Hello - I’ve looked for low-cost DVL solutions to merge into an INS for an AUV mapping project. Curious if there is a project out there I haven’t found, or a group of people that are looking to bring something to market?
Open source solutions don’t seem to have anything on it. Large corps have extremely costly solutions. Is there anything in the hobby/light industry space that’s pushing some of the big players out?
Thanks!
@jwalser thanks for the addition here. Are there any other clever sensors that could serve the same sort of correction to underwater position? Particularly anything that uses cheap sensors.
Thanks for any input!
You might be able to DIY something rudimentary with the ping echosounder: https://www.bluerobotics.com/store/sensors-sonars-cameras/sonar/ping-sonar-r2-rp/
As far as getting an absolute position underwater the most affordable off the shelf option is the Water Linked system.
See also: ROV and AUV Localization an Introduction
@jwalser thanks so much for providing the information.
The Water-Linked System is not an option for the system I’m working on. So far - my research shows that the only way to get any sort of accurate undersea positioning is with a DVL. Of course - you’re already aware of the financial challenges.
I’d be interested to see if making some sort of addition to ArduSub open source software would gather the right people and current top-talent to work on making a low-cost DVL with some off the shelf components (like your SBES).
For efforts working to map the ocean, I think this is a needed advancement to do it at scale and make enough vehicles to do it.
Initial thoughts?
My mind wanders to finding an echosounder that is powerful (and energy efficient enough) to do this 
Do you guys know anyone else who’s looking into this?
A DVL (Doppler Velocity Log) is not a positioning system, it works in conjunction with a LBL (Long Baseline) or a SBL (Short Baseline) or USBL (Ultra-short Base Line) navigational system, the control computer needs input from all in order to get you what you need. The water linked system would be considered a SBL. LBL is by far the most accurate and also the most expensive approach. USBL, the most common, will get you into centimeter accuracy (+/- 30). A DVL is basically a “smart” current meter, it allows the vehicle to automatically adjust station hold capabilities. But you still need the tracking system to make it really smart.
ArduSub is an rov control system, and it has generic (NMEA) support for any positioning system (LBL, USBL etc). There is not any support for dvl currently.
I don’t know any community efforts to build a DVL, but I have seen talk about it along the lines of what you mentioned here. It is a huge amount of work and investment.
Nortec have lent us a DVL1000 to integrate into a BR2. Are far as DVL’s go, it’s about the cheapest on the market, but would still be out of reach of the average hobbyist.
I’ll keep you posted if we get good results.
Indeed, there isn’t a low cost DVL yet. There are a few reasons: cost of components like high end ADCs/FPGAs, multiple potted transducers (think 4x the Ping sonar just for the head), but also the fact someone had to hire a lot of costly experts to develope a product with a steep learning curve and a small customer base.
All that said, I think it is feasible to get out of the $15k+ range and into the $5k range. Maybe lower if it’s a more analog based system and leave “bottom lock” into the programmer integrating it with the INS data.
Getting to your immediate need, Craig is right. If you need absolute position, it’s probably WaterLinked SBL or something like a Micron USBL coupled with your INS. If your flight path can afford it, surface and re-zero your AUV using a GPS antenna just above the water. If you have the budget, a current meter can help you INS.
I have flow work class ROVs for a decade and DVLs are handy, but they are only so accurate and are generally used to “filter” USBL data in conjunction with an INS. If you loose bottom lock in soft silt or a school of fish fly under the AUV, you can loose 10 meters in a moment and not even know it. That’s where INS and knowing you absolute position (via GPS and LBL/SBL/USBL) come into play.
The cheapest one I know of is this one: LinkQuest
In the range of 10k usd if I remember correctly.
Hi Anthony -
While not a dvl, depending on your application some off-the shelf machine vision cameras may work? The Open MV M7 and H7 have a great set of exampe scripts, including some for optical flow that might work if pointed at the bottom, providing x / y displacement? Obviously limited by water visibility and distance to the bottom…
I’ve thought at length about this. I keep coming back to power usage of a camera, and the lights required to see something at depth (>300m). Seems like a hinderance to having this work in practice. @tony-white do you agree?
Maybe? Depends on how much light you need. A single lumen doesn’t consume much, and the cameras are a watt or two? Probably similar to power consumption of an acoustic dvl…
There’s a new group out of MIT that has tested some new error correction models to get more accurate readings out of an INS and they are trying to make an open source tool from what I’ve gathered so far. http://ccom.unh.edu/publications/counteracting-autonomous-underwater-vehicle-auv-localisation-error-precursor-blue-water
From their paper the results are pretty exciting so far.
Sounds like they would be - any idea on how to access the paper?
I contacted the Author: Supun Randeni directly, and will see if he will let me share the info here! Otherwise you’ll have to reach out and see if he can provide a copy to you directly. Stay tuned.
@tony-white I asked Supun if he’d be comfortable with me sharing the overview and he said since it’s still under slight revision, he would prefer people contact him directly to get a copy.
The Pixhawk isn’t bad as far is sensors, but those small low cost MEMS, rigidly mounted to the PCB will always be a limiting factor. I’ve worked for a few ROV companies and people do use MEMS IMUs with some success, however it’s just used in a EKF that’s still receiving USBL/LBL/SBL data for zeroing. There are some nice IMUs/INSs out there that use fiber gyros, but like the DVL, the cost is prohibitive at $50-200k usd (it’s also ITAR restricted). You could probably due 1 to 2 orders of magnitude better than the Pixhawk by using a better factory trimmed/temp-compensated/calibrated part with good nulling, vibration isolation, and a dedicated microcontroller for the integration. Here’s a part commonly used in larger ROVs (where cost or export restrictions require it): https://www.digikey.com/short/pbcdcc
I am also interested in a low cost DVL. It seems to be an oxymoron as they are crazy expensice ($20K crazy). To make one you need a coherent sonar source that you can derive doppler from using a moving window FFT. I have not found even any low cost doppler capable sonar devices. most of the cheap ones are just ranging sensors like the ones on the BlueROV website. If someone can find me a low cost doppler capable sensor I can make a DVL I am sure.
There are off the shelf transducers, but they’re not cheap (about $2k/element). I’ve worked with hydraphone designers before and you get what you pay for with ceramics. The electronics, though they have come down a lot, are still going to be orders of magnitude more than a raspberry pi and a lot of it’s odd-ball low noise analog ICs and precision components (and careful layout technique required). Still, it’s possible to get cheaper if you have a more narrowly defined design criteria. It’s a sliding scale with $50k and top performance/reliability at one end, and it could extend maybe even into $2k range with the right combination of features.
Most DVLs are designed around more industrial uses, but I wonder what 80% of the BlueROV2 users out there could get by with? Maybe if a few people would chime in with the minimum DVL they actually need we could get an idea of what’s required and what it would cost?
What’s the highest altitude above seabed that would be acceptable to you (maybe 30ft)?
What’s the lowest altitude where you could still get reliable data (4in)?
What’s the lowest absolute velocity accuracy you would require (+/-1in/s)?
Would a depth rating of only 300m be sufficient?
Would it include an IMU and processing (i.e. X/Y velocity and position), or just raw Doppler shift on all 4 elements?
How would you want it connected into the system (14V battery power and USB, RS485, etc)?