Mobile Robotics Platform
by Jeff W. Dawson, February, 2012
Our Mobile Robotics Platform is a device that allows us to study how insects process and use sensory information during flight. The device is also a convenient platform for investigations of neuro-machine interfacing.
The Mobile Robotics Platform, or Locust Car, is a two-wheeled (with trailing caster) single-deck mobile robot that is controlled by a tethered insect.
How it works:
To understand how the car works – specifically how a locust can control the speed and direction of the car – you need to understand how an insect, like a locust, flies and steers.
All insects fly by flapping wings up and down (the kinematics are actually more complex than that, but I’ll spare you the biomechanics). Wings are activated by muscles and the downstroke part of the flapping, in locusts, is controlled by a small number of easily accessible muscles. Each time a downstroke muscle is activated to contract, an electrode placed in the muscle will record a small electrical ‘blip’ (called an action potential). Locusts, like all insects, have hard exoskeletons which, with a little drop of hair-removal wax (yes, bikini wax), can be used to hold an electrode in place in the particular muscle we use to control the car. Insects don’t feel pain so they don’t mind having the electrodes put into their muscles. And, when we’re done with the locust, we can easily remove the electrodes and the locust lives happily ever after.
The muscle is called the ‘forewing first basalar muscle’ or M97, and is activated each time the forewing of the locust is activated to produce a downstroke. This muscle, M97, is also the muscle used by locusts to steer. We put two electrodes in the locust – one in the left M97 muscle and one in the right M97 muscle. The timing difference in activation between the left and right side tells us which way the locust is turning (left or right), and the rhythm of the muscle activations tells us how fast the locust is trying to fly.
Electronically speaking, using these signals (the muscle activations producing the ‘blips’ or ‘action potentials’) is a bit problematic. First, the signals are small so they have to be amplified. Next the signals need to be cleaned up before they can be fed into a small on-board computer. Last, the on-board computer (or micro-controller) needs software to interpret the muscle activity and generate appropriate signals for activating the wheels on the car so the locust can actually drive and steer the thing!
The microcontroller (on-board computer) we chose for the new locust car is an Arduino-Uno microcontroller. This is a popular (open source) device that is relatively in-expensive and is very versatile and easily programmed. Also, a few years back, we designed and built a great bioamplifier for amplifying muscle and nerve activity (action potentials) which we modified for the car. This amplifier is, by the way, now used in our undergraduate physiology course for student experiments on frog nerves.
The first ‘locust car’ called M.E.L. (an acronym for MyoElectric Locust) was built in 2000 by Jeff Dawson and Ron Harding.