Dolphin-Controlled Robots


Photo Credit: TED

One of my favorite talks from this year's TED conference was Denise Herzing's talk about dolphin communication. Denise
is one of the leading dolphin researchers in the world and has been pioneering work on dolphin communication, most recently with her CHAT system. I had been following Denise's work for years, and I was lucky enough to meet and talk with her at the conference.

We got to talking about technology, and how OpenROV could aid in her work. Denise brought up the idea od creating a robot that the dolphins could actually control. It sounded crazy at first, but that longer I thought about it, the more intriguing it became.

And apparently it's already been done (although not well documented). Denise sent me a study from 1981 done by R. Stuart Mackay that tried something similar.

Here's a diagram that outlines the study:

And a photo of the dolphin controlling the electrical submarine:

A few weeks ago, Denise came to the OpenROV HQ in Berkeley, and told us more about the study. We discussed ideas on how we could build such a device. If we can, Denise has agreed to help us test it in the field.

What do you think? Have ideas on how to modify an OpenROV for dolphin control?

Full RS Mackay Study


That is one of the most incredible thing's I've read in a while (that was good anyway). Fascinating! I don't own a kit (yet) and not much of a mechanical engineer, however, the recognition and control software is something else. I'll consider some mods from my own rov and see if I can port them to the OpenROV kit (i may have to break down and get one now!).

First high level thoughts are:

-hydrophone, possibly a hydrophone array for direction tracking identification

-modified voice recognition package (off-board at first)

-not sure if the tether would get in the way. since i would think that you'd want the unit to interact with the dolphin, may want an AUV setup or high speed acoustic modem to transmit from the rov to the surface (something out of the range of the dolphin frequency)

-would need to set up a basic vocabulary. probably use the Pavlov like idea at first (from the paper) and then build from there. map rov movement to a reward, something like that.

-also should have a ping to respond to the dolphin for audio feedback, let the dolphin interact with the rov.

-also, pad the rov, pad it well. the dolphins may wish to play with it

-paint it the same color as the dolphin...they may adopt it into the pod...that would be a paper!

-expanding on that, maybe even have the rov trac the dolphin and follow...difficult but not out of reach.

I'm totally geeking out about this, I hope some great science comes from it! Awesome!


Having finally been able to see the ROV at Maker Faire, I have a better idea of the mechanics. The ROV is fairly nimble, which is good when dealing with incredibly agile dolphins. The ROV is compact and, at least to me, unassuming such as to not illicit a fear response. How does marine life in general react to a stationary and then to the moving ROV?

Should try to damping the motor noise. I'd like to analyze the frequency noise of the ROV and compare it to the range of a dolphin to see the overlap. Mitigating communication noise would probably be a good idea, or masking it, say constructively tuned for a positive signal for movement, or destructively to negate the propagation wave. The blog about brush-less and magnetic drive motors would be well suited here I think.

At first, an interesting step would be to build a small acoustic array that the ROV could determine direction of the dolphin signal to align movement. I'm thinking about real basic first principle like steps such that goal 1 would be just get the dolphin attuned to aligning the ROV. This is difficult due to reflections of the test tanks, water particulates, etc, but solutions to this exist, we just need to find them. Actually, a small hackerspace built short and long baseline array would be great projects.


Additional point: the CHAT system is described as being relatively small...but that's all I've been able to find out about it. So we know the dimensions? Can it be attached to the ROV? What buoyancy considerations will we need to make? Power requirements? ect.

One article states that the immobility of human-dolphin interaction may be one cause of seemingly disinterested dolphin involvement in communication experiments. the CHAT system was built to enable more movement and interaction in the human-dolphin study. If we add the CHAt system to the ROV, the agility of the ROV will decrease, so higher torque motor and prop redesign will most likely be needed. I recall a discussion about the ROV just at the max out point in regards to available current supply to the motors, is this correct?

I really need to get one to study. LOL