Inertia based underwater positioning system

[Wibble]

Would have thought that the speed of a diver relative to the speed of the water would make the measurements very difficult.

For example, jump in at the end of slack water and the tide reverses. If the diver's heading in a constant direction (whilst wagging around, pulsing on each fin stroke, etc.), and the tide slowly changes over 10 minutes or more in the opposite direction, then the inertial system has to be exceedingly sensitive and accurate, not to mention very high resolution to integrate all the propulsion "pulses" and subtle movements of the diver during the dive. The accuracy would very quickly degrade as that 'noise' happens.

An aircraft or submarine will be rather stable platforms with pretty constant thrust and a narrow window of speeds, thus the inertial navigation system would degrade slowly. Also on both platforms there's a lot of other inputs that may be measured ranging from altitude, compass bearing, propulsion thrust, radio beacons (flying), etc.

 

[Miyaru]

I've been playing with this concept and prototyped a unit on my dpv some years ago.

Initial positioning: gps. The signal will be gone after few centimeters of water.
Direction of movement: compas and gyro (using the same chip).
Next, you need a sensor to measure distance. For vertical distance, a pressure sensor is very accurate. For horizontal distance, a propeller with a magnet that gives a N and S pulse on each rotation.
Combined with the vertical speed and the direction supplied by the gyro, the horizontal vector can be calculated.

This only works when there is no current/flow.
Measuring the actual speed over the bottom can also be done in the same way as an optical mouse works: take a picture of the bottom, another picture several milliseconds later and compare the pixel shift.

With a mouse, it's easy since the distance to the surface is not changing. Transferring this concept underwater means that the distance to the bottom is another requirement for speed and distance calculation. The solution for that is lidar. A couple of years ago, only red lidar was available commercially off the shelve, and it turned out to be useless when more than 2m away from the bottom.
Red light should travel about 6m in crystal clear water, after 3 to 4 meters, the lidar won't read the light anymore.

Green light would work. NOAA uses green lidar for mapping the sea bed from an airplane, if I remember correctly they use 532nm. The automotive industry is developing green lidar for distance measuring to others cars in rain and fog. Once that turns into a commercially cheap product, it will be time to get my protype back in the water again....

 

[high_order1]

@high_order1 Do these orienteering groups have a website or anything? I'd love to give them a gander

This survey was a couple of years ago. I never found a makerspace or forum where they opined about how to solve their problems. It was either manufactured, or it was super home brewed. Discussion was more towards tactics/use.

Please do share if you find anything though!

I've been playing with this concept and prototyped a unit on my dpv some years ago.



Green light would work. NOAA uses green lidar for mapping the sea bed from an airplane, if I remember correctly they use 532nm. The automotive industry is developing green lidar for distance measuring to others cars in rain and fog. Once that turns into a commercially cheap product, it will be time to get my protype back in the water again....

Why not ultrasound?

Far as GPS fixes, I have seen several units with a bobber that unspools. When the operator needs to fix a position, or when the user interface tells them the system needs updating, they unspool the active antenna, it floats up, the magic occurs, and then it is reeled back in.

Perhaps not the best for a tank mount package, but just offering a way through

 

[tursiops]

What I don't understand about all these inertial navigation posts:
Accelerometers measure acceleration, which is relative to gravity (i.e., the bottom), not to the motion of the water in which one is swimming. Acceleration is the change in velocity, so is zero if one's speed or direction (relative to the bottom) is constant. The acceleration signal has to be twice integrated to give position, so an initial velocity and initial position are required. Where does that initial info come from? Also, if one is simply doing this for increments in time, each increment requires a known velocity and position to keep the calculation going. Where does all that info come from?
There is talk of using magnetometry, floating GPS units, known landmarks, etc.
There are some very good underwater navigation systems that use acoustics; why try and reinvent a square wheel?

 

[SouthernSharktoothDiver]

@tursiops At the start of your dive, your position and velocity would both be zero. Assuming a "perfect" IMU, that would be enough to know your position for the duration of the dive. Since the IMU is flawed, other metrics are considered to correct for the drift, such as magnetometry, GPS, or landmarks. Obviously, other solutions do exist, but there's always room for new innovation/ideas, and since this is a Scubaboard post (and thus consuming a negligible amount of resources) there's no harm in floating the concepts. If nothing else, an idea suggested here, as hardware improves in quality and cost, might actually be workable as a cheaper/more flexible solution than acoustic mapping.

On the off chance someone here actually has an idea I think I have the resources/knowledge to implement, I might attempt a rough prototype, but frankly that seems unlikely. I started this post with the belief that a "low drift" IMU existed and could be used for my application, and have since learned that what I thought was an IMU with low drift was actually an IMU with GPS correction, which requires both hardware and software well outside of my skillset and budget, so we're basically just floating ideas at this point.

 

[Mobulai]

floating ideas at this point.

Well.. you said it yourself:

IMU with low drift was actually an IMU with GPS correction

If the hypothetical use case is not in confined/overhead environment, a floating GPS umbilical? Something like the divevolk live stream demo but instead the tether is a GPS antenna
Highly impractical tho, I prefer the green MEMS lidars route, but still floating my idea

But I remember something we used for dGPS correction for autonomous driving ground truth data, one car has the fancy imu+dgps (egoV), and another (target/traffic participantV) has only IMU that could communicate over radio amd do its corrections based on the dGPS from EgoV
But these OxTS fancy boxes weren’t cheap for sure

 

[high_order1]

But I remember something we used for dGPS correction for autonomous driving ground truth data, one car has the fancy imu+dgps (egoV), and another (target/traffic participantV) has only IMU that could communicate over radio amd do its corrections based on the dGPS from EgoV
But these OxTS fancy boxes weren’t cheap for sure

Aerial high-precision drones use something similar called RTK. A local unit feeds corrections to the airframe.

I have no doubt that the genesis of a decent system lies in buying aerial drone parts. The magic would come in the software that interprets all the data and decides if it is true or not.

For me, I am going to continue looking for a way to use a propellered meter to provide some semblance of distance.

Wonder if you could use multiple fuel pressure sensors in a spherical array to collect pressure differentials as a way to sense water currents?

 

[Mobulai]

Aerial high-precision drones use something similar called RTK.

I think the automotive industry straight up stole that, the ground truth team experts called it that as well

Wonder if you could use multiple fuel pressure sensors in a spherical array to collect pressure differentials as a way to sense water currents?

a Bit unhinged (I love unhinged), just scared for the processor that has to process all that data coming in

How do submarines do that, the equivalent to Venturi tube thing in aviation? Or do they just skip that and depend on mapping and sonar?

 

[high_order1]

Pitot tube

They won't tell, but here's a book that may help:

Inventing accuracy : a historical sociology of nuclear missile guidance : MacKenzie, Donald A : Free Download, Borrow, and Streaming : Internet Archive

 

[Miyaru]

There are some very good underwater navigation systems that use acoustics; why try and reinvent a square wheel?

True!
Suex developed such a unit, which consists of a different nose for a dpv with all the sensors, and a screen for displaying waypoint information.
Tried it last year in open sea and it is very accurate.

The fun is in DIY.