360º always-on cloud-connected roof-mounted smart carcam

[Dashmellow]

Yeah you mentioned that before, so i will see what i can do when the weather get better again, have a lot of dashcam related projekts in the car.
And i am still on my winter rubber

The product was Turtle Wax ICE. Maybe you can find it in Denmark? If not, there's probably something similar available. It's a fairly durable polymer wax that doesn't leave any sort of white residue.

https://www.turtlewax.com/shop/products/turtle-wax-ice-paste-wax

 

[jokiin]

1. Easy and secure top of car camera
Bad idea, unless it's integrated into the car, 99% of the people will not be installing it on the outside of the car. Theft, damage, environmental factors.

without getting into the technical aspects I think this is one of the biggest challenges, not just the theft, damage etc but the willingness for people to put something on top of the vehicle, if it were targeted at commercial vehicles, Taxi, Police etc probably no issue but is the average guy in the street going to want to stick something on the top of their vehicle and then wire it inside, that would be a struggle I think, not a form factor I'd look at doing personally, I guess this idea would be pointless anywhere where you get a bit of snow in the winter as well

 

[Gibson99]

I've used liquid wax as well to polish the glass. Doesn't seem to matter what kind of wax you use as long as you do it once every few months to keep it fresh. The wipers wear it out a lot faster than the other windows and the paint.

 

[Razer]

Honestly, in my opinion this project is a non starter. This is too ambitious and or just not well thought out enough to consider. I promise I am not being a jerk, just trying to keep a realistic viewpoint in thinking this out.

Other than taxis and cops no one is going to stick a camera on the roof. Have you seen the roofs of average American SUVs and minivans? There is no way something mounted on the roof even 8 inches high will be able to see in front, behind, and the sides well enough. On a Suburban for example the roof is 10 foot long and it sits 6 foot (2 meters) plus high. Then it will be surrounded by a roof rack with rails also. Motorcycle comes by and smacks off your side view mirror you'll never see it. Someone hit your rear bumper at a light slightly? No proof at all, you'll never see it.

You would have to do multiple lenses and sensors as I work a lot with 360 cameras in the security realm and when you spread out even 5mp over a 360 degree scene it is useless for fine detail and zooming. It is good for a general overview but not nearly as good as you think it will be, and this is 5mp in a room indoors! Higher MP would be even worse low light performance. So already you need say 4 sensors and lenses, 4x the speed and processor and likely more than one SD card to record all these streams. With no other parts at all you are now 4x the price of high end dashcams. This is without solar, cellular, and other crazy features. People are not buying $99 dollar dashcams in droves, let alone a quality $200 unit. You will be talking $1000 easily.

Even here you'd find zero takers of a $1000 unknown dashcam that you have to mount to the roof.

I think it would be better to start with 360 in car cameras, the trancend 520 is a decent attempt to capture then taxi/uber/lyft type market but it has flaws and could be better itself. This would be a more reasonable area to try to start out even, and prove the concept without the problems of outdoor environmental concerns. Build something great here and you could expand. Benefit is there is a huge base of professional drivers that would be happy to get something like this! You need a market to sell to and a great product also. This would be a better start.

 

[kamkar]

Yeah i think the price will be high too, proberly in a range where the only one that can afford is the same guys that pay 20.000 dollars for a toilet seat and 10.000 for a hammer

For side view cameras i as i said would like some micro ones, here i am thinking one that can wedge down between the window glass and the seal in a door, this way you can have the window up or down as you please and the camera would be out of the way as much as you can expect.

For a camera that not only capture the road ahead but allso the cabin of the car i think there is a marked even for non professional drivers out there.
I wouldent mind a camera that allso can dokument that i am not on the phone while driving, somthing i am sure more and more ppl claim against the other part in a crash / dispute.

But first of all we need much more light sensetive color cameras, and as i said too i would not mind a FLIR camera too in the front of my car.

 

[DT MI]

Yeah i think the price will be high too, proberly in a range where the only one that can afford is the same guys that pay 20.000 dollars for a toilet seat and 10.000 for a hammer ...

You mean the US military?

 

[kamkar]

All those Black projekts have to be funded some way

 

[Webic]

Control unit connected to 4G LTE (and already started testing on 5G test rig)

As someone who works in the industry, you should think of the following:

• UE Catigory - How fast do you plan to upload videos? How frequently?
• Band Support - What operator will your equipment work on? What is their coverage map for LTE? Is this LTE only device?
• What M2M module vendor? Is that module already certified on a particular operator's network? Will you have to certify your device?
• Sim Card Support and form factor? Can the consumer pop-swap?
• What mechanism to get info from the device wirelessly? Push? Pull? How does the device check in to let the server know where it is?
• Will you subsidize your product cost by offering fleet management? Analytics to the operators / insurance companies about driving habits?

There's quite a bit to consider by throwing in "cellular" into the mix.

Also, 360 video typically means proprietary video recording formats. Meaning I always have to use your software and I have to hope you'll support it for the life of the product or the car it is installed in. I'm not a fan.

You seem like you're spreading your product over far too many features. Additionally, I would not buy anything that mounts to the exterior of my vehicle unless I can mount it under the car or in the grills.

 

[Secret Project...]

You are right there is drawbacks with current dashcams, reflections in the windscreen is one of the major ones as in the right unfortunate circumstances can make a good capture hard to get.
Offcourse you can fight this with a CPL filter or a dashmat, but both of those are not optimal but just what is available right now.

On the other hand having the camera on the outside it is also problematic, dirt can get on it, and just 1 splattered bug the wrong place and you are "recording blind"
On the windscreen you will see that big bug splattered out all over the place and can stop and get it off if the wipers cant handle it, with a camera you cant see you will drive on and not discover the problem.

Offcourse if you have a preview running on your phone you will see the bug on the camera on the roof, but i really cant be bothered with all that stuff running in my car and remembering to crank up the app and what not ( dont use apps in any form on my phone )

Cloud storage is also somthing that have been requested in here often, but it boil down to cell coverage, and even here in Denmark thats a small country with allright 3 & 4G coverage you can easy be in a place where you only have 1 or 2 G coverage.
Also the data generated by dashcams is pretty much around 400 MB for each 3 minutes of driving and this is just 1080/30, and that would eat up my 15 Gb dataplan in no time.
Offcourse i could get a mobile dataplan thats more Gb than my plan for my phone, but thats a whole other pricerange that i cant afford right now, and i think even if i was well and working i would not get such a plan, and i will ditch my cellphone when i can, ATM i am forced to have a phone but i dont like them.

Encrypted footage is also somthing ppl have asked for, this i think is somthing that can be done right here and now.

PT i am hoping for:

Much more light sesnetive sensors, what we have now is not good for anything but bright sunshine, as soon as there is grey skies above and you add a little speed you can forget all ambout capturing a small thing like a licenceplate.
Really the minimum exposure time for any of the frames in the footage must be 1:500 second or faster, and today dashcams drop to the slow 1:30 second resulting in massive motion blur in low light.

I would love a secondary camera in my windscreen and it should be flir, if it could warn about dangers on or along the road like in high end cars it would be nice.

And now we are at the fjew "smart" things i would like then ANPR i would like too, so i could enter the plate numbers of all undercover cop cars here so they cant catch me by suprice ( not that i really need it as i pretty much drive with the specs of the Danish traffic code )

Hidden main unit with a HDD or SSD for storage, transferring data by wifi or plug in USB key when i get home or need any data for my youtube channel.

For side and rear cameras i would like some ultra small ones, dont have to be really light sensetive as i see these angles as secondary, the front camera should be enuff to prove i am driving like i should so what ever go down must be blamed on the other guy.
Things rarely go to court here, for that to happen some one will have to die at least, and if thats not the case the insurance companies will duke it out among them self.

If there was to be any 3-4-5G in my dream camera it would just be for me to track the ( stolen ) car via the GPS in it. but i doubt any one would steal my small cheap ass Suzuki car.

Sorry if i am raining on your parade

Thanks for the very comprehensive feedback!

Let me see if I can address your points:

1. External mounted camera and the exposure to the weather/bugs/dirt - there would be a surface watching sensor (like a rain sensor for automatic windshield wipers) that would smart inform the user of debris interfering with recording

2. Cloud storage - this is handled intelligently. a/ no need to stream and record 100% of 360º at full res all of the time because the control unit will do the basics of analyzing the stream for object/delta detection when in motion and motion detection when at a standstill. It is planned to allow the user to always get elapsed time 360º at medium resolution as a lower priority uploaded stream. The controller will pass off the 'interesting' video segments to the cloud for storage and post storage analysis. So.... if there was a bear walking down your street, you would get a smartphone alert and video preview with the option to go 'live'. If your environment has this dynamic you can also 'mesh' with other units that you have connected with (lot of privacy here - of course) in order to create a 'neighbourhood watch' of notifications. The bear scenario did occur to me personally and I have video of it. I hooked up my test rig to an Ericcson test bed and I was getting 3.3 Gbits. It was amazing! I am building a little bit ahead of the curve.

3. Encryption - this is a core feature. What I'm playing with is moving the video that has 'interesting' metadata randomly in order that it becomes 'unsubpoena-able'

4. Light sensors - completely agree. Smart strobic, best-in-class light handling is called for, but what also is needed is a second set of sensors. Think thermal, license plate reflection tuned and IR distance sensors. HUD with FLIR would be really cool but I'm not working on that problem Volvo is (

)

5. License Plate recognition - http://www.openalpr.com is very interesting tbc

6. Hidden controller - this is pretty much the way you would do it and a way to store and offload video securely is a core feature, but it is still intended that the unit will have coverage 90% of the time. Without an active cellular data plan, yes it will unload when there is wifi

7. No side cameras - this is a single panorama camera. I will be able to show video samples soon of why this is the best approach

8. Car tracking - if the car has been stolen, tracking, telemetry, alerts and video will be available, with cell coverage and a data plan

Handling parades and rain are what I am trying to solve for!

Cheers,
JG

 

[Secret Project...]

During a rain storm how would the camera get rid of water being driven in front of the lens? Will it include a silicon windshield wiper blade? Will this include an in-car recording system? If cloud is not available how about recording to a storage device inside the car itself?

The cost of uploading constantly to the cloud is likely to be prohibitive. Look at the amount of data sent per hour. Then look at the cost of coverage for a month. i suspect that wifi or bluetooth to internal storage inside the car would make more sense.

I suspect with dirt, bugs, and rain the video would be subpar on a regular basis.

1. Hydrophobic lenses are part of the solution, the aerodynamic design would use the airstream to do part of the job, I have a test wiper system, but I want to avoid moving parts.

2. The controller is the in-car recording system, with the objective to get the video intelligently cloud stored asap.

3. The future is mobile internet. In America data-roaming between carriers is a big problem but it is being addressed. I was at the AT&T developers conference in Jan held in Las Vegas and I connected with the right people. I believe OnStar has solved for some of the connectivity issues at the low-bandwidth end.

4. Internal buffering is of course part of the design, with external monitoring (cloud to car pinging) letting the owner know of significant out-of-network issues occurring (such as parking in an underground carpark)

5. I have field tested various cameras in many different weather environments and the video results are uniformly good. The main issue is droplet spotting on hydrophobic surfaces.

 

[Secret Project...]

Could you include a couple of charging ports so that the on the top of the car cam would have the ability to charge cell phones? This would be a nice feature on a long drive.

No unnecessary wiring on top of the car. So no.

 

[Dashmellow]

The proposed technology certainly sounds intriguing but I can't see any way for the cost of this device or the data services to support it not to be entirely prohibitive.
Frankly, the whole thing sounds like the exact opposite of the KISS principle.

 

[Secret Project...]

Comments for each item below:




1. Easy and secure top of car camera
Bad idea, unless it's integrated into the car, 99% of the people will not be installing it on the outside of the car. Theft, damage, environmental factors.

2. 360º panoramic view
Interesting idea but ultimately falls short if implemented with a single lens and sensor, better to go with multiple cameras generating a 360 degree view.

3. 4K res
The more the better but needs to factor in if multiple sensors are used and if the hardware can handle processing.

4. H2O-phobic
Obviously a must, but with 24/7 outside, I don't know how long this protection/coating will last.

5. License plate and distance sensors
Purpose?

6. Powered by solar, wind, kinetic and maybe wireless transmission - Always on
Retarded idea if it's not a part of an overall vehicle system, meaning if the project is aftermarket, forget about it, only a wire run for power will work but then it'll require a hole in the car which 80% of the people will not do. Wind?-how many times do you need to be slapped? You can't transmit wireless power through the metal roof.

7. Connected to internal powered control unit
-

8. Control unit connected to 4G LTE (and already started testing on 5G test rig)
Stream all the video to the cloud, are you retarded? Do you want your bill to be $2K per month in data, or are you compressing everything into 480p?

9. Large encrypted video buffer that smart cloud purges
-

10. Smart object sensing
Purpose?

11. Partial or all 360º streaming with variable bit rates
Good luck with the data plan.

12. Chain of evidence digitally preserved and provable in court
-
13. Police can not get video off the unit or from the cloud. This is personal security video that YOU control.
-
14. Smart Alerts to your phone | Saves connected events - like initial casing by thief AND the act itself
-

I'd like to know:

1. Would you buy one?
There's a 0.5% probability that I would. ie NO

2. What price would you pay for the unit?
Proposed unit does not make sense and therefor a value cannot be assigned.

3. What price would you pay for monthly data service plan?
I would not, too many thing to pay for per month already.

4. Why you wouldn't buy one?
See all the reasons above.

5. Would you buy a unit through Kickstarter-like offering?
Not sure, I would need to trust the owner to deliver and have a proven history of deliverables. I am unlikely to pay for a dashcam, several hundred $ that might not get delivered.
I would not buy a novelty dashcam, dashcam is for security, and that's not something you pledge to buy without understanding the exact specifications and operation including all the bugs and downsides without people testing the unit first. With ak ickstarter you are pledging to buy something that no one tested and that doesn't exist outside of your imagination/lab.

Thanks for the feedback.

A couple of points:

- Most self-driving (the guys in DARPA discourage the use of autonomous because if a car was truly autonomous it might not want to drive anywhere!) cars will have additional sensors roof-mounted, so this is the future

- Multiple cameras creating a 360º view have lots of advantages - I agree

- The Nest used the power that was available (but never planned to power a computer) and the design objective here is no drilling and to grab whatever power there is. I have actually tried wireless power but it doesn't work on all roofs and it is very bad at power utilization. Lots of ideas were considered crazy initially

- You can get unlimited data plans for $50 a month. I am betting on creating a scalable video and data acquisition platform. I do think this is a 5G product that works in 4G

Thanks!
JG

 

[Secret Project...]

The proposed technology certainly sounds intriguing but I can't see any way for the cost of this device or the data services to support it not to be entirely prohibitive.
Frankly, the whole thing sounds like the exact opposite of the KISS principle.

Thanks for the feedback, but the current dashcam approach wouldn't have helped my father-in-law with his accident.

The accident occurred on a small Irish road while my father-in-law was crossing the road. He was 75% across when someone came from the right side and rear side-swiped him. This person was traveling 70 mph approx in a 50 mph zone. It was dark and damp. I went to take pictures the next day to see if I could prove their culpability, but no tyre marks were visible. It would have been unprovable without the cars present anyway. My father-in-law was the one who was deemed in the wrong, mainly because he had the best insurance and it doubled his premium for a couple of years going forward. Completely unfair and the other parties claimed whiplash also.

 

[Secret Project...]

I think roof mount is not the way to go. Flexible cameras that would just be strips on the windows or bumpers are the wave of the future
http://www.digitaltrends.com/photography/flexible-camera-lens-array/#:oXOd1Oimo02E8A

I also think cars will soon come with drones that follow you everywhere and just land on your roof to solar charge while you are parked.

Seriously, I like any advances in the dashcam market but it's a limited time before auto manufactures and insurers are including cameras and black boxes in all new cars.

Thanks for the comments.

- I like the idea but sheet cameras are still very far in the future (but not as far as the Enterprise D's phaser banks!)
- I am also interested in drones. Having a drone take off and land from the roof while underway and give you a navigational data is a very cool idea. #carcamroadmapitem
- I think car manufacturers are obsessing on self-driving tech and there is a significant opportunity in new carcam tech

 

[Secret Project...]

I have to say given that most manufacturers struggle to deliver single channel camera that functions well (high quality image day and night, GPS, parking mode) this project is too ambitious - there is no way you would be able to deliver all for a reasonable price.

Expect more! The price will be higher than normal, some people's car will merit the investment.

 

[Dashmellow]

Thanks for the feedback, but the current dashcam approach wouldn't have helped my father-in-law with his accident.

The accident occurred on a small Irish road while my father-in-law was crossing the road. He was 75% across when someone came from the right side and rear side-swiped him. This person was traveling 70 mph approx in a 50 mph zone. It was dark and damp. I went to take pictures the next day to see if I could prove their culpability, but no tyre marks were visible. It would have been unprovable without the cars present anyway. My father-in-law was the one who was deemed in the wrong, mainly because he had the best insurance and it doubled his premium for a couple of years going forward. Completely unfair and the other parties claimed whiplash also.

I agree completely about the need to cover all sides of one's vehicle. That's why I run four cameras at all times for 360° around my vehicle. (more than 360°actually as there is overlap).

See THIS thread for some discussion and examples regarding lateral (and rear) facing cameras.

I really like some of the ideas you are presenting but I believe that bringing such a product to market at a price point that will be viable at this point in time or even in the foreseeable future is unlikely considering the general dash cam buying public. Only a tiny elite portion of the dash cam buying public would be willing or able to afford what you are proposing, much less the projected 600 dollar (or more) per year data plan to support it.

The concept of a 360° coverage rooftop camera has been discussed here on DCT before. Perhaps concentrating on the camera technology alone would be more viable with the notion of adding the super advanced features at a future date as they become more practical?

 

[Secret Project...]

Honestly, in my opinion this project is a non starter. This is too ambitious and or just not well thought out enough to consider. I promise I am not being a jerk, just trying to keep a realistic viewpoint in thinking this out.

Other than taxis and cops no one is going to stick a camera on the roof. Have you seen the roofs of average American SUVs and minivans? There is no way something mounted on the roof even 8 inches high will be able to see in front, behind, and the sides well enough. On a Suburban for example the roof is 10 foot long and it sits 6 foot (2 meters) plus high. Then it will be surrounded by a roof rack with rails also. Motorcycle comes by and smacks off your side view mirror you'll never see it. Someone hit your rear bumper at a light slightly? No proof at all, you'll never see it.

You would have to do multiple lenses and sensors as I work a lot with 360 cameras in the security realm and when you spread out even 5mp over a 360 degree scene it is useless for fine detail and zooming. It is good for a general overview but not nearly as good as you think it will be, and this is 5mp in a room indoors! Higher MP would be even worse low light performance. So already you need say 4 sensors and lenses, 4x the speed and processor and likely more than one SD card to record all these streams. With no other parts at all you are now 4x the price of high end dashcams. This is without solar, cellular, and other crazy features. People are not buying $99 dollar dashcams in droves, let alone a quality $200 unit. You will be talking $1000 easily.

Even here you'd find zero takers of a $1000 unknown dashcam that you have to mount to the roof.

I think it would be better to start with 360 in car cameras, the trancend 520 is a decent attempt to capture then taxi/uber/lyft type market but it has flaws and could be better itself. This would be a more reasonable area to try to start out even, and prove the concept without the problems of outdoor environmental concerns. Build something great here and you could expand. Benefit is there is a huge base of professional drivers that would be happy to get something like this! You need a market to sell to and a great product also. This would be a better start.

Thanks for the considered comments!

My response to your valid points:

- Roof mounting - I completely agree, there are significant challenges. I am creating a statistical matrix of car/van/truck/convertible roof top configs. What is measured is unit height(s) (think dome+solar array or shark fin + solar array), placement, closest license plate viewing distance and 360º pano to ground data. I think this unit is applicable to 70% of the autos on the road and almost 90% of the personal watercraft.

- Impacts and damage are provable with accelerometer data and visual debris

- I agree with you on the uptake of dash cams and their price points. But that is because that approach has too many drawbacks. Innovation happens when you take a crappy experience and turn it into a craving experience. I am going to challenge the current model with different buying/usage models

- I agree with you that a $1,000 price point is easy to achieve with these features, but I also agree at that price point it won't sell. The job here is to innovate at a competitive price and give on-going value

- I have a list of professional drivers and fleet managers that are interested in 'the next level' of dashcam - so I agree they will more readily see the value.

- I am trying to empower the individual with personal surveillance - that is the core mission. As such I don't want to record conversations inside an auto. In this day and age; there are so many cameras pointed at us of which we have no control, a carcam allows us to have our part of the record within our control. Watching police interactions fair and unfair, accidents, extraneous events that we are called on to witness and to protect our property.

Thanks again!

 

[Secret Project...]

During a rain storm how would the camera get rid of water being driven in front of the lens? Will it include a silicon windshield wiper blade? Will this include an in-car recording system? If cloud is not available how about recording to a storage device inside the car itself?

The cost of uploading constantly to the cloud is likely to be prohibitive. Look at the amount of data sent per hour. Then look at the cost of coverage for a month. i suspect that wifi or bluetooth to internal storage inside the car would make more sense.

I suspect with dirt, bugs, and rain the video would be subpar on a regular basis.

I answered your questions in another response on this thread. Thanx!

 

[kamkar]

Your projekt do sound interesting and i for one would love to see where its going, but i have a hard time seeing how you can achive good low light / night performance with the sensors i assume you will be using ( mega/mega pixel ones )
But offcourse if you have a sensor that perform as good as say a Sony IMX185 then that would be about as sweet as can be ( for a price a regular human can pay )

And i do approve going high tech on tha ppl that do the same to you, for more or less valid reasons.
ATM it look like i have been named a crook by some US agency becuz i use the TOR browser once in a while.

Police cars here have ben fitted with phones now instead of the old radio systems, and they gotten encryption, so i have been thinking if it was possible to pick out a car like that in a crowd by its RF signature.
Not that i really have anything to hide from the police now ( decades ago it was a whole other matter ) but i do think if they go under cover and sneak up on me then i have the right to detect them sneaking up on me.

I dont really have anything to hide here, but when ppl for no reason lable me as a crook and listen in on me ( or what ever ) i cant just say like so many ppl do ( i have nothing to hide so you just log anything i do )

Dont like when ppl say that ( i have nothing to hide ) i feel ppl that say that is missing the point.