Artificial Life Milestone: OpenWorm Team Recreates Internal Muscle Sensation Entirely Through Code
There's been a mini-milestone in the OpenWorm Project that I wrote about last Summer, which is the collaborative, open source attempt to construct an artificial life form from the cellular level to the point where it's able to have basic problem-solving abilities. Watch the video closely below:
"That's a simulated worm body with muscle segments that resemble an actual C.Elegans," project advocate John Hurliman tells me. "Each muscle segment can receive a contraction signal, and although the current setup just has a hardcoded algorithm driving the muscles, its movement closely resembles published literature on how C. Elegans swims." In other words - they've artificially recreated internal muscle sensation, a building block for movement, entirely through code.
That's seriously cool, but so is what will follow: "Next steps are to continue working on performance (this is one-third of a second which took 72 hours to compute) and hook up a synthetic brain to it."
By the way, if you're curious about the algorithm that drives the musculature, and how accurately it simulates the real thing, there's hardcore geek details from Mr. Hurliman below:
"The core algorithm for the physics simulation is called PCI-SPH, which is a somewhat advanced but well understood particle simulation method. The main source of complexity is the architecture: going from brain firing signals to muscle contractions to moving particles around."
So yes, it accurately simulates the muscle algorithm for these kinds of worms:
"Any time you do a simulation like this you're trying to make intelligent abstractions," John allows. "Unless you are simulating from first principles and moving quarks and gluons around, you're going to be glossing over some detail. So you try to make an abstraction that captures the essence of what you think is happening under the hood, and measure the results. In this case, the muscle model matches a basic level of our understanding of brain to muscle signaling and the physics of contraction/expansion in this worm, and the output (how the worm moves, displaces liquid, etc) looks pretty close to the real-world measurements!"
More details at the Open Worm site.
Please share this post::