Research papers · 2025.06.22
Distributional Soft Actor-Critic with Harmonic Gradient for Safe and Efficient Autonomous Driving in Multi-lane Scenarios
Distributional Soft Actor-Critic with Harmonic Gradient for Safe and Efficient Autonomous Driving in Multi-Lane Scenarios
Overview
DSAC-H combines distributional soft actor-critic with harmonic policy updates that reduce conflicts between efficiency and safety gradients. Multi-lane simulations evaluate driving performance and constraint violations during safety-oriented reinforcement learning.
Author-supplied source information
Author-supplied information on arXiv.
IEEE Intelligent Vehicles Symposium (IV 2025)
Version: arXiv v1 — manuscript revision