M.E.L.

M.E.L. (MyoElectric Locust) is a three-wheeled robotics platform that is controlled by locust flight muscle activity. The locust is tethered to a support on the platform. Electromyographic recording electrodes are then inserted into the left and right forewing first basalar flight muscles (a pair of muscles that controls the timing of the downstroke of the wings). The recording electrodes pick up electrical impulses in the muscles when they are activated by the locust to contract. While flying, locusts steer by asynchronously activating the left and right muscles. This changes the timing of the left and right downstroke which changes the left-right balance of aerodynamic forces. The electrodes let us "eavesdrop" on the electrical activity of the muscles and we use this information to control the electronics of the platform. So...when the locust flaps its wings, the car goes forward and in the direction the locust is trying to fly.

Locusts are large insects native to equatorial regions of Africa and Asia. Locusts are most often noted for the damage they cause to crops when they aggregate into large swarms. Less well know is that because they are such excellent fliers and have an accessible nervous system, they are also used frequently by scientists interested in insect flight. They are considered a "model organism". Locusts also possess ears that are sensitive to ultrasound (sound frequencies above which humans can hear). We believe that bats, which use ultrasound to hunt, eat locusts. This is important for us because it means that we can use a small device called a "Zapper" (like an electronic dog whistle) that produces ultrasound to get locusts to steer. In the video clips below, you'll see us use a Zapper (a small grey box that looks like a remote control for a TV) to get the locust on the platform to steer.

I (Jeff) am a Ph.D. student at Queen's University in Kingston Ontario Canada studying the neural control of ultrasound avoidance steering in locusts, and Ron is a software engineer with Research In Motion in Waterloo Ontario. I got the idea to build M.E.L. when my good friend John Dower FAXed me a photocopy of a Discover Magazine article about Steve Bathiche and his Mothmobile. Bathiche modified a remote controlled car so that activity from hawkmoth flight muscles would cause it to go forward and a harness around the moth controlled the steering.

We spent approximately two months planning the design by conversing over email and four very long (and tiring) days assembling and testing. Naturally, most of the design was accomplished during the four days of construction. Initially we were going to modify a remote controlled car (as Steve Bathiche had done for his Mothmobile) but after much consideration (see explanation in the poster below) we felt that using a platform with three wheels would be better. We found a robotics kit on the internet available from HVW Technologies that was perfect. Those final four days were *very* intense and we worked nearly around the clock. I think we visited every electronics store in Waterloo that weekend. During the last night of construction I remember Ron was debugging the code for the PIC controller around 3:00 AM while I was implanting EMG electrodes in another locust to test the code. We were finally convinced the car was working properly sometime around 4:00 AM after which we decided we needed more space to be sure. At about 5:00 AM we arrived at the foyer of Research In Motion and implanted one last locust. The car worked great!

And... if you're wondering why... M.E.L. was built for the fun of it.

M.E.L. was presented as a poster at the 25th Annual East Coast Nerve Net meeting in Woods Hole MA in April 1999.

Below are some video clips of M.E.L. in action. These video clips are in AVI format compressed using the Intel Indeo video v4.3 codec (these should play without trouble with Windows Media Player).

Build1a.avi
(3.3 Mb)
This first clip was shot during the late stages of construction. Jeff is at the bench adjusting the electronics and Ron zooms the camera in to see the locust begin flapping on the tether.
Build5a.avi
(7.1 Mb)
After the car was complete we needed more space to see if it would work. We took the car to the foyer of Ron's work place. This video shows the car avoiding Jeff's stamping feet. Note that it only takes a puff of wind on the locust to get it to fly and that the locust can steer the platform in both directions. This clip was shot sometime around 5:00 AM!
Car7a.avi
(2.2 Mb)
We start the locust flying by lightly touching the end of its abodmen. This clip shows the locust steering the platform to its left but after being zapped with ultrasound steers the platform to its right.
Car9a.avi
(2.2 Mb)
This video clip shows the locust keeping the platform on a straight course but then steers it to its left after being zapped from the right.
Car8a.avi
(3.5 Mb)
Here we try to get the locust to steer the platform by jingling a set of keys (a good source of ultrasound). The locust should steer away from the keys but instead turns towards them. More interesting is that the locust subsequently steers the platform to avoid collision with chairs and back towards the centre of the room. This shows visual avoidance by the locust and that it can steer the car both left and right.

We thank Fok-Han Leung for helping with the videotapeing for the CarXa.avi files above.

 

2000 Jeff W. Dawson. No part of this WWW site or the video clips may be used without permission.