Tag
Introduces SPOT (Sampling Policy Observation Tree), a model-agnostic framework for interpreting deep reinforcement learning policies by constructing finite-horizon lookahead trees that expose downstream consequences of actions, demonstrated on SUMO-RL traffic-signal control.
This paper presents a graph-based traffic signal control interface using a shared graph neural network to assign scores to movements, with deterministic phase construction via incidence matrices. Experiments evaluate transfer across synthetic and city road networks, showing feasibility but sensitivity to distribution shifts.
OverFlowLight is a real-time framework that prevents traffic gridlock by detecting queue overflow using multi-modal sensing and inserting dedicated overflow phases via a hybrid rule-based and RL controller. Deployed across 43 intersections, it reduces overflow incidents by 60.4% and increases network throughput by 18.2%.
The paper proposes an active inference controller for adaptive traffic signal control in noisy IoT environments, outperforming DQN in idle times and CO2 emissions under sensor occlusion and adverse weather conditions.
The paper introduces OracleTSC, a method using oracle-informed reward hurdles and uncertainty regularization to stabilize reinforcement fine-tuning of LLMs for traffic signal control. It demonstrates significant improvements in traffic flow metrics on the LibSignal benchmark using LLaMA-3-8B while maintaining interpretability.