|
CommieGIR posted:Rapsberry Pi Zero with Camera module and looping file. Ship it!
|
#
?
Jul 18, 2017 17:59
|
|
|
|
| # ? May 22, 2024 09:26 |
|
|
ratbert90 posted:Ship it! I was just joking. Look foreward to seeing what you got!
|
|
#
?
Jul 18, 2017 19:24
|
|
|
ratbert90 posted:Well I talked to my power/board layout guy. He is really suspicious of a super capacitor. He is going to do some research, but more than likely the power draw of the processor would be too much. A capacitor with enough power to support offloading videos of course isn't practical, but I have to assume the amount of power required to reach a safe shutdown state is orders of magnitude less than what's required for high speed wireless data transmission. Power's actually the main problem I've had with my Pi-based experiments, it's really easy to corrupt a SD card if you're doing things with it when the power cuts off. At the moment I just have a button wired to GPIO that triggers a software shutdown, but that's obviously not something you'd want on a final product.
|
|
#
?
Jul 18, 2017 20:20
|
|
|
wolrah posted:Without a battery or supercap how do you plan to handle clean shutdowns? My G1W with its failed battery wrecks the in progress file when it loses power. While normally my car doesn't shut off the outlet until 10 minutes after I lock the doors so its no big deal day-to-day, it's not hard to envision crash scenarios where it loses power during or immediately after the event. Why not just implement a clean shutdown protocol that closes out the file when it senses loss of connection or if the battery drops below 10%?
|
|
#
?
Jul 18, 2017 21:00
|
|
|
CommieGIR posted:Why not just implement a clean shutdown protocol that closes out the file when it senses loss of connection or if the battery drops below 10%? ratbert90 posted:- NO BATTERY. Consumers are dumb, batteries don't like extreme conditions, it will have USB-C anyways if you want to add a battery pack. So no battery in the basic design + no supercap = no power source to provide safe shutdown capabilities. I'm also not aware of any USB battery packs that provide status information, though presumably that would be relatively easy to implement using the HID UPS class, so at least at the moment anyone using an external battery as suggested would still have to have something extra for safe shutdown. As noted my current car provides accessory power for 10 minutes, so in my specific case the camera could see the vehicle has been powered off by a Bluetooth connected OBD2 dongle and then it would have 10 minutes to shut down at its leisure. That's not standard though. My other car shuts off accessory power the second you turn the key. There is no warning from the vehicle, it just has power one second and then doesn't the next. A commercially viable dashcam is going to have to be able to handle that situation.
|
|
#
?
Jul 18, 2017 21:35
|
|
|
I have handled garbage shutdowns before. The caveat is that you would have to turn on the dash cam again to recover the file.
|
|
#
?
Jul 18, 2017 21:40
|
|
|
Krakkles posted:Having gmail, I would be totally ok with it sending this without the "if" statement - I can ignore the emails until I need them. Trying to put GPS and such in a pet tag makes it rather large, which limits what kind of pets you can use it for. They've also got bluetooth and RFID versions which makes for smaller packaging. "Find my device", at least on my phone, DOES provide regular position updates as long as it's connected to a network; I can also force-update at will via the website as well as ring the device (which bypasses silence functions), put messages on the lock screen, or do a remote wipe.
|
|
#
?
Jul 18, 2017 21:44
|
|
|
MikeyTsi posted:Trying to put GPS and such in a pet tag makes it rather large, which limits what kind of pets you can use it for. They've also got bluetooth and RFID versions which makes for smaller packaging. Also I have a medium to large dog and therefore don't care about smaller animals.
|
|
#
?
Jul 18, 2017 22:22
|
|
|
A dirty shutdown shouldn't be that big of a deal with streaming data should it? You'll have a few munched frames but w/e?
|
|
#
?
Jul 19, 2017 22:52
|
|
|
I would not buy a dashcam that shut down instantly when the ignition turned off and could not be configured to run on battery power for at least a few minutes, in order to record things like traffic stops, what is happening after an accident, etc. If battery life is a problem due to heat/duty cycles/etc, make the battery user-replaceable.
|
|
#
?
Jul 20, 2017 00:08
|
|
|
ratbert90, can you provide some rough numbers for dimensions for packaging? Lens/sensor assembly, PCB, battery/supercap if any. I feel the urge to flex my newfound F360 newbie skillz.
|
|
#
?
Jul 20, 2017 08:00
|
|
|
I didn’t see mention of this in the dashcam thread, but don’t neglect the camera itself. Wi‐fi upload isn’t very useful if you can’t read the plates at night.
|
|
#
?
Jul 20, 2017 08:29
|
|
|
I like the idea of a base station that takes up to 4 cameras, that can handle the 11-14.5v ->5v conversion better than a crappy USB cigarette adapter surely?
|
|
#
?
Jul 20, 2017 08:55
|
|
|
bolind posted:ratbert90, can you provide some rough numbers for dimensions for packaging? Lens/sensor assembly, PCB, battery/supercap if any. I feel the urge to flex my newfound F360 newbie skillz. Not this early in the game! I can tell you that it won't be that big though.
|
|
#
?
Jul 20, 2017 13:30
|
|
|
I think we need to visit having a small battery, even if its just to power the shutdown cycle when it loses power input. That way, it can be significantly small, just enough to power the SD Card and the Micro-controller, but not enough to power the camera. However, I think we need to consider enough battery to power the camera anyways to be able to capture during and post incident in case the power is lost/destroyed/etc. in an accident.
|
|
#
?
Jul 20, 2017 13:49
|
|
|
CommieGIR posted:I think we need to visit having a small battery, even if its just to power the shutdown cycle when it loses power input. I am rethinking the battery situation. If I can keep the power draw under... 250 mAh@5V, which I think I can, then a small battery could possibly be sourced that would run the device for around 10 minutes.
|
|
#
?
Jul 20, 2017 14:08
|
|
|
There was a Lukas car UPS on Amazon that's no longer available but that I was considering for a while. Could you (possibly in lieu of an internal battery) offer some kind of ability to hook up non-switched power as well as a switched line? The camera could start with your car, then keep recording for maybe 5 minutes (or a configurable value) after the vehicle was cut off before stopping. This would defeat the heat factor plus allow for graceful shutdown, and I think I recall the Lukas model also had an internal battery on which the camera could run until it dropped too low, then the camera would run on the car battery until the car's battery dropped to whatever value was set (obviously north of 12.4V) and then it would finally lose power. Something like this would also allow for a 'bump mode' or 'parking mode', though I admit it's going a bit outside the initial idea. Also, and this one's mostly a pipe dream: allow the ability to have the camera save audio out to a separate file. I don't even care if it syncs, Mostly want road noise on demand but with video files that can be turned over without sound.
|
|
#
?
Jul 20, 2017 19:12
|
|
|
As someone who does camera work in this sort of market professionally, I think your biggest challenge will probably be getting anywhere near decent video quality. Automotive video is really hard, because there is just so much dynamic range in a scene. So, if it were me, I'd start with the core functionality of the thing -- recording video. Be careful about getting good optics, an at least decent image sensor (there are a handful of HDR technologies out there, and you'll have to choose which one you want), and most importantly, a good image signal processor of some kind. Without those things, you can have all the Wi-Fi you like, but nothing that will ever be worth uploading...
|
|
#
?
Jul 21, 2017 08:23
|
|
|
ratbert90 posted:I am rethinking the battery situation. If I can keep the power draw under... 250 mAh@5V, which I think I can, then a small battery could possibly be sourced that would run the device for around 10 minutes. There's a battery in every single car that's far better than anything you're going to be able to source that won't adversely impact the footprint. just make something that can patch in to the radio harness and you're set for power and battery feed. Another thing, if you make an RCA output you could patch in to the video feeds on in-dash units. Could possibly be used to add back-up camera functions and maybe parking assist.
|
|
#
?
Jul 21, 2017 10:01
|
|
|
MikeyTsi posted:There's a battery in every single car that's far better than anything you're going to be able to source that won't adversely impact the footprint. just make something that can patch in to the radio harness and you're set for power and battery feed. Usb-c. 
|
|
#
?
Jul 21, 2017 13:14
|
|
|
MikeyTsi posted:There's a battery in every single car that's far better than anything you're going to be able to source that won't adversely impact the footprint. just make something that can patch in to the radio harness and you're set for power and battery feed. I think the problem is in an accident it can become disconnected or the cars crash system shuts down the power so you don't start on fire.
|
|
#
?
Jul 21, 2017 22:00
|
|
|
I'm interested. And if you need to bounce ideas off someone or find parts, I'm another embedded guy, just spent 4 years doing automotive and aerospace electronic control modules. Do not rely on even a hardwired power adapter retaining power after an accident. Some cars, for example rear-battery BMWs, have a battery disconnect squib that is fired by the ACU if it detects certain crash signatures, typically side impacts that could crush the chassis enough to short the (unfused by necessity) starter power lead to ground inside the passenger compartment. I'd like to see many of the features already mentioned, and for mounting, go with either a standard 1/4-20 thread... Or ideally make it fit existing dash cam mounts. For example I've been very happy with the mounts for my A11x series dash cams, and extra mounts complete with 3M VHB tape are a whopping 7 to 9 dollars each on Amazon, which saves you having to have them molded. Focus on video quality first though. The rest of the features, as long as the hardware design is done with their support in mind, can come later. Another idea: I would like a ground side switched MOSFET output or two capable of driving an automotive relay coil under certain circumstances. Say, to turn on the hazards if it detects an impact followed by loss of power, and another that comes on in low light so I could turn on a grille mounted IR LED array. Doesn't even need to be expensive, all it'd really take is a few MGSF1N02LT1G FETs, polyfuses if you felt like protecting them, and whatever connector you want to expose them to the outside world. E: and if you want to save the money on parts, I'd be fine with using the rigol user upgrade mantra, just put the footprint on the board, the setting in an XML file somewhere, and I'll happily void my warranty to solder them on myself 
kastein fucked around with this message at 19:25 on Jul 26, 2017 |
|
#
?
Jul 26, 2017 19:16
|
|
|
Paging joshua_ ...
|
|
#
?
Jul 27, 2017 13:59
|
|
|
kastein posted:Do not rely on even a hardwired power adapter retaining power after an accident. Some cars, for example rear-battery BMWs, have a battery disconnect squib that is fired by the ACU if it detects certain crash signatures, typically side impacts that could crush the chassis enough to short the (unfused by necessity) starter power lead to ground inside the passenger compartment. I agree with all your points. Power loss should not kill your video immediately. No fan of the thread for mounting, but very much in favor of A11x series mounts. Focus on video quality and reliability. Like being able to switch external consumers, would suggest putting the output for that not directly into the cam but into the power cord down the line.
|
|
#
?
Aug 10, 2017 10:34
|
|
|
Mo Hawk posted:I agree with all your points. Power loss should not kill your video immediately. No fan of the thread for mounting, but very much in favor of A11x series mounts. Focus on video quality and reliability. Like being able to switch external consumers, would suggest putting the output for that not directly into the cam but into the power cord down the line. These are all fair points. The project isn't dead! I have market research, a business plan, hardware and software engineering specifications, a mechanical engineer, a hardware engineer, another software engineer, and myself (systems/firmware), a build system, and started actually writing code yesterday!
|
|
#
?
Aug 10, 2017 11:19
|
|
|
Do you have an outline of the features and specs for the initial version? Also, have you further considered doing your development in the open? While it takes some courage to publish your source code, schematics etc., if it is continually developed by enthusiasts it will be a major win for your platform.
|
|
#
?
Aug 10, 2017 12:46
|
|
|
Mo Hawk posted:Do you have an outline of the features and specs for the initial version? I do and I have. But I want to make money on this. 
|
|
#
?
Aug 10, 2017 14:07
|
|
|
sharkytm posted:Paging joshua_ ... I posted upthread :-)
|
|
#
?
Aug 11, 2017 07:11
|
|
|
So my partners and I came up with this plan in regards to open source: 1) Keep everything closed source until we are ready to go into production. 2) Once production starts, release everything on GitHub. Our big fear is the Chinees stealing our poo poo and immediately implementing everything into a cheap product. Well; it's not really a fear so much as a guarantee.
|
|
#
?
Aug 11, 2017 11:21
|
|
|
ratbert90 posted:So my partners and I came up with this plan in regards to open source: Pick an esoteric programming language, I'm sure you can slow them down at least a little bit. Hell, Rust has an okayish MSP430 port now.
|
|
#
?
Aug 11, 2017 17:30
|
|
|
This is an embedded system, I demand everything be written in assembly.
|
|
#
?
Aug 11, 2017 23:14
|
|
|
Leave a couple totally undocumented (to them) gotchas in the hardware design that if you don't correct them, it nukes the board upon powerup, never send them the PCB artwork, then do final board population (just the parts to make it not suicide) and firmware burn stateside? It'd slow them down at least.
|
|
#
?
Aug 11, 2017 23:19
|
|
|
ratbert90 posted:So my partners and I came up with this plan in regards to open source: I can personally guarantee that anything you can do, they can do cheaper and quicker. HOWEVER The good news is that your killer feature (instant social media uploads) is something that they cannot implement. Because the western SM sites are blocked in China and it is illegal to access them. Thus, they can never create something with FB links because not only is there no demand in their home market, they have no access to the APIs and even if they did, they could never tell anyone about it.
|
|
#
?
Aug 12, 2017 13:13
|
|
|
ratbert90 posted:I do and I have. Would you let us know about the expected features? That should not give anyone else a leg up.
|
|
#
?
Aug 13, 2017 16:21
|
|
|
spog posted:... ratbert90 posted:I do and I have. How are you financing this thing? I'm also considering doing a HW product tied with SaaS so I'm curious how you're approaching this.
|
|
#
?
Aug 13, 2017 17:33
|
|
|
ratbert90, any progress on this? Just curious.
|
|
#
?
Sep 18, 2017 07:50
|
|
|
bolind posted:ratbert90, any progress on this? Just curious. Working on it. I have a full-time job as well so it's a bit slower than I want. Right now I have: Inter-process communication Video up and going along with stop/pause/play/record Web communication to the lower layers A build system up and running A filing system structure A flashing process and a bunch of other smaller things.
|
|
#
?
Sep 18, 2017 19:04
|
|
|
I talked to a few friends last week. One of which has a dad who is a patent lawyer. They all said that the research I have done is good and that I should stop what I am doing and completely finish the business plan to get some series A funding. So that's what I am going to do.
|
|
#
?
Sep 29, 2017 11:33
|
|
|
Double reply: Any programmers who happen to be Linux guys might be pissy at this, but I like systemD. I will be using this for Interprocess communication: http://0pointer.net/blog/the-new-sd-bus-api-of-systemd.html The main idea here is: Web frontend -> user makes request -> nginx -> authorization daemon using pam using a web socket (c++) -> systemDbus -> Necessary back end application. This makes the entire project far more simple, as I only have to deal with a single socket, and dbus.
|
|
#
?
Sep 29, 2017 11:46
|
|
|
|
| # ? May 22, 2024 09:26 |
|
|
I think the anger towards systemd is mostly overblown, and dbus is probably okay. Since it's all one machine, there's likely no need to go crazy with something like a proper message queue or whatever. There is some fun stuff with GObject that could get you off the ground moderately faster, but it's probably not worth it. You could cut nginx out of the solution altogether since you probably don't need most of its features. Depending on what you're doing from the web frontend (since I assume you'll stream video and stuff like that using a different protocol) you might be able to just get by with a tiny node/express shim or the Go built-in webserver or something like that and bump the coordination/authentication logic into that part of the stack. Seat Safety Switch fucked around with this message at 19:26 on Sep 29, 2017 |
|
#
?
Sep 29, 2017 19:23
|
|