Evaluating the Transfer of Co-Evolved Communication from 2D to 3D Simulation
Co-evolved robot communication transfers from 2D to 3D physics, but short episodes underestimate success.
The paper tests whether a communication mechanism co-evolved by two robotic agents in a discrete 2D simulator still functions after transfer into a 3D simulator with real physics. Success increased approximately linearly with episode budgets from 2,000 to 6,000 physics steps, with no plateau, so shorter episodes can underestimate controller performance. In both simulators the social cue mainly helped agents recover from jams rather than guide navigation, and eight evolutionary runs showed a consistent role asymmetry that controlling agent processing order ruled out as a physics-engine artifact.
- Co-evolved communication is transferred from a discrete 2D simulator to 3D physics.
- Success rises roughly linearly between 2,000 and 6,000 physics steps.
- The social cue mainly assists unjamming rather than navigation.
- Role asymmetry is consistent across eight runs and is not an engine artifact.
From the original article
1 min read at huggingface.co
This work examines the transfer of a co-evolved communication mechanism between two robotic agents from a discrete two-dimensional (2D) simulator to a three-dimensional simulator with real physics (3D). The study focuses on whether a communication mechanism co-evolved in a 2D environment retains its functional role after transfer to a 3D physics-based simulator.
By Fernando Montes-Gonzalez · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.