If you have ever observed an ant moving about it is a remarkable thing to see. Their exoskeletons move at the speed of light. They can push objects around many times their size. And never tempt them with a morsel of food for they will recruit all their family and friends to join the feast as they jointly devour everything on the plate. This is all done with a nervous system that contains vastly fewer cells than exists for the human brain (i.e. 250,000 neurons vs. 80 billion neurons).
So what about having an ant control an external robotic device?
First, there is a common notion in the field of brain-machine interfaces (BMIs) that 'the sum of one part of the brain equals the whole', namely by extracting signals from a dozen or so cells from one region of the brain one will be able to recreate all the signals to move a whole body with effectiveness (e.g. Pais-Vieira et al. 2013). This would be considered an absurd proposition for any scientist who has ever studied a whole brain system for the generation of movement (e.g. Schiller and Tehovnik 2001).
Second, it is known that after some 50 neurons the signals extracted from the brain begin to saturate for the control of external devices (Tehovnik et al. 2013; Tehovnik and Teixeira e Silva 2014). This bottleneck is partially related to the fact that recordings are made from one part of the brain without any consideration of how these signals interact with other parts of the brain to produce movement.
Third, since movement is generated by an entire system of neurons (Schiller and Tehovnik 2001), it might be more productive to use a 'systems neuroscience' approach for the development of BMI. The study of the ant brain with its fewer neurons than that of the human brain but with its highly advanced locomotor repertoire might be one way to approach this problem.
As pointed out already above, this was done already several times. But here are other approaches, non-invasive, leading to similar results:
You can place an animal on an air-floated styrofoam ball which is mounted on the robot and record in which direction the animal runs, putting this then on robot wheels. To close the sensor-actor loop, you can just interface (touch, smell) to the antennae without any section.
Advantage: You can include their physic-chemical sensor-fusion properties into that loop. You would loose those aspects if you directly go to the nervous system.
Or you can take the "social approach" that uses normal social animal interaction to make a bio-hybrid. See: http://assisi-project.eu/
First: would you want to do this? An ant's brain evolved to accommodate the precise circumstances that ants live in: environment, sensor attributes, ant social 'needs', etc. Unless you could completely understand the ant brain, any use of its functionality is a 'best-guess' that could likely be accomplished by other means.
Second: the attraction seems to be the potential mental flexibility, learning ability, of the brain. And yet, by harnessing the brain you would be adding constraints to those very abilities simply because you want the brain to do what _you_ want, not what _it_ wants. In which case, why not simply build a device that does what you want it to do in the first place?
Third: why hybridize organic and silicon (or other substrate)? An organic brain evolved to meet the needs of an organism AND the organism evolved to meet the needs of that brain (nutrition, etc.) within the constraints of its environment (organic tissue is easily damaged and must self-repair, etc.). Silicon-based computers face completely different requirements including a very different set of environmental concerns: electricity vs glucose, temperature indifference, humidity sensitivity, physical wear, etc. A hybrid system however has to accommodate constraints of _both_ component systems ... at which point one might ask, again, why would you want to do this?
Conclusion: Yes, an ant brain could control a robotic device. And yes, in turn, you could control an ant brain to control a robotic device. But ask yourself first, why would you want to do this?
[Note: This reply does not cover the very thorny ethical question of enslaving another species, a topic that is very relevant to anyone dealing with real or artificial intelligence.]
Yes, with reference to Research Scholar of UTS Faculty of Engineering and Information Technology, said the behavior of social insects, like ants, allows scientists to understand the body's electrical signals and use the knowledge to create a robotic prosthetic device that can be operated by human thought, like a flesh and blood limb.
They said that, we can use the behavior of the ants to enhance the quality of the control systems that we employ with the robotic limbs. The biggest problem in our field is the amputee acceptance rate — they are disappointed if the system is not fast and accurate enough,”
According to Hölldobler and Wilson (1990) ants have up to 20 letters in their communication, scent code which is just short of the number of letters in the English alphabet. To identify 1 of 20 letters represents 4.3 bits of information [bits = 3.3219 x Log10 (20)]. It takes an ant about 3 seconds to discriminate a chemical (McIndoo 1914). Therefore, the information transfer rate for ant chemical-communication is 1.4 bits per second. Now transferring information is something that humans are much better at exhibiting rates of up to 60 bits per second when based on an advanced reader of books (Reed and Durlach 1998).
Hölldobler B, Wilson EO (1990) The Ants. Harvard University Press, Cambridge MA, pp. 1-732.
McIndoo NE (1914) The olfactory sense of hymenoptera. Proc Acad Nat Sci Philadelphia 66:294-341.
Reed CM, Durlach NI (1998) Note on information transfer rates in human communication. Pres Teleoperat Virtual Environ 7:509-518.
Bees isolated in a transparent chamber were required to observe a demonstrator bee pull a string attached to an artificial blue flower to get a sucrose reward as conducted over 10 trials (Alem et al. 2016). Of 25 bees observing the event, some 15 bees (60%) could successfully perform the task on the first trial. The string-pulling of the observer bees was very slow at 181 seconds on average (which was far greater then the latency of highly trained bees at ~ 20 seconds). Of 110 bees put in a chamber with a string and a flower only 2 of 110 learned to perform the task spontaneously following two bouts of a 5 minute exposure time (totaling 10 minutes). Thus the spontaneous learning of string-pulling was rare. In all these experiments the bees knew that the blue flower contained sucrose based on prior exposure.
The most intriguing part of the string-pulling skill is that it was shown to be transmitted across generations of bees such that the original teachers were long dead as the skill was being passed on. Perhaps the transmission of culture by human beings has been overrated, and most importantly: let’s start taking care of our bees.
Alem, S., Perry, C.J., Zhu, X., Loukola, O.J., Ingraham, T., Chittka, L., 2016. Associative mechanisms allow for social learning and cultural transmission of string pulling. PLOS Biology doi:10.1371/journal.pbio.1002564.
http://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.1002564
- Badges
- Science topic