AI‑Powered Urban Mobility Forecasting: Adapting City Transportation to Post‑Pandemic Trends

The COVID‑19 pandemic upended traditional commuting patterns virtually overnight. In the years since, remote work and flexible schedules have fundamentally changed how people move around cities. Rush‑hour peaks have flattened, transit ridership dropped then slowly rebounded, and many former commuters now travel at different times or not at all on certain days. City transportation agencies are grappling with a new question: how do we plan for transit and road usage in this “new normal” when past trends no longer apply?

This is where AI-powered mobility forecasting comes into play. By leveraging machine learning on vast mobility datasets—from traffic sensors, public transit smart cards, cellphone location data, and more—cities can predict emerging travel patterns and adapt transportation services proactively. AI excels at detecting subtle shifts and combining diverse factors (like remote work rates, school schedules, weather, etc.) to forecast demand. The result is a more responsive transportation network, tuned to post‑pandemic realities.

The Shifting Commute Paradigm

First, it’s important to understand how stark the change has been. Before 2020, only about 5% of U.S. employees worked remotely most of the time. During the pandemic, that spiked above 60%, and even in late 2022, a stable 35–40% of the workforce continued working from home at least one day a week. In practice, many office workers now commute only two or three days per week, often Tuesday through Thursday, with Mondays and Fridays at home.

This hybrid pattern breaks the old mold of five-day-a-week peak commuting. Data confirms the effect: as of early 2025, office occupancy is only ~54% of pre‑pandemic levels on average, yet transit ridership has climbed back to about 85% of pre‑pandemic trips. That means people are using transit and roads for a variety of purposes beyond just going to the office—travel for errands, appointments, or social reasons has filled some of the gap. Weekend and midday ridership are up in many cities, while Monday/Friday rush hours remain lighter. These complex changes are hard to capture with traditional planning models, which is why agencies are embracing big data and AI to continually update their forecasts.

Learning from Data in Real Time

Artificial intelligence can sift through massive, real-time data streams to identify new patterns far faster than manual analysis. For example, machine learning models can ingest public transit fare card swipes, rideshare trip records, bike-share usage, and highway traffic speeds, and then find correlations—perhaps discovering that midday travel between residential suburbs and local retail centers is rising as work-from-home employees go out for lunch or errands.

AI can also factor in external data like COVID case rates, gas prices, or even Zoom meeting statistics to predict how many people will travel. Importantly, these models are not one-time; they continuously learn. If a city observes that every time there’s a major event or a shift in public health guidance more people return to offices, the AI model will incorporate that into forecasts.

This level of real-time modeling aligns with Urban SDK’s mobility intelligence platform, which enables agencies to integrate live data streams—from traffic, transit, and demographic sources—into continuously updating forecasts and dashboards.

Adapting Transit Schedules and Routes

One direct application of AI forecasting is optimizing transit operations for new demand patterns. Instead of running the same number of trains every rush hour, agencies can use AI insights to allocate service when and where it’s needed most. For example, if data shows the 7am downtown express is underutilized but the 11am and 2pm trains are getting crowded (a possible scenario with staggered office hours and hybrid workers), schedules can be tweaked accordingly.

Urban SDK supports this level of decision-making with travel time delay and traffic volume data, helping planners identify when and where congestion builds or transit demand shifts. This data can guide changes in frequency, rerouting, or even the creation of midday microtransit options.

Traffic Management in the Hybrid Era

It’s not just transit—city traffic engineers are also using AI to manage roads amid fluctuating commuting patterns. Traffic volumes dropped significantly in 2020, then rebounded as people avoided transit, and are now finding a new equilibrium. Interestingly, remote work has brought some environmental benefits: one study estimates that each 1% increase in remote work reduces urban transport emissions by ~1.8% as fewer people drive.

However, there are also counter-trends like increased freight and delivery traffic from online shopping and more off-peak car trips. AI-based traffic modeling systems can better predict these flows. For instance, if a city knows that many employees now drive to the office only on Wednesdays, it can adjust traffic signal timing or HOV lane operations on those days specifically.

Urban SDK enables this through real-time traffic speed data and scenario-based modeling tools. Cities can simulate "what-if" scenarios and adjust traffic control strategies dynamically, whether that’s for signal timing or congestion pricing trials.

Case Study – Data-Driven Recovery

A concrete example comes from New York City Transit’s pandemic recovery efforts. The Metropolitan Transportation Authority (MTA) brought in an AI analytics platform to help regain riders and deploy service efficiently. The system ingested millions of data points from bus passenger counters and MetroCard swipes and learned the new usage patterns.

Planners discovered, for example, that some residential routes had different peak times than pre‑2020, so they adjusted timetables. The result was that buses were less often empty or overcrowded—resources were rebalanced to meet actual needs. This type of real-time responsiveness reflects what Urban SDK delivers through our traffic congestion analytics, allowing agencies to match service to demand even as behavior shifts.

In Summary

AI-powered forecasting is becoming an indispensable tool for city transportation planning in the post-pandemic era. Traditional models built on past decades’ habits simply don’t fit a world where remote/hybrid work, on-demand travel, and other flexible arrangements reign.

By using AI to continuously learn from fresh data, cities can detect how travel is evolving—and then respond with nimble adjustments. Urban SDK’s data ecosystem—spanning mobility intelligence, traffic safety, and predictive analytics—helps cities modernize their transportation strategies in line with real-world behavior.

The ultimate benefit is to travelers: buses and trains arrive when needed, traffic flows are smoother, and planners are prepared—not surprised—by new trends.

‍