CARL: autotelic RL agent discovers and steers Lenia solitons in closed loop
pyoudeyer · x · 2026-11-17
Researchers from Inria, Tufts (Michael Levin's lab) and Inria's Moulin-Frier team present CARL (ALife 2026): an autotelic RL agent that samples its own goals and learns goal-conditioned policies to intervene in Lenia, a continuous cellular automaton, via minimal local perturbations. It discovers stable solitons across many update rules at higher rates than heuristic baselines, steers existing solitons with few interventions, and follows human directional commands in real time to guide patterns through a maze. Paper and code are open, with in-browser demos. Unlike open-loop exploration methods, CARL's closed-loop framework offers a new paradigm for exploring and controlling self-organizing complex systems.
More from Research
- Physicist pans AI "superconductivity breakthrough" as repackaged 1975 result — ZeeshanZiaML · 2026-11-17
- Studying retrosynthetic disconnection from Vogel's textbook: "really freaking cool" — CatAstro_Piyush · 2026-11-17
- AI drives 30% surge in arXiv theoretical physics submissions; models solve ~60% of private-problem benchmark — binarybits · 2026-11-17
- FrontierCS 2 benchmark challenges AI agents to beat published CS research blind — arena · 2026-11-17
- Statistician: Three Systemic Incidents in One Month Is Enough to Reject Low Base Rate (p=0.002) — wfithian · 2026-11-17
- Isomorphic Labs raises $5B, its second mega-round in six months — The Decoder · 2026-11-17