouldn't say the app can do.
* Very, very few people are well suited to working on these problems. You've met your share of sighted people who make weird assumptions about blind people, and sighted people just like the ones you know are developing tech.
* Lots of money is required to prototype what can seem to be the simplest, smallest component of a system such as you're describing.
* Machine learning has issues, and these issues are very problematic for safety-related systems.
* Coordinating all the different features you mention is itself a very hard problem. That takes some explaining I'll have to save for later.
* Investors may support the idea of creating a guidance system, but they're not necessarily keen on putting money into it. The BVI community is comparatively small, and the ranks of BVI folks don't include many with deep pockets. Investors tend to put their money in companies that grow fast by selling stuff to people with money ready to spend. That's a gross oversimplification, but I can go into it more later.
* It'll be expensive. No two ways about that.
* Microsoft, Google, and Apple are unlikely to develop this tech, although they have the means to develop much of it. That takes some explaining becomes it's not necessarily obvious.
* When someone with the right background talks about this kind of thing with enthusiasm, it's hard to do so concisely. If you have 30 seconds to 3 minutes to convince someone that a need exists, that you have a solution, that there's money in developing that solution, and that investing money in the development is a good idea, well, that's hard to pull off. I've been hacking away at that problem for years, with partial success.
* The necessary hardware isn't quite there yet. My company has early access to some of the necessary hardware, but using early stage hardware also means dealing with early stage features. You spend more time struggling with someone else's prototype than you do building your own prototype on top of it.
So that's my answer for now, but now I have to get back to a full day of working on exactly these things.
spaceship4parakeet [OP] 2 points 11m ago
Yes, I’d be happy to be involved. I can Zoom though my timezone is odd.
As for liability, GPS have warnings that it’s up to the user to make the decision whether or not to drive into a lake just because the GPS said to turn a sharp right. I realize it’s more complex than a simple disclaimer, but for the number of times I’ve heard of guide dogs pulling someone into the street, I have to think that most of us are prepared to accept that our lives are simply more dangerous than most, even with help. Soundscape doesn’t have a disclaimer and it would have directed me off a cliff and into the ocean in order to get me to the end of the jetty. I don’t blame it because I know it’s up to me to use my cane.
I also think even having something that says only “obstacle ahead,” “obstacle approaching from left” would be helpful enough. That alone would far surpass radar which is often foiled by the doplar effect.
It’s too bad Tesla can’t make a self-driving scooter or something in the meantime.
As for liability, GPS have warnings that it’s up to the user to make the decision whether or not to drive into a lake just because the GPS said to turn a sharp right. I realize it’s more complex than a simple disclaimer, but for the number of times I’ve heard of guide dogs pulling someone into the street, I have to think that most of us are prepared to accept that our lives are simply more dangerous than most, even with help. Soundscape doesn’t have a disclaimer and it would have directed me off a cliff and into the ocean in order to get me to the end of the jetty. I don’t blame it because I know it’s up to me to use my cane.
I also think even having something that says only “obstacle ahead,” “obstacle approaching from left” would be helpful enough. That alone would far surpass radar which is often foiled by the doplar effect.
It’s too bad Tesla can’t make a self-driving scooter or something in the meantime.
Rethunker 2 points 11m ago
All very good points, and well put! I’ve had team members tell me they’re willing to accept some risk because they have to do so anyway. But then the insurance guy tells a good horror story. Finding the balance will take some time.
I’ll send you a private message So that we can figure out a time for a Zoom call. Thanks!
I’ll send you a private message So that we can figure out a time for a Zoom call. Thanks!
smarthome_fan 2 points 11m ago
I would absolutely love to be involved in testing as well, and would love to at least have a Zoom call to discuss whether we would be a good fit working together. Please feel free to shoot me a PM.
Ok_Scar9881 2 points 11m ago
I would be interested in testing, if this were to become a thing.
Rethunker 2 points 11m ago
Would you be interested in answering questions on a Zoom call?
Although half my team is blind, we're always interested in reaching out to new people. Some people have very specific tasks they want to accomplish. Others may wait until there's a prototype or product they can test in person in their hometown. Out of necessity, our hardware testing takes place in the city where we're based.
One problem in developing a navigation system is determining what capabilities to prioritize. Each incremental bit of functionality we offer must be useful multiple times per week for as large a group of people as possible. If the tech isn't used often, it's hard to stay proficient with it, and/or it's easy to forget it in a drawer.
There's also the matter of what spaces to prioritize for navigation. For example, navigating airports can be miserable, **but** few people fly very often. Some people are interested in navigating shopping malls, and others don't care about malls.
Surveys and interviews help us determine what to prioritize so that we can develop what people need, but also make enough money to survive and grow. The cost to develop a basic navigation system is much, much higher than most people would assume. To make development affordable, to mitigate risk, and to ensure we're developing the right technology, we have to tackle one piece at a time.
So if there's some particular navigation task of interest to you, we'd like to discuss that. But if you're not certain, we can also talk through it.
Although half my team is blind, we're always interested in reaching out to new people. Some people have very specific tasks they want to accomplish. Others may wait until there's a prototype or product they can test in person in their hometown. Out of necessity, our hardware testing takes place in the city where we're based.
One problem in developing a navigation system is determining what capabilities to prioritize. Each incremental bit of functionality we offer must be useful multiple times per week for as large a group of people as possible. If the tech isn't used often, it's hard to stay proficient with it, and/or it's easy to forget it in a drawer.
There's also the matter of what spaces to prioritize for navigation. For example, navigating airports can be miserable, **but** few people fly very often. Some people are interested in navigating shopping malls, and others don't care about malls.
Surveys and interviews help us determine what to prioritize so that we can develop what people need, but also make enough money to survive and grow. The cost to develop a basic navigation system is much, much higher than most people would assume. To make development affordable, to mitigate risk, and to ensure we're developing the right technology, we have to tackle one piece at a time.
So if there's some particular navigation task of interest to you, we'd like to discuss that. But if you're not certain, we can also talk through it.
smarthome_fan 2 points 11m ago
Yes—technology should be quite a bit more advanced from where it is today. Even something like a robotic guide dog, where you'd still make all the decisions about when to cross etc. but the robot would perform traffic checks would be extremely helpful.
There are also issues like planned obsolescencewhere the tech you use and rely upon to literally go about your life, goes out of business and possibly stops working. Just look at what happened to Second Sight Medical Services.
There are also issues like planned obsolescencewhere the tech you use and rely upon to literally go about your life, goes out of business and possibly stops working. Just look at what happened to Second Sight Medical Services.
Rethunker 1 points 11m ago
We have fun plans for dealing with obsolescence. When I first started in the assistive tech industry I heard stories about people stranded with tech that lost support. An approach that's part of my company's strategy will address this, which is one of many interrelated problems.
Robotic guide dogs are something that has come up before. I've got plenty of thoughts on that, too. Surveys, interviews, and group meetings over the past few years have allowed us to review a bunch of potential applications. Some ideas lead to prototypes, others were put on the backburner. Now we're preparing to release a product and talking about what's next.
I sent a private message to you about following up with a chat. Thanks!
Robotic guide dogs are something that has come up before. I've got plenty of thoughts on that, too. Surveys, interviews, and group meetings over the past few years have allowed us to review a bunch of potential applications. Some ideas lead to prototypes, others were put on the backburner. Now we're preparing to release a product and talking about what's next.
I sent a private message to you about following up with a chat. Thanks!
retrolental_morose 2 points 11m ago
Read the intro of the novel Kill Decision by Daniel Suarez.
That's what we need.
That's what we need.
Rethunker 1 points 11m ago
>Kill Decision by Daniel Suarez
I'll check it out. In the meantime, if you had a drone, what's the least it would need to do for you to be worth having one? That is, if you spent $500, or $1000, or $5000, and the drone could only handle two or three tasks, reasonable well, what would those tasks be?
To bias your answer, I'd say that one of the tasks would likely not be related to safety. So using a drone to confirm when it's safe to cross a street probably won't happen any time soon. If someone were injured or killed while using a drone in this fashion, then even if the drone weren't at fault, lawsuits could very well kill the product and the company making it.
To my knowledge, even AIRA's sighted operators aren't allowed to guide someone across the street. The necessary liability insurance could be prohibitively expensive, assuming any insurance company would even offer a policy. Even for lower-risk applications getting liability insurance isn't a given.
That aside, I'm curious to know what one or two specific tasks you'd want a drone to handle. This has been a serious topic of discussion with my team over the past few years.
I'll check it out. In the meantime, if you had a drone, what's the least it would need to do for you to be worth having one? That is, if you spent $500, or $1000, or $5000, and the drone could only handle two or three tasks, reasonable well, what would those tasks be?
To bias your answer, I'd say that one of the tasks would likely not be related to safety. So using a drone to confirm when it's safe to cross a street probably won't happen any time soon. If someone were injured or killed while using a drone in this fashion, then even if the drone weren't at fault, lawsuits could very well kill the product and the company making it.
To my knowledge, even AIRA's sighted operators aren't allowed to guide someone across the street. The necessary liability insurance could be prohibitively expensive, assuming any insurance company would even offer a policy. Even for lower-risk applications getting liability insurance isn't a given.
That aside, I'm curious to know what one or two specific tasks you'd want a drone to handle. This has been a serious topic of discussion with my team over the past few years.
retrolental_morose 1 points 11m ago
ooh, good question.
One of the most irritating things for me is pathfinding. I can get off a bus and walk down a street quite happily, but say I need to get to a specific front door, cooler in a supermarket, counter at a post office etc, having a clear and consistent route to these things is impractical because of other people. In an ideal world, I'd have a wearable with strong images of the target and some sort of system to ensure I can get to it. Indication of clear line-of-sight at head-height is no guarantee of success walking but it'd be a start.
One of the most irritating things for me is pathfinding. I can get off a bus and walk down a street quite happily, but say I need to get to a specific front door, cooler in a supermarket, counter at a post office etc, having a clear and consistent route to these things is impractical because of other people. In an ideal world, I'd have a wearable with strong images of the target and some sort of system to ensure I can get to it. Indication of clear line-of-sight at head-height is no guarantee of success walking but it'd be a start.
SiriuslyGranger 1 points 11m ago
Sounds interesting. I ran in to a sleeping human on the sidewalk once and was like oh what’s that. Why would someone lay something over a sidewalk and it’s soft and squishy. So I ran my cane over him and even poked him a little. Then a man woke up and he’s like. Hey! What’s that for! I ran for it or started to and the guy was like are you okay.
So yeah, it may be nice.
So yeah, it may be nice.
Rethunker 1 points 11m ago
Detecting people is feasible under normal circumstances, but as your story indicates, one has to consider the unusual circumstances, too.
A system might have some use even if it only recognizes people in good lighting, up close, when the person is still and standing up, but making that system more useful gets difficult fast. The difficulty can be addressed with marketing, engineering, test, and customer support along with a boatload of cash.
A system might have some use even if it only recognizes people in good lighting, up close, when the person is still and standing up, but making that system more useful gets difficult fast. The difficulty can be addressed with marketing, engineering, test, and customer support along with a boatload of cash.
SiriuslyGranger 2 points 11m ago
Yes I agree.this was after dark one wonders if a light like headlights or flashlight option would aid in this.
Rethunker 1 points 11m ago
There are some problems with using lights or flashlights at night. If the light's on continuously, it would be a problem for other pedestrians and drivers.
Light outside the visible range can work, but it would still have to be very bright, cover a large area, etc.
There are solutions, but more exotic and more expensive.
Long story short, detecting and perhaps identifying people or other obstacles on the ground might work as a feature of a system with other capabilities, but likely the cost of the system and the cost of development would prevent it from being created as a stand-alone product.
Light outside the visible range can work, but it would still have to be very bright, cover a large area, etc.
There are solutions, but more exotic and more expensive.
Long story short, detecting and perhaps identifying people or other obstacles on the ground might work as a feature of a system with other capabilities, but likely the cost of the system and the cost of development would prevent it from being created as a stand-alone product.
SiriuslyGranger 1 points 11m ago
Very true. That’s interesting. Not sure what else would work though.
This nonprofit website is run by volunteers.
Please contribute if you can. Thank you!
Our mission is to provide everyone with access to large-
scale community websites for the good of humanity.
Without ads, without tracking, without greed.
scale community websites for the good of humanity.
Without ads, without tracking, without greed.
©2023 HumbleCat Inc • HumbleCat is a 501(c)3 nonprofit based in Michigan, USA.