“… although the true location of the vessel is known, it is not accidental,
it’s there, but it is unknown at what position”.
V.G. Aleksishin et al., Practical Navigation, 2006. p. 71
With this feature we open a series of articles on the practical application of various navigation systems of our own production.
The RedNav LBL navigation set that bore the serial number 1 was produced and delivered to the Customer in the summer of 2015, already far. We have sold dozens of such sets since then, however, up until April 2019 we couldn’t carry out full-fledged tests on our own robot its availability since 2017 notwithstanding, having been focusing on development, improvement and production of both old and new devices and systems. In this article we will tie some loose ends and describe how it works from the developer’s point of view.
Flying and rolling, crawling on the ground and swimming on the surface drones transmit videos and often GPS-coordinates from their cameras in real time. The operator can easily find out where his piece of apparatus is located; moreover the operator can flat-out see it.
It is somewhat different with underwater vehicles such as ROV. Having released the device into the water, the only thing you can be sure of is that it is definitely under water.
A Bit Deeper Into The Issue
Underwater devices can be categorised into different classes, depending on size and tasks. The most simple and small ones, the viewing kind we have, is just a videocamera on a cable with engines. The more complex and large ones can be equipped with manipulators and other elements of interaction with external environment. Cable lengths may vary between dozens and hundreds of meters for small devices and may amount to thousands of meters for serious devices of the so-called working class.
The classic control of ROVs is carried out by visual feedback. The operator sees the image transmitted from the cameras of the device via cable, that is often equipped with sonar sensors since the visibility in the research and work area is oftentimes poor and does not exceed 1-3 meters.
The most significant drawback in this approach is that most of the time it is impossible to locate the device just by looking at these images.
This disadvantage can be overcome by applying hydroacoustic navigation system. As a rule either a pinger (a device that emits a special signal) or a responder-beacon are installed on the device. The bearings of pinger signal are taken, the distance is being determined, and then, based both on the signal arrival angle (or two angles – horizontal and vertical) and distance the location of the apparatus is being determined. These kind of systems are called SBL, short for short baseline systems. They are classified as AoA (angle of arrival) + TOF (time of flight) system and have a number of shortcomings, especially in relation to this task.
In order to determine the horizontal angle of arrival of the respondent or pinger signal one requires a direction finding (DF) antenna. In older systems it is also necessary to determine the vertical angle of signal arrival, in new systems, including ours, the beacon-responder transmits its depth, which simplifies the task and increases the accuracy of the complex. A DF antenna is a complex device in itself and needs to be mounted on a rod attached to a vessel. The distance, depth and horizontal angle or the distance and two angles determine only the relative position of the device. However the accuracy decreases with increasing distance.
The accuracy of the angle depends on the following:
Once the distance and the angle of arrival of the signal is determined it should all be tied to geographical coordinates. To do this one needs to know a geographical position of the DF antenna and the direction of its zero relative to the true north – in other words add to that a compass and GPS on the antenna – a system of course and position. After that one can solve a direct geodetic problem and determine the exact position of the underwater vehicle in geographical coordinates.
In terms of accuracy the use of long baseline system is preferable to an ultra-short baseline system. We only recommend using a USBL when it is impossible to use LBL systems. An example of this can be the positioning of a towed object with a need to cover a very long distance. In this case the elements of a LBL will have to be moved very often which will lead to a waste of too much time and effort. Or another example, when the LBL buoys are impossible or too difficult to install on the surface because of the great depth on the site. In all other cases it is advised to use RedNav LBL navigation system, it is more reliable and more accurate than a USBL system.
In LBL systems a navigation base is formed by several receivers or transmitters dispersed in the area. GPS and GLONASS are the best known examples of LBL navigation systems, but there are other satellite navigation systems as well. The advantage of LBL navigation systems lies in their invariable accuracy throughout the entire area within the base, they are significantly less susceptible to the effects of rocking and all in all provide much more accurate results than USBL systems.
In practice, however, we see that users prefer to purchase and install USBL systems, based on the assumptions that USBL systems are easier to deploy than LBL systems. This stereotype exists not least because the overwhelming majority of LBL systems available on the market are represented only by the so-called deep-water base in which, unlike our RedNav system, the elements of the base are non-floatable and are installed on the sea bed on site. The deployment of such deep-water base requires significant amounts of time and money.
In our LBL system we combine the advantages of easy deployment with the high accuracy of the result.
Let’s get back to the highlight of today’s testing – RedNode System. The Navigation system consists of the actual navigation base, formed by four floating buoys transmitting GNSS signal:
Before operation the buoys are deployed on the water body with the help of anchors and rope. All you need to do is to bring the buoy to anchor, turning it on beforehand. That is actually it! No calibration, presynchronyzations, etc., just turn it on that’s all. During testing our engineers deploy the navigation system buoys on a row boat in less than half an hour.
Another element of the system is a navigation receiver installed on a positionable underwater object:
This yellow cylinder is a RedNODE navigation receiver is installed abaft the apparatus. (желтый цилиндр) установлен на корме аппарата. It is powered by the robot’s on-board network and transmits data via the vehicle cable.
Given that the buoys only emit and the receivers only receive – a solution based on Time Difference Of Arrival (TDOA) measurements, it is possible to ensure the service of any number of such receivers on one set of buoys in one water area. In other words a whole fleet of submersibles and divers can navigate simultaneously, each of which will receive its own location with a nominal frequency of 1 Hz.
We simply installed a navigation receiver and added some buoyancy to level out the robot’s balance.
The data from the receiver goes straight to the control panel from where it is possible to transfer it to any laptop via RS232<->USB converter (the “Sonar” outlet in the photo above).
The coordinates in our LBL system are generated on the receiver, that is to say strictly terminologically it is a navigation system not a positioning system. But since the ROV operates on cable it is easy to transmit the location calculated on the device by cable upwards. Navigation receiver of the system can operate in a mode of emulation of the usual “land-based” GNSS receiver, which allows you to connect it to any software that can work with GNSS receivers. An example of this can be the popular application SAS.Planet that works well on-line with our navigation receiver.
The test bench is quite simple:
Despite the apparent simplicity we wouldn’t have been able to work with USBL system on this watercraft, because to mount a USBL antenna you need a rather large boat to withstand the rocking with a hard bottom and a rod firmly fixed on board. Furthermore it would have been necessary to stay on the water all the time which in some cases, such as low temperatures, winds and waves, can be uncomfortable or unacceptable.
The so-called Command Station was set in 10 minutes and in our case looked like this:
As usual all field tests are conducted at the mouth of the Pichuga River at the place ot its inflow into the Volgograd reservoir.
As it was mentioned above it takes less than 30 minutes to deploy the system on water is. This time two people on a rowing boat did it in 24 minutes, on oars, fighting the wind and excitement.
In the photo above shows the buoys in a small inflatable kayak. All four of them.
One would think that the submersion is also made from the boat but that’s not necessary. Under different circumstances diving can also be carried out from the shore, the unit is simply put into the water:
And here are the first pictures of the underwater world:
Yeah, this isn’t the Red Sea. The water seems clear but the truth is the visibility close to the shore does not exceed 1-2 meters.
We were upset but the fact remains, it is not possible to operate a ROV based only on the image because of the extremely low visibility under water.
Those who wish to assess underwater visibility at the test site can do so by viewing this footage recorded from the camera on the device:
The video is posted without any filters, special effects or editing. Запись дана без какой-либо обработки и монтажа. You can reach the conclusion about the ease of controlling the device and performing meaningful action underwater, searching for something for instance, using only the image from the camera without navigation.
And here comes the first touch of the waterbed and some elements of the so-called moon landscape at the depth of 13 meters:
Just a couple of seconds later, having gone a little further, the device came up against the flooded log overgrown with small shells:
The test scenario was based on the following sequence of actions: an easily identifiable object is flooded from the boat, the coordinates of the flooding site are saved using a cellphone, and the operator’s task is to come to the place of flooding using navigation system and controlling the robot and try to visually detect a flooded object.
We have found that it is absolutely impossible to operate the ROV using only the image from the camera. We were mainly guided by our navigation system, which provided the current location of the device on the map online.
The professionals will be pleased with the system’s resolution in a real water body, which is about 30 centimetres as you can clearly see from the grid obtained by determining the position:
Just as with other tests of the same kind the range of motion points is 1-1,5 meters which provides sufficient accuracy for most inspection works:
The final track of the device movements looks like this:
We were pretty off with our estimation of the work site range and nearly half of the track (everything on the left of the red lines) lies outside the navigation base, in other words outside of the buoys’ geometry, where the system should work much worse. However, with only a few exits the system had worked with a very high accuracy.
During one of its passings the apparatus was very close to the object’s flooding site:
But having carefully watched more than an hour-long footage recorded by the on-board camera, not once have we seen the sought object which indirectly confirms our observations about the impossibility of performing any meaningful actions relying only on the video image.
The video report from our testing:
We invite you to discuss this article. if you are interested to pursue it further individually here is the link to the tracks obtained during these testings. They are in kml-format.