Evolved Virtual Creatures (1994)

Channel:karlsims · Length:4:08 · Published:2019-12-16 · Views:43117

Watch on YouTube

Summary

A restored 1994 video of Karl Sims' research on evolved virtual creatures, where simulated block creatures are bred through artificial evolution to swim, walk, follow and compete. It is a landmark demonstration of genetic algorithms in physics simulation.

Why HN recommends it

HN brings it up as a classic example of evolutionary algorithms, including how the creatures exploited physics-simulation bugs, which parallels metric gaming in AI.

they would absolutely exploit any bugs in the physics simulation that they could — HN, 2024

Machine learning & deep learning · Computer graphics & rendering · Evolution & genetics

Subtitles (excerpt)

this demonstration shows virtual creatures that were evolved to perform specific tasks in simulated physical environments swimming speed was used to determine survival most of the creatures are results from independent evolutions some developed strategies similar to those in real life once they're evolved multiple copies of these creatures can be made and simulated together in the same environment the next group of creatures were evolved for their ability to move on a simulated land environment with gravity and friction some simple solutions with just two parts were found some seemed like they could use some assistance while others were fairly efficient such as this rowing like behavior here is an odd cousin of the previous a mutation caused him to tumble some creatures evolved to incorporate contact sensors in their control systems here is another inch worm like creature that tends to go in circles this was actually a creature first evolved for its ability to swim in water then later put on land and evolved further a successful sidewinding ability resulted here is one with a hopping style the protrusions on its arms seemed to help prevent it from tipping over this was the fastest with a successful galloping like stride this group was evolved for their

On Hacker News

What was shown in the video seems to be something like Karl Sims' work[0], which is completely different (genetic algorithm for optimising black-box locomotion within a physics simulation framework) from AI image generation / what we're currently calling "Ghiblification". [0] [VIDEO] — pixelpoet, 2025-03
The discussion reminds me of Karl Sims’ Evolved Virtual Creatures. In the Siggraph talk he mentioned that they would absolutely exploit any bugs in the physics simulation that they could. That’s all that’s happening with this article’s LLMs - the metric is dictator, and simply and predictably leads to any and all possible behaviors that achieve higher scores. [VIDEO] You know what’s really weird … — dahart, 2024-12
Beautiful video. I would love to see this animation technique combined with an evolution simulation similar to Karl Sims' Evolved Virtual Creatures project: [VIDEO] Previous HN discussion: [LINK] — owenpalmer, 2024-07

Related videos