h a l f b a k e r yIf ever there was a time we needed a bowlologist, it's now.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Voluntary driving monitor tutor with 2000% return on investment.
Stick on computer goes on back of rearview mirror; road camera and accelerometer records all your driving; the seat camera records the person in the seat; tor forwarding of all car data anonymizes; Software on cloud suggests education; the "no-power" ePaper 1.4-2 inch display that sticks onto the center of the steering wheel advises, and/or the speaker provides talk/text based tips. alibaba parts cost $1.74-0.44 | |
Reducing vehicle accidents is beneficial, fuel efficiency is beneficial, and reducing injuries is beneficial.
This very cheap thing you stick/clip on both sides of your car rearview mirror does all three. The cheapest version is 70 cents (from alibaba parts). It monitors your driving, and tells
you how you can improve your driving. two kinds of insurance companies like saving money on you, and you save money each year from fuel effciency.
The bold theory is that these numbers have some source[link] findability: 2010 2.2 million car accidents; 220 million cars, interpolating 1 car per 100 +/-20% has an accident. Many accidents are very mild, so each accident costs the vehicle insurer $200 on average. The health insurer, if there is one also averages $200 expenditure per accident, perhaps because only 1 per 20 vehicle accidents has a medical component but that one is near $4000 (made up number) of medical care.
100 stick-ons at the road, prevent, at most, 1 accident a year per hundred cars.
So, for each of the 100 cars it is worth $2.00 + $2.00 annually to insurers if it were to improve driving so much there cease to be any accidents, which it would not. Sadly, if accidents only went down 10% the stick-on is only worth about 40 cents per installation to combined insurers.
The driver also benefits. When they utilize the teaching function of the stick-on they learn to drive fuel efficiently and save 2% of $1000 worth of annual fuel; $20. Looking useful.
So, the 1 year value of the stick-on (I will call it a "Gnats", for [G]nag To Save) that monitors driving and teaches driving is $20.40/year. It costs only $1.74-44 cents to make from alibaba parts. To the gas consumer, at $1 a gnat, this is a 2000% return on investment.
The actual technology:
Front and back VGA resolution cameras for seat and road: 2 x 10c each
1 cent CPU
1 cent accelerometer
.5 cent rechargeable battery
5 cent photovoltaic that charges all day but just runs cameras for 1-2 hours driving/24 hours
10 cent Internet of things bluetooth module
Teaching technology:
.5 cent rechargeable battery
5 cent photovoltaic that charges all day but the gnat says 7 or fewer things an hour (a hint every 10 minutes might even be overdoing it)
1 cent piezospeaker
Expensive stuff, options:
2 cent CPU does tor on the Gnat, just like the tor browser anonymizing data reporting so the deep learning AI at the cloud can see a drunk person and nobody has to freak out. the AI just posts it's individualized updated advice profile, say once each 3 to 7 days, to driveadvice.com /anonuserLivelyTriwordNmemonic and the Gnat (or person at their PC, who remembers their easy three english word nmemonic) looks it up.
You can put a display sticker on the center of the steering wheel if you like. A 50 cent 1 inch square Epaper or 2" unlit LCD makes happy face emojiis and teaches with words. This makes a new text driver's educational hint once every 30 seconds-minute without distracting, unlike the piezo speaker.
720p resolution video sensor 90 cents (or less). Along with the anonymity of tor data communications to the cloud, is the tossup between give the interpretive cloud software great video, or bask in the zero-fault effects of doing it all with VGA computer resolution. What matters is the effectiveness of the advice.
So, as an actual thing, the efficacy of the drivers education at the Gnat would matter a lot. I figure drivers can learn, but I do not know the User interface they learn best with.
another reason it's called a Gnat is because although I strive to make it as nonpesky as possible ...it still makes a screen flutter when changing words and emojis. The speaker version, while cheaper just seems potentially too intense, perhaps if it only vocally teaches driving on car start and car destination.
Turning the car on or off could provide a brief educational communication chunk: "Last time You completely matched traffic flowspeed, but even at flowspeed there was an opportunity to see if you could drive under the speed limit about 10% more of the time."
"Nobody, no person, and no government can tell, but please, do not drive until you feel less lightheaded. I can't tell if you've had anything or not but a blood alcohol meter is just a dollar online, and could save your life."
Gnat stick-ons are applicable to 1.2 billion car drivers (2020) globally, and a different version at (imaginably) 1.2 billion moped/motorbikes in the developing world could also be possible.
[link]
|
| |
Computers in your car which monitor your driving and report back to
your insurance company are widely known to exist. I don’t see
much
difference between those and this idea. Many cars also already
report to the driver about how economically they’re driving. No
additional technology is needed, even if it comes from alibaba. |
|
| |
WKTE, aside from the cost you imagine. Beyond the cost of (some of) the components you also have to pay for: assembly, ule certification, software development, batteries or power cord, shipping, monitoring, antenna, wireless connection, and computer power. |
|
| |
A device without any wireless connection could be made but then it wouldn't have enough local processing power to be useful. A device using just a tiny battery and a solar panel can be imagined but it would barely have the power to run an LED strip and accelerometer. A dirt cheap clip-on g force meter would be fun, (and possibly useful for post-facto accident analysis) but that's not really what you're envisioning. |
|
| |
This is definitely WKTE! Oops |
|
| |
It is a little like saying "car computer" all the software ideas that could run on a car computer could be new actual inventions though. |
|
| |
Anybody have any software they think could run on on a Gnat? |
|
| |
Smart phones have accelerometers, can connection to servers (possibly the car itself), dual cameras, large capacity storage and memory and can run off a cigarette lighter. Oh and have a large communities backing IDEs. |
|
| |
So couldn't you phone it in, to a driving analytical server, with little lag 5G? |
|
| |
[WJT] I think all cars having voluntary internet connectivity is assumable. The Tesla vehicle has it now. Elon Musk's global wifi satellite network, already under construction, could do it too. |
|
| |
Doing it with phone data is fine too, People with unlimited data plans might go "woot!". It will all depend on if running the software is cheaper on a car computer or the cloud. |
|
| |
Making a guess, since storing a years driving is 40 cents (2020) on a flash drive, a car computer might do two pass interpretation of driving. Driving and commenting while live gets its own 2 cent CPU. Then, basking in the luxury of entire seconds of relative lack of hazard at stoplights and when parked at the library, the car's main computer could do exhaustive deep analysis that used literally quadrillions of times more computing cycles (30 seconds-3 minutes) to do deep analysis like with 2020 Deep neural network AI. |
|
| |
This provides relatively recent human connection to the words of "deeper value" coming from the Gnat. "You tend to speed in the mornings, just a heads-up", or "You used to use the "fastest commute" pathfinder app, but haven't this week. It saves like 10% of your gas and I can even optimize for good scenery" |
|
| |
So, either the software on the car's main PC-like computer, or cloud based AI could do deep long data analysis for things like that. |
|
| |
I never learned how to drive. If people use Gnats then it could be insurer prompted, so the cheapest one gets pushed. |
|
| |
If it is technology enthusiast, unIncentivized perhaps 60-90% of people might get Gnats. The Gnat could also be positioned as something good you get for your children, sort of like the advertising of life insurance or smoke alarms. |
|
| |
I didn't do the numbers, but a life insurer might literally give away a $1 Gnat. Nicely though the cheapest Gnat is a calculated 44 cents, so that's not a lot of "responsibility advertising" to put up with. |
|
| |
If I were going to improve the technology more it would be to make a developing world version where, at 2020 there are lots of motorbikes, and unless I misremember accident rates are more than 10 times higher, and accidents more serious. |
|
| |
Yup, there's no doubt: motorcycle drivers are more dangerous. |
|
| |
What would be the development, testing, manufacturing setup, and legallese costs of this 1$ Gnat? Does the dedicated Gnat really deserve maanufacture? We should really use the stuff we have fully before making more. Isn't that what software is for? |
|
| |
Insurance trip sensing apps for mobile phones are in the wild and work for all those wide ranging vehicles that are hardware centric. |
|
| |
If a Gnat using the seatview camera noticed people making out, kissing, or even having road sex It could say something naunced (and tested*) like "I hear fun pulled over on the roadside is actually twice as much fun ...as fun on the road!" |
|
| |
"kissing on the shoulder, while on the shoulder, saves lives!" |
|
| |
Thinking of teenagers and romantics here. There might or might not be a correlation between driving while intoxicated and making out/road sex in a car, so these folks having a wonderful time might actually be double dangerous as compared with mere DUI. The software's goal could be to continue the making out while reducing the driving risk by half (just intoxicated after finishing making out on the shoulder). |
|
| |
[WJT] One exciting thing about the Gnat is gains big with localization. Traffic regulations in a 10x accident zone like India or even China, and perhaps something called "defensive driving" might feature much more prominently at both car and scooter Gnats in the developing world. Localization is really big with a Gnat. |
|
| |
That suggests that the usual USA/Europe first product development could be developing world simultaneous. It is much cheaper to develop the Gnat, at an authentically working version, in a place like India or China or Brazil. The US has the right culture to develop the Gnat though. |
|
| |
Also, The very simple version might get 80% of the benefit from 20% of the effort. Just notice a mere 4-11 things at a rapidly developed version of the gnat. There are clever people out there (Like halfbakers), and someone like a halfbaker making up 11-24 ways to say 4-11 identical vocal cues (or display phrases) could be really fast to build. It is only 121 phrases to type, or recordings to make. |
|
| |
The part where it recognizes danger has, at just 5 (80/20 effect), some imaginable/inventable software forms. |
|
| |
*Don't speed*
Someone hundreds of times better than me at prgramming takes the suggestion "PC(car) OCR the entire JPG from a printscreen made once each second by the Gnat, Look for the OCR phrase "speed limit" and a number; use bluetooth to ask the actual car how fast it is going. Sophisticats use the accelerometer (1c) on the gnat. Are they Speeding Y/N? Update textarea at sticker, If they speed more than 10% of the time, let them know not more than once every 1/2 hour about it unless it's doublespeed" Actually the big challenge is knowing what to say to a speeder, and how often to speeder to get them to drive the limit. *note has ideas on that. |
|
| |
You're (very) intoxicated, call and have someone pick you up, Very intoxicated rivers might have cars that gently ~~~~ sway back and forth a little. <->; Two accelerometers + could sense this at the gnat (3 axis accelerometer 1c at alibaba). Or, again, JPG of freeze frame, only 20 FPS freezeframes, and look to see if [2 color monochrome version, light distribution or (photoshop fourier plugin/menu item graph) changes a certain %] The actual astute programmer, rather than referencing photoshop might edge-detect a centerline on the road. |
|
| |
Ooops! missed a signalling opportunity
20 FPS JPG printscreens (as if it were a PC) from video; loops starts at scanning BMP from topmost line then proceeds downward. This finds all light brighter than median sky brightness that is green, yellow, or red, and larger than say 80 pixels diameter in front of the car. It likes the idea that might be a traffic light, but of course takes continuous samples. OCR all the printscreens 20 times a second and earnestly prefer that it already be able to recognize "->" characters. |
|
| |
Remember, *the Gnat does not drive*, it advises. So if it is unsure about the meaning of a glowing green "-->" at one of those glyph-on-light fixtures it just ignores it. The cumulative improvement in driving comes in days, months or a year+. |
|
| |
Gnat computer sees and processes: (4 varieties of drivers seat person occluded; making out, food in lap, laptop on lap,). When you first get the Gnat it says "Hi, let's get started! Is there anything on your lap?" Y/N and gets a reference image. This one is actually a little harder because of what about a purse on a lap? What about printed T-shirts?, if (if?) they are public domain those functions from libraries that turn a 2D image to interpretive 3D could perhaps make "imitation parallax", and even I could note that virtual camera 1 - virtual camera 2 = bulk offset = not a T-shirt, is most likely an actual thing. |
|
| |
Software that spots anything that occludes the face could be from 1) knowing what a fleshtone is 2) making a BMP from a 20 FPS screengrab 3) using the reference photo, or perhaps better a photo from the very first moment the car was, turned on, that trip to find out what the usual size of the biggest blob of fleshtone is (200-400 pixels@720p) if that is >5-20% blocked its OK for the Gnat to say something nice, like, "I'd love to see your face while you drive!". It accumulates 20 FPS of this (20 times more likely to be right) getting a full second of samples becasue the person may just be glancing at the passenger seat of be in profile from fully looking out the side window. 60-100% facial bloackage suggests romantic effects, something really dangerous like reaching into the backseat at speed, or
maybe they are using a tablet while driving.
finding the best match to circles at a screengrab |
|
| |
Gnat and Phoning while driving. This is easy to spot with the seat facing camera because 20 FPS screengrab BMPs get processed to look for things that emit more light than anything else in the image. Phonescreens glow. It is likely they glow brighter than anything in the rearview mirror. Also, the 2D-3D library that can find imitation parallax from 3D-zing an image could figure out t ignore things as far away at the backseat plane or farther. |
|
| |
An interesting alternative to using video processing to spot a phone is monitoring audio. amazingly peculiar audio signals compared with voice? Are they at low volume? Perhaps the software can even checkwith the car to find out if its playing music? Do waveforms like normal talk alternate with the low volume tinny stuff? It might be a phone conversation. |
|
| |
Summary: The 80:20 effect make-believe suggests that a gnat that starts with 4-11 things could generate 80% of the accident reduction benefit. My Programming skills are horrible, no, back that up, I no longer even know if !#/usr/bin/perl starts off a perl program. But in 3 months I could write a couple of these. Some halfbakers would take 2-3 days each. But fortuntely |
|
| |
Pro-$ benefit Gnat is very easy:
Gnat say/writes, "Hey! that gas station coming up has cheaper gas than 9 out of ten gas stations! around town Saves $2 on a fill up compared with your usual!" |
|
| |
"I talked with your car, oil changes every 40-60,000 miles keep it all smooth and increase resale value about $190. Cheapest non-coupon oil change in town is at this place coming up on your right." |
|
| |
Gnat only needs to know GPS of where it is, which is queryable at perhaps almost every new car, and have a premade, hardcoded list of 20 cheapest gas stations from a single snapshot that "gasbuddy.com" took sometime. It stsrts with data, and, if at all possible (Hey that IoT chip is 10c) it gets updates. Same thing with oil changes, what the internet knows about where to re-vend a car. |
|
| |
The luxury of mean, median and mode of * ratings is available to the Gnat at Gnat Headquarters. There they webscrape all the reviews from the top 700 largest municipalities gas/oil/repair/vend/autoparts store and then send out updates to Gnats via the 10c bluetooth that talks with either the person's car or the person's phone. Perhaps you get a text once every 3 months that says, "You've got a Gnat but we can't reach it! tap the URL (at the text) next time you are in the car to update your Gnat! Most people save $20-60 a year on gas and all things car. (Pretty unobtrusive) |
|
| |
Some other Gnat money savers:
"You are just one block from free parking" |
|
| |
"cleaning out the back of your truck or car saves weight and saves Gas. If you left 2 Kg/5Lb of stuff back there that's $N (YMMV) a year. |
|
| |
Studies show shorter commutes are linked to greater happiness with life. All things being equal, if you halve your commute you'll be happier. That's above and beyond saving $499 in gas a year" |
|
| |
"did you now higher grades come with cheaper car insurance? Study, Get that GPA up and if we may be so bold, split the savings between the student and the parent."
Voice recognition:
"Are you thinking about selling your car?
(((person says something)))
I haven't looked it up but of the 40-100 places in town to do that they get rated from 2.5-5 starts on Yelp; check out yelp before you sell your car." |
|
| |
"Hey I've got more ideas on selling your car, Just like you'd pay more to go to mechanic in the poshes of all the mini-cities in the vastopolis metropolitan area, selling your car in Santa Barbara might pay 7-14% more than selling it in San Luis Obispo"; an 80 mile drive picks up $700-1400 of money in your pocket." |
|
| |
These are spurious numbers, but you get the drift of some gnat software opportunities. |
|
| |
There’s a money-making opportunity here, in that people would pay
to not have one of these in their car |
|
| |