Ability to fly grid pattern


#1

Hey everyone. Wondering if the trident will be able to fly a grid pattern using pre programmed waypoints for mapping purposes like aerial drones can do. This would be a very handy application. Thanks for replies!

Chris.


#2

Almost all drones rely heavily on GPS for waypoint navigation. Unfortunately, you can’t receive GPS signals underwater, so GPS location could only be determined at the surface aboard the buoy that Trident tows. This would give you a (large) ballpark location of the ROV, but not nearly enough accuracy for autonomous navigation. If the ROV’s direction and distance from the buoy could be determined with sensors, then the ROV’s coordinates could be calculated, but I don’t know how/if that could be done; maybe someone else has a good idea.


#3

Have a look at some of the active work @Jim_Trezzo and others have been doing in relation to underwater localization system


#4

Have been giving this one some thought as i am hoping to use my Trident for shallow water survey when we get it. The areas we get our students to work in for teaching exercises are usually shallow, near shore and from 2 - 8m deep.

First thing i’ll try is shortening off the tether length (bundling it up, NOT cutting it short) and trying to do a survey pattern with the Trident 1-1.5m below the surface (if its depth keeping is good enough).

I’m hoping that with a downward facing GoPro / lasers as an external payload, that will mean that the GPS in the WiFi buoy will give me a useful (in my context) level of position accuracy out of the box. If it works, then i would get a short, 2m or so, tether made up and try that.

I have thought about a pole / frame, with a GPS antenna rigged as an external payload above and behind the Trident? Fly the Trident at a constant depth so that the antenna is just out of the water and a constant distance and orientation relative to the Trident? Will try the short tether first though as is simpler and should have less drag / weight distribution issues to deal with…i think?

Deeper water and most research applications would mean acoustic positioning gear though.

Much to play with and try out with this beast. :slight_smile:


#5

It won’t be able to do a decent (accurate) grid out of the box, but use of the buoy to refresh an IMU would make it capable of very accurate grids.

Another option would be a downward looking camera and assuming the bottom could be resolved, it could use an image correlation system to determine it’s position with respect to the bottom.


#6

Hi Ian @i.mckernan

Reading between the lines it sounds like you are doing some Structure from Motion mapping of shallow water sites

Big picture the SfM software is pretty good and you don’t need the GPS location of the camera as a first off location for the software to work out the camera pose as software can work it out by pure computer grunt power (albeit a fair amount of time). Additionally as long as you get a good amount of over coverage of the area the lines don’t have to be straight and you will still get an output.

Given what you are suggesting with the shortened tether (a great idea) you should also be able to upload the track file and then imply a image location (if you can time sync the two sets of data somehow) which (I can’t remember the correct Agisoft terminology) should allow for an “adjacent pair selection type camera location” to be used to shorten the processing time

Above and beyond this if you can drop a say 3 or 4 known ground control points (say different coloured weighted objects at known locations) you should be able to add this into the SfM model (not sure of the accuracy you are chasing but the assumption is that you are only after OK accuracy but not too tight as the GPS would only be ±2.5m with a 95% confidence unless you go survey grade units) to geo-reference the model.

In the future a plug in to interface the Tridents WiFi to the GoPro WiFi and then trigger the frames and record the location (or add it into the image data file via Exif) would be nice

@HR1

In relation

The Laser realtime Distance Measurement plugin that @Joakim_Karlsson had working a while back would pretty much do this without significant modification (hence GPS (sort of X and Y location and Z being set off the bottom via the laser distance plug in)

Scott


#7

I am also interested in semi-autonomous control, line surveys and structure from motion using Trident.

I sigh when I see the cost of underwater navigation systems and locators like the Water Linked Underwater GPS system stocked by Blue Robotics (https://www.bluerobotics.com/store/electronics/underwater-gps/aps-wl-11001/). If another 19 community members want to buy a share in such a system then please let me know hahaha. No seriously, LET ME KNOW!

The external payload modules I plan to build will incorporate provisions for vertical and oblique GoPro cameras and additional Lumen Subsea lighting from Blue Robotics (https://www.bluerobotics.com/product-category/electronics/lights/) to help collect imagery for use in building structure from motion models. The limitation I see in this case is the bandwidth intensive requirement for real time and/or near real time processing of imagery to provide control of Trident in response to position fixes if the need arises.

I have many hurriedly written notes, sketches and ramblings from the past about semi-autonomous systems incorporating tethers to solar powered (or other creative energy type) surface vessels that allow Trident, with a little AI and remote guidance from interested individuals anywhere in the world, to work its way around the nearshore fringe looking for cool things to find, identify, document and share with the world.