Evolved e-puck robots show communication emerges with information value
uv-mex · hf · 2026-10-12
An evolutionary robotics study examines when communication emerges between autonomous agents. Two e-puck robots (rebuilt in PyBullet from their Webots model) must locate a feeding zone detectable only by driving over it; each can emit a one-bit directional signal, controlled by a GRU network with a residual path.
- When robots receive a compass pointing to food, signaling disappears entirely.
- Without a compass, three independent runs in a small arena evolved different strategies: referential signaling, a permanent beacon, or no signal use.
- In larger arenas the referential-signal strategy generalizes best; evolving directly in the large arena doesn't increase communication frequency but clearly favors referential signaling over solitary search.
- Within an Embedded Action Selection Architecture (EASA), innate reflexes constrain the network and hurt performance, and evolution consistently weakens them.
Conclusion: communication emerges in proportion to the value of information it transmits; innate scaffolding designed for solitary agents can impede social tasks.
More from Embodied
- TouchScale: 500-Hour Vision-Tactile Dataset Lifts Robot Manipulation Success from 22.5% to 57.5% — 2077AIDataFoundation · 2026-10-12
- Robotic systems and real-time connectivity let surgeons operate on patients thousands of km away — CurieuxExplorer · 2026-10-12
- Humanoid robots learn Nadal and Federer's tennis styles straight from broadcast footage — CurieuxExplorer · 2026-10-12
- Humanoid hands are the hardest problem: Shadow, Mimic and PSYONIC specialize in them — xmercury_one · 2026-10-12
- Claude ports Crash Team Racing natively to the New 3DS with stereoscopic 3D — zando95 · 2026-10-12
- Atomic Machines exits stealth with Matter Compiler, an AI-native manufacturing system — mikeflache · 2026-10-12