OverFlowLight: Real-Time Gridlock Prevention and Traffic Signal Optimization for Urban Intersections

2026-06-29T08:52:13Za846694e891e36f30b983d1f2cfee8613071b7703dc584e503c01cdebdbb9774
DartsFourier-neural-operatorMarkovian-banditsPairSAEPrism-TransformerRANSACattentionbandit-algorithmsboundary-condition-fidelitycomputer-visionfoundation-modelsgeological-carbon-storageinlier-scale-marginalizationmechanistic-interpretabilitynonlinear-Schrödingeroverflow-detectionprotein-structurereal-time-systemsreinforcement-learningreservoir-simulationrobust-estimationsmart-citytime-seriestraffic-signal-controltransformer-architecture

What happened

Batch of CS/ML arXiv submissions (29 Jun 2026) covering applied ML, algorithms, and scientific computing. Highlights: OverFlowLight — a deployed, real-time traffic-signal framework using camera/radar sensing and hybrid rule/RL control that inserts overflow phases; deployments across 43 intersections report a 60.4% reduction in overflow incidents and +18.2% network throughput. RANSAC Scoring Done Right — closed-form marginalization of inlier scale under an Inverse-Gamma prior yields a scale-free RANSAC score computable in O(N log N) and improves robustness to threshold miscalibration. Unified D

Why it matters

A reviewed impact interpretation has not been published for this record.

Evidence and limitations

Source ID
arxiv_cs_lg
Record identifier
a846694e891e36f30b983d1f2cfee8613071b7703dc584e503c01cdebdbb9774
Enrichment time
2026-06-29T08:52:13Z
AI-assisted enrichment
Yes

This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.