//
you're reading...

Fisheries

Echoes in the deep: Robots with fish finders

Article: Mark A. Moline, Kelly Benoit-Bird, David O’Gorman, and Ian C. Robbins, 2015: Integration of scientific echo sounders with an adaptable autonomous vehicle to extend our understanding of animals from the surface to the bathypelagic. J. Atmos. Oceanic Technol., 32, 2173–2186. doi: http://dx.doi.org/10.1175/JTECH-D-15-0035.1

Background

You may know that some animals use sound to ‘see’ their environment, like bats and dolphins, in a process called echolocation. And you may know that humans have produced our own echolocation devices like sonar and commercial fish finders. In this study, scientists have passed this technology on to underwater robots. Using sound to identify objects under water is useful because sound travels much farther than light in water. This means that even if water is murky or simply too deep for much light to penetrate, echosounding devices can still identify objects.

The basic principle behind an echosounder is a device that emits a short pulse of sound, and then listens for the returning echo to bounce back off of nearby objects. The standard technique today builds on this concept by emitting the pulse from multiple transducers (think of them like speakers that can also listen for noise) simultaneously, in slightly different directions. By analyzing how long the sound takes to return to each transducer, the echosounder can determine both how far away and in which direction an object is. Additionally, the intensity and other properties of the sound returned can help determine the size and shape of the object (Fig. 1), and sometimes even its internal properties. For instance, many fish show up well on echosounders because they have swim bladders (pockets of gas) that return very strong echoes. Lastly, using different frequencies of sound ensures that echosounders can identify all the things a researcher might be interested in, as different materials reflect certain frequencies better than others. Higher frequency sounds can also reveal more detail than lower frequency sounds. The technology used in these devices has progressed to the point where creatures as small as zooplankton can be identified (sometimes to the species level) and counted.

Figure 2: The REMUS 600 AUV, with echosounder components in the middle.

Figure 2: The REMUS 600 AUV, with echosounder components in the middle.

There is a problem that arises however. High frequency sound doesn’t penetrate water as far as lower frequency sound (although still usually farther than light), so any multifrequency echosounder is limited in depth by its highest frequency. Typical scientific echosounders on ships can only penetrate to about 600 meters, but many interesting fish and zooplankton species live deeper. The simplest way to address this problem is to get the echosounder closer to the depth of the species of interest. Scientists have tried to tow echosounders off of ships, but the platforms are sometime unstable and the cables involved are a bit of a hassle.

This paper reports research by a team that integrated echosounders into a REMUS 600 Autonomous Underwater Vehicle (AUV), an underwater robot they think will provide a stable, nimble, and long lasting platform to get their echosounders where they need to be.

Methods

The REMUS vehicle (Fig 2.) is large enough to fit off-the-shelf echosounders, while still being small enough to easily deploy and recover from a research ship. It can also dive to 600 meters, essentially doubling the depth the echosounders could reach if they were on a ship. The team constructed a custom housing (Fig. 3) to put the echosounders into, and also packed in two separate computer systems to process the acoustic data. These are used to process the acoustic data gathered in almost real-time, so the vehicle can be programmed to identify an object of interest and follow it. In addition to this custom payload, the REMUS has a compass and GPS for navigation, and a suite of standard sensors for sensing the bottom and various water properties, as well as enough power for up to 70 hours of travel. All these features make it a very adaptable robot, and its only about 3 meters long!

Figure 2: The REMUS 600 AUV, with echosounder components in the middle.

Figure 2: The REMUS 600 AUV, with echosounder components in the middle.

Figure 3: A detailed look at the custom echosounder setup (A-C), and an image of the vehicle preparing to dive before a mission (D).

Figure 3: A detailed look at the custom echosounder setup (A-C), and an image of the vehicle preparing to dive before a mission (D).

 

 

 

 

Findings

To test their custom system, the researchers conducted two test missions to check that everything worked, and two primary missions, which had actual scientific goals (Fig. 4). They found that the setup was very stable and level on both shallow and deep missions, features that are very important for the echosounders to perform well. If the echosounder platform moves too much (e.g. from waves at the surface) in-between pulses, the sensor may receive only a portion of the returned sound, producing an inaccurate intensity reading. Based on some of the ship’s roll during their testing, the echo intensity returned could vary by up to 50%, while the AUV had almost no roll at all. The vehicle stuck to its programmed path fairly well, surfacing periodically to get a GPS fix to correct its course.

Figure 4: The four missions discussed (B has two missions) and an image of the craft being deployed.

Figure 4: The four missions discussed (B has two missions) and an image of the craft being deployed.

The echosounders also performed very well. Background noise from surrounding electronics and things like bubbles near the surface were reduced compared to a ship based echosounder; the detection range for an object increased by around 30 – 40% in the AUV setup (Fig. 5). Ship based echosounders also frequently encounter bubbles at the surface that obscure the data, a problem that the AUV never had once it dove a few meters down. Getting the echosounders closer to the targets with the AUV also clearly worked well, as shown in Figure 6. The ship could barely detect even a few organisms at depth with the 120 Hz transducer, whereas the AUV detected an entire layer of organisms. Even when the ship could detect organisms, the AUV returned a much higher resolution image, giving betters count and size estimates.

Figure 5: A side-by-side comparison of ship and AUV data from both echosounders. You can see that the AUV does a better job of capturing separations between different layers. The green section in panel C is essentially all noise, obscuring any useful data, but the AUV avoided any noise problems.

Figure 5: A side-by-side comparison of ship and AUV data from both echosounders. You can see that the AUV does a better job of capturing separations between different layers. The green section in panel C is essentially all noise, obscuring any useful data, but the AUV avoided any noise problems.

Figure 6 : A detail of one of the missions. It’s clear that the AUV captures more data at depth than the ship, and with much more detail.

Figure 6 : A detail of one of the missions. It’s clear that the AUV captures more data at depth than the ship, and with much more detail.

Broader Impacts

This modified REMUS AUV has essentially opened a whole new region of the middle ocean (a.k.a the mesopelagic) to acoustic study. This area is home to many important commercial fish species and their prey, and studying them acoustically allows for larger areas to be surveyed faster. This proof of concept device also paves the way for projects that ships can’t perform, like studying how some fish might avoid ships (it’s hard to tell if fish are avoiding the ship you’re on because they aren’t there!). Although I only mentioned them briefly, the additional sensors on the AUV can provide more data on what type of water certain organisms are observed in. The programmable nature of the AUV also makes it much less tedious for researchers to cover large areas or easier to track a particular target animal without having to maneuver a ship.

Engage

What do you think the researchers might find using this system? What other environments or situations would an autonomous vehicle be useful in?

Austen Blair is a MS candidate at the University of Rhode Island Graduate School of Oceanography. While his current research focuses on the influences of wave fields in a hurricane-wave-ocean model, he enjoys the many interdisciplinary opportunities the field of oceanography provides. When not doing research, you can find him on the water, rock climbing, or on his bike.

Discussion

Trackbacks/Pingbacks

  1. […] – from using sound in the ocean. Tomorrow, Austen will cover an article discussing the use of autonomous robots for monitoring fish populations. Wednesday is a double header: Megan will look at whale earwax and […]

Talk to us!

oceanbites photostream

Subscribe to oceanbites

@oceanbites on Twitter