A curriculum that progressively adds traffic and a collision penalty, plus a speed-encouraging reward, increases the average speed of a PPO+VAE driving agent in CARLA while keeping lap distance comparable to a baseline.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CuRLA: Curriculum Learning Based Deep Reinforcement Learning for Autonomous Driving
A curriculum that progressively adds traffic and a collision penalty, plus a speed-encouraging reward, increases the average speed of a PPO+VAE driving agent in CARLA while keeping lap distance comparable to a baseline.