JL-GAT extends grounded action transformation to multi-agent traffic signal control by feeding each agent's grounding models with neighboring state and action information, reducing the sim-to-real gap in simulated rainy and snowy conditions.
Antoine Cully, Jeff Clune, Danesh Tarapore, and Jean-Baptiste Mouret
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Joint-Local Grounded Action Transformation for Sim-to-Real Transfer in Multi-Agent Traffic Control
JL-GAT extends grounded action transformation to multi-agent traffic signal control by feeding each agent's grounding models with neighboring state and action information, reducing the sim-to-real gap in simulated rainy and snowy conditions.