The Traffic Optimization Problem
Traffic signal coordination is fundamentally a complex optimization problem. In a network of hundreds or thousands of intersections, each signal change affects traffic throughout the connected network. Traditional fixed-timing plans — calculated based on average conditions and updated infrequently — cannot respond to the variation in traffic patterns that occurs daily, seasonally, and in response to events. Adaptive systems that respond to real-time conditions offer significant performance advantages.
Sensing and Data Collection
Effective traffic optimization requires comprehensive sensing. Inductive loop detectors embedded in pavement count vehicles and measure speeds. Video-based detection using computer vision can provide richer data — vehicle classification, turning movements, pedestrian and cyclist counts — from camera infrastructure that also serves other purposes. Increasingly, aggregated probe data from navigation apps provides a real-time picture of vehicle speeds across entire networks without physical infrastructure at every point.
Reinforcement Learning for Signal Control
Reinforcement learning approaches — where AI agents learn optimal signal control policies through simulated and real-world experience — are showing promise as alternatives to rule-based adaptive systems. Unlike rule-based systems designed by traffic engineers for specific intersection configurations, RL agents can discover control strategies that human experts would not have designed while performing well in the specific conditions of their deployment environment. Glidonce uses RL-based signal optimization at scale across city networks.
Measuring Impact
The impact of traffic optimization is measurable through before-and-after travel time analysis using probe data, intersection delay measurements, and fuel consumption and emissions estimates derived from vehicle operating data. Cities that have deployed Glidonce's platform consistently report 15-25% reductions in average intersection delay and 10-20% improvements in network-wide travel times in the deployment area. These improvements translate directly to economic value through reduced congestion costs and environmental value through emissions reductions.