Researchers at EPFL’s Neuroengineering Laboratory, led by Pavan Ramdya, purpose to duplicate the workings of the mind of the widespread fruit fly, Drosophila melanogaster. EPFL spoke with Ramdya in regards to the thrilling prospects for robotics.
On the display screen, white on a black background, a fly is magnified hundreds of instances and walks calmly throughout a spherical floor on its six legs. “Watch, in a second it will do the moonwalk.” We’re within the coronary heart of the EPFL Neuroengineering Laboratory with Pavan Ramdya, head of the lab, and a postdoctoral researcher, Maite Azcorra. She shines tiny targeted pulses of laser mild on the fly utilizing a method referred to as optogenetics, which makes use of mild to activate particular neurons. As if on command, the fly strikes its legs backward. And it seems to be identical to a dance.
Ramdya’s 14-person analysis group has been finding out the nervous system of those 2-millimeter-long bugs since 2017. “Maite is at the moment finding out how neurons that descend from the mind management motor features,” says Ramdya. The group hopes to ultimately reverse-engineer the fly’s mind and to mannequin it for robotics.
One main step ahead was the event of a digital twin that the researchers can use to precisely simulate a fly’s behaviors; one other was an necessary breakthrough in understanding how neural networks flip mind indicators into coordinated actions. EPFL sat down within the workplace of the New York-born neuroscientist to speak about his work.
Are you able to describe the overall concept behind your analysis program?
People have been making an attempt for hundreds of years to construct machines that may behave like animals or folks. In Historical Greece, for instance, automated marionettes had been fairly widespread—these had been easy objects however they had been already a type of biomimetics, in that they imitated how an precise physique strikes. That is the identical concept we’re pursuing right here, besides that we use way more superior strategies and programs that may really bio-mimic animals just like the fruit fly.
Why are you finding out Drosophila melanogaster particularly?
There are extra sophisticated animals after all, like mammals, however they’re tougher to review. And there are easier ones like C.elegans, a worm with solely round 300 neurons [flies have about 100,000 and humans have approximately 86 billion] however we will not study as a lot about conduct from them. Not like worms, flies have legs, and so they do lots with them—stroll round, clear themselves, manipulate obstacles and extra.
It is rather more attention-grabbing for purposes in robotics and neuroprosthetics to understand how a creature with each wings and legs works. They’re good specimens from that perspective: easy sufficient to review, but complicated sufficient to supply many insights.
In your latest TEDx discuss, you stated that sooner or later robots used to discover and colonize new planets may look lots like these flies.
Sure, robots for area exploration might want to full quite a few duties on their very own and make selections autonomously whereas transferring in unknown, hostile environments. Engineers have been working to construct such robots for many years, however for now, even essentially the most refined machines have nowhere close to the agility of the fruit fly.
Flies can do unimaginable issues. Not solely can they fly, they’re additionally extraordinarily steady as a result of their six legs. They’ll transfer in all three dimensions whereas performing different duties with their legs. They seem to be a main supply of inspiration!
How may the work you are doing affect the event of robotics and AI?
Many engineers are engaged on the {hardware} facet of robots—for instance, batteries and motors. That is not our focus. We’re in search of to design their controllers. With the thought of creating a robotic fly, our foremost focal point is to grasp the way it can management its limbs. That is why we’re finding out the fruit fly’s nervous system—to realize insights that can assist us develop neural networks that can be utilized in robotics and AI.
I would additionally level out that the robots utilizing these controllers do not must be the scale of a fly. So long as it’s scaled appropriately, they are often any measurement—whilst large as a home, though that may be slightly scary!
However your analysis consists of different facets, too.
That is proper. One distinctive attribute of flies is that their legs are coated with mechanical sensors. How do flies use all the data they collect to grasp their surroundings and detect objects round them? How do they determine when to raise a number of of their legs over obstacles?
These are the sorts of questions we wish to reply. And to do this, we’re making an attempt to develop supplies patterned after the fly cuticle with built-in sensors that can be utilized in robots.
Many robotics and AI consultants have stated that to create machines genuinely able to studying, the machines will need to have our bodies that may transfer round and discover their surroundings.
Sure, that is a central principle held by scientists finding out neurobiology and conduct. And it must be a central principle in AI too, since animals can behave extra flexibly than robots. Engineers working in machine studying typically level out that human infants always transfer round and contact issues, exploring their environment to study in regards to the world round them.
This course of is rather more efficient than if we simply introduced them with movies of their surroundings. The sensors I discussed earlier—these on flies—additionally serve this function.
What are at the moment the largest hurdles to creating programs that may study by exploring their surroundings?
One hurdle is to create algorithms that may course of sensory knowledge. If they cannot contextualize these knowledge, it will be very troublesome for machines to study the suitable behaviors. It is necessary to emphasize that the answer does exist—it is simply hidden in animals’ nervous programs. That is what we’re making an attempt to uncover. As a substitute of spending many years trying to design an answer from scratch, why not have a look at what already exists in flies?
Will that essentially be a better and quicker strategy?
What we’ll in all probability want is a mix of various approaches. Particularly since animals have a whole lot of constraints and aims that are not related for our functions. As an illustration, robots do not want to have the ability to reproduce or defecate.
That is why we should additionally embrace biologists in our work and never simply engineers. They’re higher in a position to know which elements of an organism we are able to ignore, equivalent to neurons used to evacuate meals, in order that engineers do not deal with them. An interdisciplinary strategy is important for the work that we’re doing.
Is the objective to ultimately map the human mind?
I will have to present you a egocentric reply: for me, personally, that is not my objective. I’ve round 40 extra years to stay if I am fortunate, and I would actually prefer to see main breakthroughs in my lifetime that might present how organic programs work. That appears potential with the fruit fly, however it will be way more sophisticated for the human mind.
Maybe it is only a matter of scale—possibly we merely have to take a fly’s mind and multiply it by 1,000,000. That may give us one thing clever and will surely be very attention-grabbing. However I am undecided that it will seize human intelligence. I do not assume we might use precisely the identical strategy for the human mind as for the fruit fly, as a result of it will take too lengthy.
How is your strategy to neuroscience totally different from that of different neuroscientists?
In neuroscience, I would say that over 99% of persons are engaged on subjects associated to human well being and drugs. Most research that study how neuroscience can inform the remedy of a illness, as an example, are carried out utilizing mice or rats, as a result of they’re mammals like us. I feel what our analysis group is doing can shift folks’s views in two methods.
First, we have a look at neuroscience not simply when it comes to human well being, however quite in how it may be utilized in robotics to construct machines in new methods. And secondly, we draw consideration to the smaller fraction of neuroscientists finding out bugs.
We must always take into account that most of the bugs on our planet are below risk. Take a look at bees, and the necessary function they play in pollination. It is a essential situation. Fruit flies aren’t a threatened species however they will present perception into ones which are, serving to to help conservation efforts. This will encourage folks to view the world from a extra eco-systemic perspective and admire the important contribution made by biodiversity.
Quotation:
Q&A: Fruit flies are a significant supply of inspiration in robotics (2025, April 4)
retrieved 25 April 2025
from https://techxplore.com/information/2025-04-qa-fruit-flies-major-source.html
This doc is topic to copyright. Aside from any honest dealing for the aim of personal research or analysis, no
half could also be reproduced with out the written permission. The content material is offered for info functions solely.