HiLLTS: Zero-Shot Hierarchical LLM-Guided Traffic Signal Control for Sustainable Transportation

Reinforcement learning for traffic signal control in large-scale transportation networks: a systematic literature review

Effective Traffic Signal Control (TSC) in large-scale transportation networks is essential for enhancing urban mobility, reducing congestion, and improving safety. However, traditional control methods often fail to effectively address the complexity, dynamic conditions, and multimodal demands of modern urban traffic systems. In recent years, Reinforcement Learning (RL) has emerged as a promising solution for achieving adaptive and scalable TSC. This paper presents a systematic and up-to-date review of RL-based methods for large-scale TSC. We analyze representative studies published between 2013 and 2025, presenting a comprehensive analysis of traffic simulation environments, transportation modalities, and advances in methodologies. Key aspects include multi-agent paradigms, state and action representations, reward mechanisms, RL frameworks, as well as advanced representation learning and cooperative strategies for large-scale transportation networks. We also provide a critical discussion on performance evaluation and opportunities for improvement, and conclude by summarizing the current challenges and outlining future research directions. This review aims to inform and guide the development of next-generation RL-based TSC systems that promote sustainable, safe, and efficient urban transportation.

A Decentralized Traffic Signal Control Framework Using Multiagent Reinforcement Learning With Real-World Operational Constraints

Deep reinforcement learning-based traffic signal control in multi-intersection environments: a comparative study of DQN variants

The findings demonstrate the potential of DRL-based traffic signal control in controlled simulation conditions and highlight that algorithm performance is strongly influenced by traffic policy design and environmental complexity.

Low-Carbon Urban Freight Optimization: Per-Order Adaptive Mode Mixing with Demonstration-Regularized Constrained Reinforcement Learning

Urban last-mile delivery is a rapidly growing source of city-centre emissions, and decarbonizing it without eroding service quality has become imperative for climate goals. Operators are turning to multimodal systems that integrate road vehicles, off-peak metro freight, and electric drones—yet the optimal delivery channel varies dynamically with location and time. Current RL-based schedulers handle constraints via manually tuned penalty weights, lacking formal safety guarantees, and the feasibility of online carbon-cap enforcement under partial observability remains an open question. To address this, we model the problem as a Constrained Markov Decision Process (CMDP) and propose a demonstration-regularized Lagrangian deep RL algorithm. Our approach learns an online policy that is model-free at deployment—it controls emissions in expectation against a hard carbon budget, makes per-order decisions using only state observations, and operates without an emission model at test time (the demonstrator used at training time does access the emissions model, so “model-free” refers strictly to the deployment phase). Experiments on synthetic benchmarks and a Nanjing-inspired scenario—grounded in real metro topology and population-weighted demand—show that our policy achieves emissions within 1.3% of the offline optimum. It robustly tracks a ±17% carbon-budget band across a threefold daily volume range and a threefold city-scale range, with zero per-instance tuning. By contrast, a standard PPO with fixed penalty weights consistently degrades to single-mode selection. Our findings suggest that hard carbon budgets can be controlled in expectation online at modest cost—a step toward operator-facing low-carbon logistics whose average emissions honour a binding carbon budget, though external validation on operational data and a risk-sensitive formulation that upgrades this average control into per-day compliance are still required before deployment.

Soft Actor-Critic based regional traffic signal control in connected environment and its application in priority signal control

TLMOF: An Operating Framework Inspired Hybrid Kernel for Signal Control

Urban traffic congestion imposes significant economic and environmental costs worldwide. Conventional adaptive signal controllers-including actuated systems and deep reinforcement-learning (DRL) approaches-either collapse under saturation or lack formal verifiability, making them unsuitable for safety-critical infrastructure. This paper presents the Traffic Light Management Operating Framework (TLMOF), a signal control architecture that maps classical operating-system (OS) scheduling theory directly onto intersection management. Vehicles are modeled as processes, approach lanes as ready queues, and the intersection controller as a CPU scheduler. The TLMOF Hybrid Kernel selects signal phases by maximising a modified pressure function combining Max-Pressure (MP) throughput control with Weighted Fair Queuing (WFQ) anti-starvation. Gridlock is formally characterised as a Coffman deadlock, detected via Wait-for-Graph (WFG) depth-first search in O(V+E) time, and resolved through a formally specified Flush Phase. With fairness weight β > 0, the kernel provably guarantees bounded delay for all movements (AntiStarvation Theorem). SUMO 1.26 validation across three demand scenarios demonstrates a 377× average-delay reduction over fixed-time under saturation and a 25% Jain's Fairness Index improvement over pure Max-Pressure (0.909 vs. 0.727) under asymmetric demand while reducing average delay by 56% (3.62 s vs. 8.15 s). A sensitivity analysis of (α, β) confirms that the recommended default (0.7, 0.3) achieves a robust balance between throughput and fairness.