DARPA’s latest call for proposals merited an interesting article in the Toronto Star.  The advanced projects agency has requested labs to submit research plans that will lead to :

the controlled arrival of an insect within five metres of a specified target located 100 metres away. It must then remain stationary indefinitely, unless otherwise instructed … to transmit data to sensors providing information about the local environment.

The preferred methodology is to implant computational units in the larvae and allow them to integrate with the nervous system through pupation.  A high goal indeed and one that is so far from any proven science that it certainly qualifies as ‘blue sky’ research.  Last year DARPA gave out 3.1 billion for equally outside the box, high concept research.  The Star article compares this to $325 million in similarly unconventional funding from the Canadian government.  John Polanyi, the Nobel Prize winning Canadian chemist points out that :

The long-term, out-of-the-box approach is why the U.S. is the world leader in science. Canada thinks in the short term. It’s all about wealth creation here, having business models, setting milestones for work even before it’s begun.

It is my strong belief that the most basic sort of research has to be funded at very high levels to feed ideas and capabilities into the engine of commercially applied research. DARPA is clearly an agency that manages to do this quite well. A strict accounting of the money they spend would reveal a host of failed initiatives, the by products of these are staples of research science and have had wholly unintended consequences.

TheStar.com - Uncle Sam’s scientists busy building insect army.

An interesting PDF by Roderick Wallace in which he makes a prediction about the uses of the first biomimetic artificial intelligence systems…to wit that they will be terribly unstable and prone to psychoses.

“The most likely use of the first generations of conscious machines will be to model the various forms of psychopathology, since we have little or no understanding of how consciousness is stabilized in humans or other animals.”

Psychopathic Robots Predicted. (robots.net)

The Defense Advanced Research Projects Agency (DARPA) has been funding several labs interested in basic sensory biology.  Their goal is to define a system that will allow remote control of a free swimming shark.  The project has been under way for two years now with the major focus on the olfactory and the electrosensory systems.   A recent meeting of the research group held in Hawaii has been reported in the New Scientist.  Jelle Atema, better known for working on lobsters, has successfully implanted an electrode in each olfactory tract that causes a shark to veer one way or another.  Tim Tricas is attempting similar things with the electrosensory system but is not yet able to build the interface with the shark.  There are apparently trials scheduled for the Atema system to be held in open ocean off Florida. 

New Scientist Article

Tricas Lab

The co option of animal sensory systems for robotics or as disembodied sensors seems likely to be a successful use of biomimicry. In this instance David Macmillan of Melbourne University has been exploring the sensory modalities of the crayfish as it negotiates dark tunnels.  The antennae, usually thought of as chemosensory structures, are apparently also adept touch sensors.  When faced with a y-maze the edible little fellows will make a random choice.  But, if an antenna is either denervated or tied back along the length of the animal it will pick the side it can ‘feel’ on.  The multi jointed antennae of crustaceans seems a worthwhile place to start looking for ‘feeler’ designs for wall following robots.

Bio Bull article

Macmillan’s page

Science New Article

Bristol and Sheffield universities are working on a biomimetic sensory system for robots.  Whiskers are exquisitely sensitive in determining texture and contact forces.  Better sensitivity is clearly needed in the robotics world and the this is far from the only project looking at biomimetic solutions.    

Their idea is to attach strain gages to the bases of fine glass ‘whiskers’ that will be swept back and forth to assess surface irregularities.  I can see how the system might work, but whiskers are not nearly so active in biological systems.  Perhaps it would be useful to try to understand how animals appreciate texture without moving so violently. It would be easier on the whiskers if they were not flailing. 

Bristol Robotics Lab

The WhiskerBot Project Page

This is a very entertaining and interesting take on the biomimetic robot.  Garnet Hertz at UCI has designed a robot that acts as a waldo for a cockroach.  That is, the roaches movements are amplified and serve as the control inputs for a wheeled robot.  Sensors on the sides of the robot shine lights at the roach when it approaches walls. Hertz hopes these lights will cause the roach (and its attached robot) to move away from the light.  The endearing thing about the project is that there is no processing between the roach and the robot. What the roach does with its legs is the input for the ‘bot.  There is likely some interesting applications that could come out of this approach, but this is the only project of its kind I have ever seen.  It reminds me of the plastic habitrail ball you could slip your hamster in.  The hamster could roll around the living room, scare the dog, tumble down the stairs…all without input from the owner.

Web Site

One of the key inputs to biomimetic robot design is the kinematics of the biological system being imitated.  There are several different ‘gecko’ robots that currently imitate the lizard by spinning various configurations of a sticky tire to climb walls.  The biomimetic models are united in their treatment of walls and floor in the same way.  This paper shows that the geckos are not so limited.  When running up walls they use a fundamentally different gait in which the legs are pulled into the body to set the clinging hairs (setae) into the surface.  On the ground they use a superficially similar gait but the forces on the feet push out rather than in on impact.  This important difference needs to be realized in the next generation of gecko-bots. 

Inside JEB article

Two books entirely devoted to the biologically inspired robots I have been writing about.  They cover all the biggies here: the lobster, the cockroach, even the tuna. These robots have a long way to go before they will be a really useful example of biomimicry, but its fun to keep an eye on this field.

Biologically Inspired Intelligent Robots (SPIE Press Monograph Vol. PM122)

Biorobotics

 Japanese researchers have coupled a control system to a live cockroach.  Appartenrtly they are having some problems with the interface to the control system.  Random cell phone users are making the roach misbehave!

Article

Joeseph Ayres at Norteastern, the same fellow with the robot lobster, has another aquatic bot.  This one is modeled on a lamprey.  It is easier to model a lamprey than other fish because they do not have a heavily segmented vertebral column. It is perfectly appropriate to approximate them as a strip of bendy plastic, which is just what Ayres does.  The actuators are shape memory metal (Nitinol) and the power and control come through a tether.  This one is not swimming free any time soon.  Control is provided through oscillatory finite state machines that produce a wonderful wave along the body while rather faithfully recreating the activation patterns that might be seen in the muscles of a swimming lamprey. 

Web Site