Getting Ready for Deployment: Transformer-Enabled Robust Adaptive Traffic Signal Control
Bibliographic record
Abstract
Urban traffic signal control must contend with rapidly changing demand patterns, limited sensor coverage, and strict safety regulations - challenges often overlooked by lab-focused reinforcement learning (RL) research. We introduce eMARLIN-Transformer-Robust, a decentralized multiagent RL framework built for real-world deployment. We frame these issues as partial observability and out-of-distribution challenges. To reduce partial observability, each agent employs a Transformer encoder over a brief history of local observations augmented by neighbor embeddings. The architecture captures temporal context and extends agent's effective field of view. To achieve out-of-distribution robustness, we apply domain randomization techniques, exposing agents to diverse traffic scenarios to prevent overfitting. In a simulated field test spanning 75 out-of-training scenarios (225 runs), eMARLIN-Transformer-Robust reduces total stop delay by$17.7 \% \sim 27.8 \%$compared to the expert-tuned industry-standard TransSuite system. An ablation study shows that robust training further improves performance by 12.2 % over agents lacking it. These results demonstrate that only the combination of Transformer-based history encoding and robust training yields a single, adaptive policy capable of seamless, plan-free operation across diverse, real-world conditions.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".