EvoFlock: evolved inverse design of multi-agent motion
Craig Reynolds unaffiliated researcher [email protected] · 2026
Read EvoFlock: evolved inverse design of multi-agent motion by Craig Reynolds unaffiliated researcher [email protected] (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-25280.txt.
Description
Simulating flocks, crowds, and traffic means tuning many interacting numbers, and changing one behaviour often disturbs another. Craig Reynolds presents EvoFlock in 2026, an automatic tuning method that flips the task around. Instead of hand adjusting a flocking model, the user states the desired group behaviour as a scoring function. A genetic algorithm then evolves the model's parameters until the simulated flock scores well, rewarding good spacing, a target speed, and obstacle avoidance.
One surprise emerged from the evolved solutions. The striking alignment of real bird flocks, where neighbours match direction, appeared on its own once proper spacing was maintained. Reynolds frames the method as inverse design for multi agent motion generally, covering herds, schools, and vehicle flows as well as birds. Code and data are released openly alongside the work.
Catalog record
| CREATOR | Craig Reynolds unaffiliated researcher [email protected] |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 34 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.25280