Show HN: Nematoduino – Implementing a Worm on the Arduino UNO | Hacker News Reader

Show HN: Nematoduino – Implementing a Worm on the Arduino UNO

github.com

101 pointsnategri14 comments

I was expecting a worm that would infect computers that tried to program Arduinos and then would infect more Arduinos.

The actual content is cool in its own right, just not at all what I expected.

One step closer to Ghost in the Shell!

Great work, and seems like a very good learning device. Can't wait for more complicated organisms to be mapped and more complex sensors and electronics plugged in. The debate about what is and isn't consciousness is going to be kept alive for a long time.

Thanks for reminding me this exists..

It's cool but something about it still makes my hair stand on end.

I wonder would it be possible to build onto worm connectome?

add ways for it to handle visual stimuli?

As somebody with no idea how nematodes work, what does this model do? Does it basically try and find things that stimulate the chemotaxis neurons?

It looks like it's using the neural map from http://www.openworm.org/ to drive the robot around. What researchers have done is made a complete connection map of all the neurons in a nematode and what they do, allowing a simulation to be run that essentially emulates the brain and nervous system of the worm.

Author here. Thanks for all the awesome feedback, it's been a wonderful day :)

For a 1 minute clip showing nematoduino's reaction to nose touch stimulation, with a comparison to a biological worm, check out this YouTube video I just posted: https://youtu.be/HIs7IbhaW7Y

Is it self-reinforcing/learning, by any chance?

The weights on the neural connections are fixed, so it probably has no capacity for learning in the way that you're thinking.

However, this might actually be the case in the biological worm. This organism is simple enough that the only 'training' it receives may be through the evolutionary process.

Semi-related, I was reading a book called Wetware - A computer in every living cell, and it mentions about William Grey Walter, who created simple robot 'tortoises'.

There's some intriguing videos of them - https://www.youtube.com/watch?v=wQE82derooc from the 50s.

I'm just trying to find some schematics at the moment, it sounds like they used vacuum tubes.

Obviously Grey Walter's robots will be far more simplistic, but I found them intriguing.

The OpenWorm project is really cool too, I'm curious how well their model maps to reality now, whether it's possible or will be possible for the simulated model to produce near identical movement to a real c. elegans under the presence of certain chemicals.

> I'm just trying to find some schematics at the moment, it sounds like they used vacuum tubes.

Yes - vacuum tubes and relays, cds cells (or maybe selenium cells). Transistors would have been very expensive and fragile, if even available.

Here are some links that may help you:

http://cyberneticzoo.com/category/cyberneticanimals/grey-wal...

http://www.roboticsportal.it/wp-content/uploads/2014/01/Spec...

http://4.bp.blogspot.com/-exHhj_kM47Y/VVknwK5WPsI/AAAAAAAAA5...

Now, it's just a short loop, but the gif in the project repo seems remarkably organic when compared to a similar bot that just uses 20ish lines of code to turn at walls. It even seems to pause and "think" at points

I'm pretty sure that's it just running up against a wall :) But your point still stands! Part of the joy of this project has been interacting with and observing a biological system that for all intents and purposes behaves like it is alive.

I'm very inexperienced with C, but why does this repo include so much of the implementation in header (.h) files?

Arduino sketches don't have a "project file" to list which .c files to compile, so instead you just include all the code into one compilation unit. This works fine as most projects don't take long to compile anyway.

Reply on news.ycombinator.com