Inertia based underwater positioning system
- Thread starter SouthernSharktoothDiver
- Start date
Please register or login
Welcome to ScubaBoard, the world's largest scuba diving community. Registration is not required to read the forums, but we encourage you to join. Joining has its benefits and enables you to participate in the discussions.
Benefits of registering include
- Ability to post and comment on topics and discussions.
- A Free photo gallery to share your dive photos with the world.
- You can make this box go away
Shore dive entry point, at the anchor/base of the tie in, any known point on the bottom. Much depends on the OPs goal... I got the take that it is meant as a log for after the fact, not real time nav... In which case you wouldn't even need to initialize at a stop, you could have time stamped start and stop GPS points and rectify for initial motion after the fact.
OP
SouthernSharktoothDiver
Contributor
@James79 Any idea how I could mitigate the sensor drift, either by adding more inertia sensors or introducing another form of sensor for comparison. I'd be ok with it needing calibration on start and stop, though ideally I'd eventually have a system that works in real time. Of course, if the tech can't do it, it can't do it. Though I'd love to get my hands on a drift free IMU. It'd turn this into a probem an 11th grader could solve.
@SouthernSharktoothDiver, search the interweb for an intro to Kalman filters to fill in the basics: many intros start with navigation as the first example. An IMU requires periodic independent, external sample data to correct/control/limit the accumulation of sample error (i.e. drift).
I've sorta/kinda napkin-designed/built something like you're asking about - if I understand your use-case correctly. What I have/had in mind is a dive track recorder to produce a 3d track of the dive that can be added to the dive log after the fact, and used as the basis for a better dive site map. A system for safe reliable navigation during the dive is WAY more - not conceptually, but costwise and risk of use. So, not an INU like @James79 is referring to; just a logger using an IMU and whatever external data inputs are possible.
My cobbled-together prototype had significant limitations/requirements - notably:
1. require some underwater structure "in view" (eg, bottom, coral wall, wreck, etc.) at least "most" of the time.
2. (really a corollary of point 1): require relatively good visibility.
3. require stable (no, obviously not fixed) mounting w.r.t the environment.
In the theme of "if you have a hammer everything looks like a nail", what I did was add a laser and camera to detect travel (distance and direction) against stationary structure (eg bottom). Why the "hammer/nail" analogy? I spent a lot of my career implementing, later designing, later architecting both passive and active vision systems for various 3-d measure/profile/track/target tasks. I tried to interest one employer in the project but the response (as expected) was "a cool idea is neither a product nor a business plan."
Anyway, in use, the laser/camera data, pressure (depth), and the inertial data from the IMU were all logged, along with GPS start/stop location. There was no processing on board (not budget possible as a diy project at the time). Kalman filter based processing was done offline on a desktop computer. These days an embedded system might be able to do the job in real time for a navigation system, but that's another level.
My one and only prototype seemed to generate nice data in shallow super-easy conditions, but the first (only) test with any challenges resulted in:
1. A leak in the housing about 1 minute after reaching planned bottom (12m).
2. Shorted and sparking/smoking electronics as soon as the salt water made it to the circuit board.
3. A brief fire (nice thing about a housing underwater - very limited O2).
4. No data.
I think I still have before/after pictures of the unit. It's kinda funny.
So bottom line, I think a system like that is possible, and I think about a Mk2 once in a while now that I'm retired. Given the giant advances in multiview photogrammetry it can (and should) be done without the laser. Of course, any other method that gets periodic distance and direction travelled independent of the IMU should also in principle be possible.
I've sorta/kinda napkin-designed/built something like you're asking about - if I understand your use-case correctly. What I have/had in mind is a dive track recorder to produce a 3d track of the dive that can be added to the dive log after the fact, and used as the basis for a better dive site map. A system for safe reliable navigation during the dive is WAY more - not conceptually, but costwise and risk of use. So, not an INU like @James79 is referring to; just a logger using an IMU and whatever external data inputs are possible.
My cobbled-together prototype had significant limitations/requirements - notably:
1. require some underwater structure "in view" (eg, bottom, coral wall, wreck, etc.) at least "most" of the time.
2. (really a corollary of point 1): require relatively good visibility.
3. require stable (no, obviously not fixed) mounting w.r.t the environment.
In the theme of "if you have a hammer everything looks like a nail", what I did was add a laser and camera to detect travel (distance and direction) against stationary structure (eg bottom). Why the "hammer/nail" analogy? I spent a lot of my career implementing, later designing, later architecting both passive and active vision systems for various 3-d measure/profile/track/target tasks. I tried to interest one employer in the project but the response (as expected) was "a cool idea is neither a product nor a business plan."
Anyway, in use, the laser/camera data, pressure (depth), and the inertial data from the IMU were all logged, along with GPS start/stop location. There was no processing on board (not budget possible as a diy project at the time). Kalman filter based processing was done offline on a desktop computer. These days an embedded system might be able to do the job in real time for a navigation system, but that's another level.
My one and only prototype seemed to generate nice data in shallow super-easy conditions, but the first (only) test with any challenges resulted in:
1. A leak in the housing about 1 minute after reaching planned bottom (12m).
2. Shorted and sparking/smoking electronics as soon as the salt water made it to the circuit board.
3. A brief fire (nice thing about a housing underwater - very limited O2).
4. No data.
I think I still have before/after pictures of the unit. It's kinda funny.
So bottom line, I think a system like that is possible, and I think about a Mk2 once in a while now that I'm retired. Given the giant advances in multiview photogrammetry it can (and should) be done without the laser. Of course, any other method that gets periodic distance and direction travelled independent of the IMU should also in principle be possible.
OP
SouthernSharktoothDiver
Contributor
@TooCold Ok, this is actually amazing, and props to you for doing it. Shame the prototype burned, but if you ever did make a 2.0, I'd love to hear about it. For my usage case, sadly, the vis is lousy enough that even with lights, there's no way the system could sight in on any landmarks. I did, however, have two thoughts:
First, hypothetically, given the current advances in fishfinder/recreational grade sonar tech, you could probably scan the bottom of a shallow dive site (and maybe a deep one, not sure on the range) to sufficient degree to find and isolate your landmarks. Sort of a survey of the bottom, which you could then enter into your mapping system. Make a quick pass over the site in the boat and even a brand new dive site could be ready to go by the time the code's done compiling.
Second thought: Hypothetically, how feasible do you think a system of magnetic landmarks would be? Like, imagine I dropped four strong magnets (like, the kind used in magnet fishing, or, if we're imagining this is a commercial product for people with a large budget, some form of electromagnetic beacon) at known locations around a site. Could these then be used as points of reference at a distance use a magnetometer? I'm unfamiliar with current state of that tech, and as I said in an above post, I was never this kind of engineer, so I'm flying blind as you might say.
First, hypothetically, given the current advances in fishfinder/recreational grade sonar tech, you could probably scan the bottom of a shallow dive site (and maybe a deep one, not sure on the range) to sufficient degree to find and isolate your landmarks. Sort of a survey of the bottom, which you could then enter into your mapping system. Make a quick pass over the site in the boat and even a brand new dive site could be ready to go by the time the code's done compiling.
Second thought: Hypothetically, how feasible do you think a system of magnetic landmarks would be? Like, imagine I dropped four strong magnets (like, the kind used in magnet fishing, or, if we're imagining this is a commercial product for people with a large budget, some form of electromagnetic beacon) at known locations around a site. Could these then be used as points of reference at a distance use a magnetometer? I'm unfamiliar with current state of that tech, and as I said in an above post, I was never this kind of engineer, so I'm flying blind as you might say.
I hear you. Same for me with local diving most of the time. As a buddy says about our diving, "Beyond 10 m every dive is a night dive". So my system would have needed godawful spotlights, and who wants to dive (or be around a diver) like that? I was looking more for use tropical diving.
Very true. But I don't see a way to link the sonar map data to the tracker's inertial data while on the dive, unless it was manual like "I'm passing a car in the mud on the bottom; time to enter a waypoint on the tracker." I think you'd be better off (i.e. simpler, easier, cheaper, faster to have a usable system) just using your sonar map and tracing your likely route based on traditional notes/sketches/landmarks during the dive. Also, I really didn't like needing a laser, and I personally wouldn't want to use sonar or any active system. There is research showing disturbance to marine life including mammals, due to underwater ultrasonics, depending on the frequencies involved. A laser powerful enough to reach a more distant bottom is a danger to nearby eyes including humans. Say I got whacked by a curious seal and lost my mask - what's my orientation and where's my laser pointed the whole time I'm waving around hoping to catch the mask as it floats away, or getting out my spare?
Never thought of that. In principle it would work. Off the top of my head triangulation using a pair of magnetometers on a wide beam would provide samples of each magnet's range and direction, within detection distance, to augment the IMU. System calibration would be a bitch since you'd ideally have to have the magnetometers in fixed position relative to the metal in your gear. Think calibration of ship's compasses. If there are no (practical, political, regulatory, environmental) barriers to dropping landmarks: again, simpler would be to use highly visible landmarks and manually waypoint them during the dive as you get to them. Say a bright slightly boyant float tethered 1 m off the bottom to a smallish weight. Easy to recover, and the system doesn't need much beyond the IMU, and inputs for depth and maybe compass heading.
high_order1
Contributor
Interesting thought experiment.
I researched the concept a little bit, and found a sport called underwater orienteering. They make some very novel and 'garage level' systems to participate.
As far as IMU systems, my aerial drone has one. They are not very expensive, and fairly well hashed out, I am told. Not ring gyro precision, but using essentially a barometer and a compass and talkback to the control for corrective inputs.
Subs use IMU. I am too busy poking the coners to learn about nukes, but I seem to recall that the classified systems need occasional fixes based on certain things.
Lastly, I would not concern myself with making it waterproof. There are people that do amazing things there (I follow a university that makes ghetto DIY fabulous sensors, the stuff they do with US big box store products is impressive.
make it work, someone else will figure out how to keep it from being a box of fire like the other guy (realizing it was on fire, then remembering being surrounded by extinguisher fluid)
I still want to make an orienteering board, but I have bigger fish to fry with my air compressor and hookah sources and rigging gear to my facemask and
I researched the concept a little bit, and found a sport called underwater orienteering. They make some very novel and 'garage level' systems to participate.
As far as IMU systems, my aerial drone has one. They are not very expensive, and fairly well hashed out, I am told. Not ring gyro precision, but using essentially a barometer and a compass and talkback to the control for corrective inputs.
Subs use IMU. I am too busy poking the coners to learn about nukes, but I seem to recall that the classified systems need occasional fixes based on certain things.
Lastly, I would not concern myself with making it waterproof. There are people that do amazing things there (I follow a university that makes ghetto DIY fabulous sensors, the stuff they do with US big box store products is impressive.
make it work, someone else will figure out how to keep it from being a box of fire like the other guy (realizing it was on fire, then remembering being surrounded by extinguisher fluid)
I still want to make an orienteering board, but I have bigger fish to fry with my air compressor and hookah sources and rigging gear to my facemask and
high_order1
Contributor
realized I had some pics I cribbed:
OP
SouthernSharktoothDiver
Contributor
@TooCold You're absolute correct about most of what I'm floating being impractical, but hey, if I wanted a practical, simple way to navigate, it'd be as easy as dropping weighted strobes (perhaps with numbered tags) in a grid pattern and using those for visual markers. "Oh yeah, i found this one near beacon 22-B." And that sort of thing.
I'd love to hear more about the calibration reguirements for the magnetic guidance system. If I could make that work, even theoretically, it might well actually be practical. You'd obviously recover the magnets post-dive, and if you made the final product use electromagnets, you could switch them off at the end to avoid accidentally sticking them to side of a shipwreck forever.
@high_order1 Do these orienteering groups have a website or anything? I'd love to give them a gander
I'd love to hear more about the calibration reguirements for the magnetic guidance system. If I could make that work, even theoretically, it might well actually be practical. You'd obviously recover the magnets post-dive, and if you made the final product use electromagnets, you could switch them off at the end to avoid accidentally sticking them to side of a shipwreck forever.
@high_order1 Do these orienteering groups have a website or anything? I'd love to give them a gander
Sorry about the late reply. I'm not a software guy, I'm an instructor pilot that flies an aircraft using INUs with GPS aiding (stupid accurate inertial sensors with frequent GPS updating to keep it on track).
So my first thought on it would be to run 4 cheap IMU sensors in a purely data logging mode to start (a la @TooCold ), and take them for a walk around a high school track. Then use desktop software to mess with data filters, averaging the 3 closest (drop the data from the furthest "out of common range" at any point) kind of stuff. See if that can get to a filter that gets it to map pretty close to the track you walked. If that can pan out, then a simple box with an orient north to and push the initialize button, and time stamp button for any GPS fixes you can pull at the surface for data massaging after. More or less building up to recreating an INU with aiding, but the updates being manual and using redundant sensors and voting logic in place of large and stupid expensive ring laser gyros.
I kind of envision the end state as a harness mount box (less movement and therefore sensor drift than wrist mounting) with buttons for initialize, markpoint (i.e. your neat find on the seafloor being marked in the data stream to locate in the log), and an update position button (gps, or your magnet based Loran-esque system).
I'd do topside data log/desktop computing testing, then known dive site data logging/desktop compute testing, then see if the software you end up with can reasonably run on an SBC (rpi or similar board).
Of course, if it pans out then a "point me to the initialization point" pointer on the screen would be a good first "real time" use implementation.
Respectfully,
James
Similar threads
