EvoSignal: LLM-Guided Evolutionary Design of Modular Traffic Signal Control Programs

Leizhen Wang; Peibo Duan; Zhenlin Qin; Yancheng Ling; Jian Xu; Yue Wang; Hao Wang; Zhenliang Ma

cs.LGOct 7, 2026

EvoSignal: LLM-Guided Evolutionary Design of Modular Traffic Signal Control Programs

Authors: Leizhen Wang, Peibo Duan, Zhenlin Qin, Yancheng Ling, Jian Xu, Yue Wang, Hao Wang, Zhenliang Ma

Organizations: Monash University · KTH Royal Institute of Technology · Chinese Academy of Sciences · Zhejiang Dahua Technology · Southeast University

Abstract

Effective traffic signal control (TSC) requires policies that respond to changing traffic demand and network conditions while meeting different control objectives. However, adapting existing strategies often involves repeated manual design and adjustment, making it difficult to systematically explore better control rules for a target network. Large language models (LLMs) can automate this process, but directly using them to select signal phases leaves decision rules embedded in black-box models and incurs recurring inference costs and latency. This paper formulates TSC as a modular program design problem and proposes EvoSignal, an LLM-guided evolutionary framework using traffic knowledge and performance feedback. The modular representation separates traffic feature extraction, local phase prioritization, and optional network-based priority adjustment. Starting from several established strategies, EvoSignal improves programs through feedback on congestion and signal operation, retaining strategies with different performance trade-offs. The resulting programs operate without online LLM inference. Simulation experiments across five scenarios on two real-world road networks show that the selected default EvoSignal program reduces waiting time by 16.8--49.2% relative to the lowest waiting time achieved by the 20 conventional, reinforcement learning-based, and LLM-based baselines in each scenario. A program prioritizing travel time and queue length outperforms all 20 baselines on all three metrics in the search scenario and remains among the top three on each metric when transferred unchanged to the other four scenarios. These findings support automated design of inspectable control programs that transfer across the evaluated road networks and traffic demands.Code is available at https://github.com/georgewanglz2019/EvoSignal.

Figures & tables

Figure 1: Modular control program at decision step

t

. One controller is instantiated per intersection, with controller

i

expanded; the selected phases jointly control the traffic environment.

Figure 2: EvoSignal framework for evolving modular control programs from multiple initial strategies, guided by traffic diagnostics and reusable strategy summaries. The selected program controls signals without online LLM inference.

Figure 3: Traffic networks, selectable phases, and demand profiles. (a,b) Jinan and Hangzhou layouts. (c) Four selectable phases. (d,e) Observed demand on a common scale. (f) J-high on a separate scale. Arrivals use five-minute intervals.

Demand scenario	ID	Signalized intersections	Hours	Vehicles	Arrivals (vehicles / 5 min)
Demand scenario	ID	Signalized intersections	Hours	Vehicles	Mean	SD	Min	Max
Core demand profiles
Jinan 1	J1	12	1	6,295	524.58	98.53	256	672
Jinan 2	J2		1	4,365	363.75	74.66	237	493
Jinan 3	J3		1	5,494	457.83	46.22	363	544
Jinan Extreme †	J-high		1	24,000	2000.00	44.21	1,933	2,098

Table 1: Traffic demand statistics for the two road networks.

Traditional controllers
Method	J1			J2			J3
	TT (s)	QL (veh)	WT (s)	TT (s)	QL (veh)	WT (s)	TT (s)	QL (veh)	WT (s)
Random	597.62	687.35	99.46	555.23	428.38	100.40	552.74	529.63	99.33
FixedTime	481.79	491.03	70.99	441.19	294.14	66.72	450.11	394.34	69.19
Maxpressure	281.58	170.71	44.53	273.20	106.58	38.25	265.75	133.90	40.20
Reinforcement learning

Table 2: Traffic control performance on Jinan. Lower metric values and ranks are better.

Traditional controllers
Method	H1			H2
	TT (s)	QL (veh)	WT (s)	TT (s)	QL (veh)	WT (s)
Random	621.14	295.81	96.06	504.28	432.92	92.61
FixedTime	616.02	301.33	73.99	486.72	425.15	72.80
Maxpressure	325.33	68.99	49.60	347.74	215.53	70.58
Reinforcement learning

Table 3: Traffic control performance on Hangzhou. Lower metric values and ranks are better.

Figure 4: Relative improvement over Maxpressure for two EvoSignal programs evolved on J1 and reused unchanged across five scenarios. Positive values indicate improvement.

Figure 5: Evolution-framework comparison on J1 under the default objective. (a) Mean best-so-far score with SD bands over three runs. (b) Distribution of final best scores.

Figure 6: Component ablations on J1 under the default objective: (a) best-so-far score over 200 evolution iterations; (b) final best scores. Each setting uses three runs, with the same plotting conventions as Figure 5 .

Figure 7: Objective-space distributions under the default and TT–QL objectives on J1: (a) travel time versus queue length; (b) queue length versus waiting time. Additional views and plotting details are provided in Appendix E .

Component	EvoSignal	EvoSignal (TT–QL)
M1: features	Pressure, waiting and approaching traffic, upstream occupancy, downstream load, and time since a phase was requested.	Pressure, waiting and approaching traffic, maximum downstream occupancy, and time since a phase was observed active.
M2: service	Service-age compensation grows with both age and waiting demand.	Service-age compensation is capped; the waiting weight adapts to congestion.
M2: congestion	Downstream load attenuates the pressure term; upstream occupancy amplifies demand.	High downstream occupancy attenuates the combined phase score.
M2: persistence	Applies different discounts to current and competing phase priorities at regular decision points.	Adds a continuation bonus when approaching demand remains relatively high.
M3: network adjustment	Favors the through phase with greater network directional pressure; adds a common offset based on neighboring intersection loads.	No additional adjustment.

Table 4: Control rules in the three modules of the two selected J1 programs. Details are provided in Appendix C .

Appendix figures & tables8 assets

Supplementary material from the paper’s appendix.

Appendix

Configuration	Runs	Mean $\pm$ SD	Best run
EvoSignal	3	0.397314 $\pm$ 0.000489	0.397742
OpenEvolve	3	0.394427 $\pm$ 0.002986	0.397484
ShinkaEvolve	3	0.395840 $\pm$ 0.001751	0.397672
Without SPA	3	0.394508 $\pm$ 0.002684	0.396541
Without PSA	3	0.396607 $\pm$ 0.000639	0.397309
Without 3M	3	0.393988 $\pm$ 0.003230	0.397682

Appendix

Table 5: Final best search scores on J1 under the default objective (Section 4.1.2 ).

EvoSignal
Scenario	Programs	TT (s)	QL (veh)	WT (s)
J1	3	275.97 $\pm$ 1.86	160.64 $\pm$ 2.50	35.64 $\pm$ 1.52
J2	3	268.21 $\pm$ 0.51	101.75 $\pm$ 0.50	33.06 $\pm$ 1.43
J3	3	261.57 $\pm$ 1.00	128.56 $\pm$ 0.56	34.34 $\pm$ 0.96
H1	3	316.15 $\pm$ 2.18	61.76 $\pm$ 1.32	29.88 $\pm$ 0.47
H2	3	329.69 $\pm$ 1.62	177.71 $\pm$ 2.69	34.45 $\pm$ 4.92

Appendix

Table 6: Fixed-program results: EvoSignal, EvoSignal (TT–QL).

EvoSignal (H1, TT–QL)
Scenario	Programs	TT (s)	QL (veh)	WT (s)
J1	3	276.66 $\pm$ 3.74	162.54 $\pm$ 4.26	40.44 $\pm$ 1.67
J2	3	268.40 $\pm$ 0.26	102.08 $\pm$ 0.00	37.29 $\pm$ 1.36
J3	3	260.31 $\pm$ 1.06	127.36 $\pm$ 1.43	38.97 $\pm$ 2.06
H1	3	307.77 $\pm$ 0.59	55.86 $\pm$ 0.59	32.34 $\pm$ 2.17
H2	3	332.12 $\pm$ 3.71	186.25 $\pm$ 6.36	52.92 $\pm$ 2.34

Appendix

Table 7: Fixed-program results: EvoSignal (H1, TT–QL), ShinkaEvolve.

Input component	Role in program revision
Task and interfaces ( $G_{n}$ )	Define the control task, editable modules, and current M2 focus.
Parent and references ( $p_{a},R_{n}$ )	Supply executable code and alternative control examples.
Performance and SPA ( $m_{a}$ )	Report parent metrics, congestion patterns, and suggested changes.
PSA summary ( $b_{a}$ in $m_{a}$ )	Describe retained strategies and their measured trade-offs.
Output instructions	Request explanations followed by exact SEARCH/REPLACE edits.

Appendix

Table 8: Components of the recorded revision prompt.

Metric	Parent	Revised program
TT (s)	277.70	276.37
QL (veh)	163.89	161.73
WT (s)	36.52	35.64
Search score	0.392953	0.396055

Appendix

Table 9: Recorded parent and child performance for the illustrated revision.

Figure 8: Additional objective-space views of the default and TT–QL searches on J1: (a) three objectives; (b–d) pairwise projections over wider display ranges.

Setting	Value
Program evolution and LLM generation
Evolution attempts after initialization, $N$	200
Initial programs (full / without Multi-init)	11 / 1
LLM sampling probabilities (Flash / Pro)	0.7 / 0.3
Generation temperature / maximum output tokens	0.7 / 16,000
Population size / database archive size / islands	40 / 15 / 2

Appendix

Table 10: Core evolution and traffic-control settings.

Initial program	Design emphasis	Main control idea
Pressure–demand	Pressure and demand	Combine queue pressure, waiting vehicles, and approaching vehicles to rank phases.
Efficient pressure	Pressure-based control	Use efficient queue pressure directly as a simple initial priority rule.
Continuation-aware pressure	Phase continuation	Favor the current phase when approaching demand justifies continuation; otherwise follow pressure priorities.
Demand–supply balance	Demand and capacity	Combine estimated demand–supply ratios, saturation, and waiting vehicles to assess service urgency.
PID-inspired	Feedback control	Combine pressure, an accumulated-waiting proxy, and queue growth to respond to current and past congestion.
Fairness-aware	Service fairness	Increase priorities for phases not recently selected and reduce preference for a phase retained for a long period.

Appendix

Table 11: Initial control programs and their complementary design principles.

Explore similar work

HiLLTS: Zero-Shot Hierarchical LLM-Guided Traffic Signal Control for Sustainable Transportation
Jul 17, 2026Yue Ding, Tendai Mukande, Mingming LiuTraffic Signal Control Congestion
Research Ireland Centre for Researching Training in Machine Learning at Dublin City University
DGLight: DQN-Guided GRPO Fine-Tuning of Large Language Models for Traffic Signal Control
Apr 28, 2026Chenbo YuTraffic Signal Control Reinforcement Learning Control
National University of Singapore Singapore
LLM-Augmented Traffic Signal Control with LSTM-Based Traffic State Prediction and Safety-Constrained Decision Support
Apr 26, 2026Jiazhao ShiTraffic Signal Control Accurate Traffic Forecasting
Tandon School of Engineering, New York University, 6 MetroTech Center, Brooklyn, NY 11201, USA

Jul 17, 2026, cs.AIJ/K move · Enter open · S save

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

Yue Ding, Tendai Mukande, Mingming Liu

Research Ireland Centre for Researching Training in Machine Learning at Dublin City University

Traffic Signal Control Congestion Urban Mobility Trajectory-Level Credit Ridge

Open paper AskPDF arXiv

Urban traffic congestion significantly increases fuel consumption, greenhouse gas emissions, and commuter delays, resulting in substantial economic losses and environmental harm in modern cities. Traditional traffic signal control strategies such as fixed-time scheduling, actuated control, and reinforcement learning (RL)-based methods, offer different degrees of adaptability; however, RL-based methods can require extensive retraining, careful reward design, and substantial simulation data when transferred across networks or demand regimes. To address these challenges, we propose HiLLTS, an LLM-guided traffic signal control framework that employs a hierarchical three-layer architecture consisting of a central coordination agent, a district layer and multiple cluster-level intersection agents. Experimental results demonstrate consistent improvements in both congestion and environmental performance. Compared with the strongest non-LLM baseline in each scenario, HiLLTS reduces average waiting time by 36.73% under the low-congestion scenario and 14.71% under the high-congestion scenario, while reducing average CO2 emissions by 7.87% and 8.57%, respectively. Larger gains are observed against weaker baselines: under low congestion, HiLLTS achieves reductions of up to 18.00% in emissions and 62.07% in waiting time relative to Fixed-Time control; under high congestion, reductions of up to 28.89% in emissions and 40.36% in waiting time are observed relative to Max Pressure. The ablation study further validates the contribution of LLM-guided coordination over rule-based control

Abstract

Explore similar work

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

DGLight: DQN-Guided GRPO Fine-Tuning of Large Language Models for Traffic Signal Control

LLM-Augmented Traffic Signal Control with LSTM-Based Traffic State Prediction and Safety-Constrained Decision Support