From paratransit optimization to school bus routing — building AI that moves real communities.
Project WebsitePublic transit agencies across America face a common problem: delivering reliable service with tight budgets, variable demand, and aging infrastructure. The challenge is especially acute beyond dense urban cores. Paratransit riders depend on services that are expensive to operate. Rural communities near new employment hubs lack transit entirely. School districts manage hundreds of buses with little technological support. And when disruptions hit — a breakdown, a road closure, a pandemic — agencies fall back on ad-hoc decisions made under time pressure.
The SCOPE Lab has spent over a decade building AI systems that address these challenges — not as research prototypes, but as software tested on real vehicles, serving real riders, and evaluated by the transit professionals who use them daily.
Our transit research began in 2015 with an NSF-funded effort to build a decision support system for Nashville’s public transit network. Transit Hub brought together real-time vehicle location data, passenger counts, and schedule information into a single analytics platform, giving dispatchers visibility into system performance across multiple timescales. The system predicted bus arrival times using SVM-Kalman models, reducing prediction error by 30–65% compared to schedule-based baselines.
Transit Hub established two principles that have guided everything since. First, transit AI must be built on real operational data — not synthetic benchmarks. Second, systems must serve the people who actually run transit: dispatchers, supervisors, and planners whose expertise is essential but whose tools have not kept pace.
When the Department of Energy began funding smart transportation research in Chattanooga — through HD-EMMA ($1M, 2018–2021), AI-Engine ($1.8M, 2020–2024), and Autonomy-Aware ($6M, 2024–2027) — we extended this foundation to energy-aware fleet optimization. Working with CARTA’s mixed fleet of electric, hybrid, and diesel buses, we built real-time energy consumption models that predict fuel and electricity usage as a function of route, schedule, weather, and driving patterns. As fleets electrify, optimizing for energy cost alongside ridership and on-time performance becomes critical.
Every planning and operations algorithm we build depends on the ability to simulate what will happen before committing to a decision. A dispatcher considering where to position a reserve bus needs to evaluate dozens of possible futures. A microtransit router confirming a trip needs to estimate how that commitment will affect later requests. A school district rerouting around a road closure needs to see which alternative minimizes delay across the entire network.
This is the role of city-scale digital twins — virtual representations of the transit system that are continuously updated with real data, maintain predictive models of how the system will evolve, and allow decision-makers to evaluate alternatives through counterfactual simulation. In transportation, digital twins integrate heterogeneous data sources (vehicle positions, passenger counts, traffic conditions, weather, historical patterns) with system dynamics (route structures, schedules, fleet constraints) and decision processes (dispatch rules, routing algorithms) into a coherent planning environment. They span both long-term planning — where should new zones or stations be located? — and real-time operations — which bus should be dispatched right now?
We have built several simulation and evaluation platforms that serve as digital twins for our transit partners. TRANSIT-GYM provides a realistic simulation environment for evaluating dispatch and routing strategies on actual transit networks, allowing us to test hundreds of disruption scenarios without risking real service quality. BTE-Sim achieves approximately 13x speedup over conventional simulation while maintaining accuracy, making it practical to run the kind of large-scale scenario evaluation that real-time decision-making demands.
These digital twins are not standalone tools — they are embedded directly into our operational systems. When our Monte Carlo Tree Search planner decides where to position WeGo’s reserve buses, it rolls out thousands of simulated futures within the digital twin, evaluating how each positioning choice performs under different disruption scenarios. When SchoolRide reroutes buses around a road closure, it uses the twin to evaluate candidate routes and select the one that minimizes delay propagation across the entire network. When SmartTransit.AI decides whether to confirm a microtransit trip, the twin provides the demand forecasts and capacity projections that inform the reinforcement learning agent’s accept/reject decisions.
The key challenges we are actively addressing — as outlined in our research roadmap for transportation digital twins — include maintaining correctness under non-stationary conditions (demand patterns shift, new construction changes traffic), bridging the sim-to-real gap so that decisions optimized in simulation transfer reliably to the field, and scaling these models to operate across multiple agencies and cities simultaneously.
Before a single vehicle can be dispatched, a more fundamental question must be answered: where should on-demand service exist, and how should it connect to the fixed-route network?
This is the problem at the heart of CARTA Connect — the AI-Powered Autonomy-Aware Neighborhood Mobility Zones project ($7M DOE, 2024–2027), a collaboration between CARTA, Vanderbilt, Penn State, Cornell, and Spark the Firm. The project reimagines public transportation for mid-sized cities through an AI-driven multi-modal framework that integrates fixed-line buses and on-demand microtransit into a single responsive system.
At its core is the concept of neighborhood mobility zones — virtual, data-driven operational areas that dynamically organize transportation resources without requiring costly new infrastructure. These zones centralize around Chattanooga’s downtown core, connecting neighborhoods, universities, hospitals, and economic centers. Within each zone, riders can access on-demand microtransit shuttles, electric car shares, and bike shares alongside CARTA’s fixed-route buses. Zone boundaries are not drawn arbitrarily; our AI algorithms identify optimal zones by analyzing travel patterns, population density, existing transit coverage, available resources, and revenue sustainability — placing microtransit where it genuinely complements fixed-route service rather than competing with existing lines. The system uses real-time multi-modal routing, predictive demand modeling, and autonomy-aware fleet planning to deploy resources where they are most effective.
Beyond the technical design, the project is building sustainable business models — including subscription-based mobility packages and monetization of CARTA’s 4,200 parking spaces — to ensure long-term viability. The goal is to increase total trips served through the system by five times, providing a replicable blueprint for mid-sized U.S. cities to modernize transit using AI.
Once the network is designed, the harder problem begins: running it well, every day, in real time.
Paratransit and microtransit present some of the hardest vehicle routing problems in practice. Each new ride request must be inserted into an existing schedule in real time, balancing wait times, vehicle capacity, driver hours, and detour limits — while coordinating transfers where microtransit meets fixed-route service.
Our SmartTransit.AI platform handles this through a modular architecture that separates routing algorithms from the operational interface, so agencies can adapt the system to their specific service rules. At the algorithmic core, our offline-online hybrid approach computes baseline route plans from advance bookings, then continuously adapts as same-day requests, cancellations, and traffic updates arrive. The offline phase solves a vehicle routing problem with time windows; the online phase uses reinforcement learning to make non-myopic accept/reject decisions — balancing the cost of serving a request now against the opportunity cost of committing capacity that might be needed later.
A key recent innovation is prompt confirmation with continual optimization (ICCPS 2026). Riders need immediate confirmation when they book a trip — they cannot wait for a batch optimizer. Our system confirms trips quickly using insertion search, then continuously improves the full schedule in the background. Reinforcement learning guides confirmation decisions, learning when to accept immediately versus when expected future demand makes declining worthwhile. On real microtransit data, this achieves significantly better trade-offs between confirmation speed and service rates than either pure optimization or greedy insertion alone.
Even well-designed fixed-route services degrade without effective real-time operations. Buses break down, traffic builds unexpectedly, and demand surges at particular stops. The question facing dispatchers is: where should reserve vehicles sit, and when should they move?
Working with WeGo Public Transit in Nashville, we formulated this as a coupled stationing and dispatch problem solved via Monte Carlo Tree Search. The stationing component pre-positions reserve buses across the network in anticipation of disruptions; the dispatch component deploys them as disruptions materialize. The planner searches over thousands of simulated futures within our digital twin, making proactive decisions that account for uncertainty rather than reacting after service has already degraded. In validation on three years of WeGo data, this approach served 7% more passengers while reducing deadhead miles by 42% compared to greedy heuristics. This work received the Best Paper Award at ICCPS 2024.
To put these recommendations in dispatchers’ hands, we built Vectura — a real-time dashboard that combines live vehicle tracking, predictive disruption alerts, headway visualization, and explainable dispatch recommendations. Dispatchers see where buses are, where delays are developing, and what the AI recommends — with full authority to override any suggestion.
Our ongoing Fixed-Line Transit 2.0 project with WeGo, funded by USDOT, tackles a specific operational headache: bus bunching. High-frequency routes like Nashville’s Route 55 Murfreesboro see buses cluster together during peaks and leave long gaps during valleys, despite stable underlying demand. Working with Penn State, Swiftly, and LYT, we are building a decision support system that combines real-time headway sensing with adaptive transit signal priority to smooth service and reduce wait times.
We also developed graph neural network models that predict occupancy and ridership across WeGo’s network by incorporating spatial relationships between routes and stops — enabling predictive overcrowding alerts and evidence-based capacity planning.
Transit systems are inherently vulnerable to disruption, and our work has repeatedly confronted this reality — with the digital twin providing the simulation backbone for each response.
When COVID-19 struck in 2020, ridership collapsed but essential workers still depended on service. We conducted a comprehensive analysis of the pandemic’s impact across Nashville and Chattanooga, documenting initial ridership drops of 66% and 65% respectively. Recovery patterns differed significantly by socioeconomic group. We also built models predicting transit load to estimate social distancing violation probabilities, giving agencies data for service adjustment decisions during a period of extraordinary uncertainty.
SchoolRide (ICCPS 2026) addresses disruption management for school bus operations — where delays directly affect children’s safety and parents’ schedules. Working with Williamson County Schools, which runs 228 buses serving over 50,000 students, we built a platform that integrates live vehicle telemetry, real-time driver status updates, and dynamic rerouting. When a disruption hits — a road closure, a breakdown, severe weather — the system evaluates candidate reroutes within the digital twin and selects the one that minimizes delay propagation across affected routes. SchoolRide consistently outperforms the human-intuition baselines that districts currently rely on.
We measure our work by what happens when it meets the real world.
The Clifton Hills Microtransit Pilot (2024) tested SmartTransit in the Clifton Hills neighborhood of Chattanooga over 27 service days. A vehicle, driver, and booking agent operated between 9 AM and 3 PM, providing on-demand service as a feeder to CARTA’s fixed-route network. The results, reported at IJCAI 2024, showed substantially higher shared-ride rates and reduced vehicle miles compared to baselines — validating microtransit as a last-mile complement to high-capacity fixed routes.
The SmartTransit system was also piloted for CARTA paratransit operations, handling hundreds of daily requests and reducing empty-seat miles by 18%. The real measure of success was adoption: dispatchers chose to use the system because it made their decisions better.
In Nashville, the Vectura dashboard has been piloted with WeGo dispatchers, delivering the 7% ridership improvement and 42% deadhead reduction documented in our evaluations.
PATH-TN — an $8.6M USDOT-funded consortium — brings our individual agency projects together into a unified vision for multimodal transit across Tennessee. Nashville, Chattanooga, Memphis, and Knoxville each face the same problem: how to integrate fixed-route buses, on-demand microtransit, paratransit, and park-and-ride into a seamless network. PATH-TN brings together four transit agencies (WeGo, CARTA, MATA, KAT) and academic partners (Penn State, UTK, University of Memphis) to build the AI planning tools, data infrastructure, and operational workflows to make this real — moving from single-agency optimization to multi-modal, multi-agency coordination across an entire state.
The REACH project extends this to entirely new contexts. Blue Oval City — Ford’s new manufacturing hub in West Tennessee — will employ approximately 10,000 workers across a large rural area with strict shift schedules. We are developing multi-modal transit systems coordinating shuttles, rideshare, and microtransit to connect workers to employment in a community that has historically had no transit at all.
The technologies developed across a decade of research are now being carried forward by Mobius AI, a company spun out of SCOPE Lab research, commercializing our transit optimization and decision support systems for agencies nationwide.
This work reflects sustained collaboration across institutions. Our team includes members from Vanderbilt University, Cornell University, Penn State University, University of Washington, University of Tennessee at Chattanooga, and University of South Carolina. Our transit agency partners — WeGo Public Transit, CARTA, MATA, and KAT — are not just data providers but active collaborators who shape our research questions and validate our solutions in the field.
Funding has been provided by the National Science Foundation, Department of Energy, Federal Transit Administration, USDOT, and the Tennessee Department of Economic and Community Development.
Selected Publications:
@article{talusanTCPS2025,
title = {An End-to-End Solution for Public Transit Stationing and Dispatch Problem},
author = {Talusan, Jose Paolo and Han, Chaeeun and Rogers, David and Mukhopadhyay, Ayan and Laszka, Aron and Freudberg, Dan and Dubey, Abhishek},
year = {2025},
month = jul,
journal = {ACM Trans. Cyber-Phys. Syst.},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
doi = {10.1145/3754454},
issn = {2378-962X},
url = {https://doi.org/10.1145/3754454},
note = {Just Accepted},
keywords = {public transit, dispatch optimization, stationing problems, Monte Carlo tree search, disruption management, resource allocation},
what = {This work presents an end-to-end solution for public transit stationing and dispatch problems, formulating the dynamic scheduling and dispatch challenge for fixed-line transit under disruptions. The research develops a semi-Markov decision process framework that solves for optimal routing and dispatch policies using Monte Carlo Tree Search. The platform integrates a simulator for evaluating both synthetic benchmarks and real-world transit data, enabling principled evaluation of stationing decisions and dispatch policies under realistic operational constraints.},
why = {Public transit systems face significant challenges from operational disruptions and the need to maintain service quality under uncertainty, yet most research assumes fixed infrastructure and deterministic conditions. This work is innovative because it provides a unified, scalable framework that simultaneously optimizes both strategic stationing decisions and dynamic dispatch policies, using simulation-based validation on real agency data. The approach bridges planning-time and operation-time decisions in transit systems.},
results = {Evaluation on real WeGo Public Transit data from Nashville demonstrates that the proposed approach increases passenger service by 7% while reducing deadhead miles by 42% compared to greedy baselines. The Monte Carlo Tree Search-based planning provides significantly better performance than myopic policies, validating the effectiveness of principled decision-making under operational uncertainty.},
project_tags = {transit, planning, CPS}
}
Public bus transit systems provide critical transportation services for large sections of modern communities. On-time performance and maintaining the reliable quality of service is therefore very important. Unfortunately, disruptions caused by overcrowding, vehicular failures, and road accidents often lead to service performance degradation. Though transit agencies keep a limited number of vehicles in reserve and dispatch them to relieve the affected routes during disruptions, the procedure is often ad-hoc and has to rely on human experience and intuition to allocate resources (vehicles) to affected trips under uncertainty. In this paper, we describe a principled approach using non-myopic sequential decision procedures to solve the problem and decide (a) if it is advantageous to anticipate problems and proactively station transit buses near areas with high-likelihood of disruptions and (b) decide if and which vehicle to dispatch to a particular problem. Our approach was developed in partnership WeGo Public Transit, a public transportation agency based in Nashville, Tennessee and models the system as a semi-Markov decision problem (solved as a Monte-Carlo tree search procedure) and shows that it is possible to obtain an answer to these two coupled decision problems in a way that maximizes the overall reward (number of people served). We sample many possible futures from generative models, each is assigned to a tree and processed using root parallelization. We validate our approach with both real-world and scaled-up data from two agencies in Tennessee. Our experiments show that the proposed framework serves 2% more passengers while reducing deadhead miles by 40%. Finally, we introduce Vectura, a dashboard providing transit dispatchers a complete view of the transit system at a glance along with access to our developed tools.
@inproceedings{talusan2024ICCPS,
author = {Talusan, Jose Paolo and Han, Chaeeun and Mukhopadhyay, Ayan and Laszka, Aron and Freudberg, Dan and Dubey, Abhishek},
booktitle = {Proceedings of the ACM/IEEE 15th International Conference on Cyber-Physical Systems (ICCPS)},
title = {An Online Approach to Solving Public Transit Stationing and Dispatch Problem},
year = {2024},
address = {New York, NY, USA},
publisher = {Association for Computing Machinery},
series = {ICCPS '24},
contribution = {lead},
note = {Best paper award},
acceptance = {28.2},
location = {Hong Kong, China},
numpages = {10},
what = {This work develops a software framework for public transit stoning and dispatch that solves the problem of optimally assigning substitute buses when the fixed-line fleet experiences disruptions. The system models the problem as a semi-Markov decision process and uses Monte Carlo tree search to find good dispatching decisions. The approach includes both offline optimization for planned scheduling and online components for responding to real-time disruptions, with integration into a complete transit management system.},
why = {When transit buses break down or experience incidents, agencies must quickly decide which substitute vehicles to dispatch to cover affected trips. This decision-making problem combines aspects of scheduling, resource allocation, and real-time optimization. The work is important because it addresses the practical challenge of making good decisions under uncertainty with limited time and information, using both planning and learning techniques to balance the need for speed with solution quality.},
results = {The MCTS-based approach successfully solves the stoning and dispatch problem for real transit instances, outperforming greedy baseline approaches. The system demonstrates the ability to handle both pre-planned scheduling for known trip patterns and dynamic reallocation when disruptions occur. Results show how tree search methods can effectively explore the space of alternative dispatching strategies to find solutions that minimize passenger impact.},
keywords = {transit dispatch, vehicle routing, disruption response, online optimization, Monte Carlo tree search, resource allocation, real-time decision-making},
project_tags = {transit, emergency, POMDP, middleware}
}
Public bus transit systems provide critical transportation services for large sections of modern communities. On-time performance and maintaining the reliable quality of service is therefore very important. Unfortunately, disruptions caused by overcrowding, vehicular failures, and road accidents often lead to service performance degradation. Though transit agencies keep a limited number of vehicles in reserve and dispatch them to relieve the affected routes during disruptions, the procedure is often ad-hoc and has to rely on human experience and intuition to allocate resources (vehicles) to affected trips under uncertainty. In this paper, we describe a principled approach using non-myopic sequential decision procedures to solve the problem and decide (a) if it is advantageous to anticipate problems and proactively station transit buses near areas with high-likelihood of disruptions and (b) decide if and which vehicle to dispatch to a particular problem. Our approach was developed in partnership with the Metropolitan Transportation Authority for a mid-sized city in the USA and models the system as a semi-Markov decision problem (solved as a Monte-Carlo tree search procedure) and shows that it is possible to obtain an answer to these two coupled decision problems in a way that maximizes the overall reward (number of people served). We sample many possible futures from generative models, each is assigned to a tree and processed using root parallelization. We validate our approach using 3 years of data from our partner agency. Our experiments show that the proposed framework serves 2% more passengers while reducing deadhead miles by 40%.
@inproceedings{iccps2026_prompt_confirmation,
author = {Sivagnanam, Amutheezan and Mukhopadhyay, Ayan and Samaranayake, Samitha and Dubey, Abhishek and Laszka, Aron},
title = {Dynamic Vehicle Routing with Prompt Confirmation and Continual Optimization},
year = {2026},
booktitle = {Proceedings of the HSCC/ICCPS 2026: 29th ACM International Conference on Hybrid Systems: Computation and Control and 17th ACM/IEEE International Conference on Cyber-Physical Systems},
location = {Saint Malo, France},
keywords = {dynamic vehicle routing, on-demand transportation, prompt confirmation, optimization, stochastic requests, anytime algorithms, reinforcement learning},
note = {Acceptance rate: 28\%; Regular Paper; Track: Systems and Applications},
series = {HSCC/ICCPS '26},
what = {This paper introduces a novel computational approach for dynamic vehicle routing with prompt confirmation of advance requests. The work addresses the problem of on-demand transportation services that must make real-time decisions about accepting or rejecting trip requests while continuously optimizing vehicle manifests and routes. The research formulates this as a two-stage optimization problem: first deciding whether to accept or reject incoming requests with immediate response requirements, then continuously improving route plans to accommodate future requests between arrival of consecutive requests.},
why = {Real-world on-demand transit services face a fundamental challenge: agencies must provide prompt confirmation of whether requests can be accepted, yet future requests are unknown and will influence optimal route plans. Most prior work either provides immediate confirmation without optimizing or continuously optimizes without addressing the confirmation timing problem. This work is innovative because it bridges this gap by combining quick insertion search for rapid decision-making with continuous optimization, enabling both high service rates and operational efficiency while managing computational constraints.},
results = {The proposed computational approach demonstrates significantly better trade-offs between confirmation timeliness and service rate compared to existing methods on real-world and synthetic problem instances from a public transit agency. The anytime algorithm with continuous optimization provides prompt confirmation while also improving subsequent route plans, achieving higher service rates than approaches that simply optimize without considering confirmation requirements.},
project_tags = {transit, planning}
}
Transit agencies that operate on-demand transportation services have to respond to trip requests from passengers in real time, which involves solving dynamic vehicle routing problems with pick-up and drop-off constraints. Based on discussions with public transit agencies, we observe a real-world problem that is not addressed by prior work: when trips are booked in advance (e.g., trip requests arrive a few hours in advance of their requested pick-up times), the agency needs to promptly confirm whether a request can be accepted or not, and ensure that accepted requests are served as promised. State-of-the-art computational approaches either provide prompt confirmation but lack the ability to continually optimize and improve routes for accepted requests, or they provide continual optimization but cannot guarantee serving all accepted requests. To address this gap, we introduce a novel problem formulation of dynamic vehicle routing with prompt confirmation and continual optimization. We propose a novel computational approach for this vehicle routing problem, which integrates a quick insertion search for prompt confirmation with an anytime algorithm for continual optimization. To maximize the number requests served, we train a non-myopic objective function using reinforcement learning, which guides both the insertion and the anytime algorithms towards optimal, non-myopic solutions. We evaluate our computational approach on a real-world microtransit dataset from a public transit agency in the U.S., demonstrating that our proposed approach provides prompt confirmation while significantly increasing the number of requests served compared to existing approaches.
@inproceedings{iccps2026_schoolride,
author = {Nath, Vakul and Liu, Fangqi and He, Guocheng and Rogers, David and Chhokra, Ajay and Talusan, Jose Paolo and Ma, Meiyi and Mukhopadhyay, Ayan and Dubey, Abhishek},
title = {SchoolRide: A Data-Driven Platform for School Bus Disruption Management},
year = {2026},
booktitle = {Proceedings of the HSCC/ICCPS 2026: 29th ACM International Conference on Hybrid Systems: Computation and Control and 17th ACM/IEEE International Conference on Cyber-Physical Systems},
location = {Saint Malo, France},
keywords = {school transportation, disruption management, vehicle routing, optimization, cyber-physical systems, transit operations, real-time decision-making},
note = {Acceptance rate: 28\%; Short Paper; Track: Systems and Applications},
series = {HSCC/ICCPS '26},
what = {SchoolRide is a comprehensive cyber-physical system platform designed for school bus disruption management and operational resilience. The system integrates live telemetry, real-time status collection, and dynamic bus status monitoring to detect and respond to disruptions such as vehicle breakdowns, traffic congestion, and driver absences. Using an integrated pipeline that combines baseline routing with travel-time prediction and constrained optimization, SchoolRide automatically recomputes routing plans when disruptions occur. The platform serves as a testbed for evaluating data-driven optimization strategies for real-world school transportation systems with practical constraints.},
why = {School transportation is a societal-scale transportation cyber-physical system serving 26 million students daily, yet it remains vulnerable to operational disruptions despite strict schedules and regulations. Most existing disruption management relies on manual coordination, while SchoolRide advances the state-of-the-art by demonstrating that systematic, data-driven optimization can enhance operational resilience at realistic scale. This work is innovative because it balances competing objectives—student service quality (waiting time, delays, schedule adherence) with operational efficiency—while respecting institutional constraints and preserving privacy through synthetic data generation.},
results = {Experiments on synthetic benchmarks and real district data demonstrate strong performance and scalability of the SchoolRide optimization approach. The AdVIns insertion heuristic consistently outperforms baseline human-intuition policies on student-centered metrics, achieving substantially lower average stop and school delays. Across large-scale synthetic instances and real district scenarios, the system effectively handles realistic disruption patterns while generating high-quality rerouting solutions that balance feasibility with optimality.},
project_tags = {transit, CPS, planning}
}
As a societal-scale transportation Cyber-Physical System (CPS), school transportation integrates large-scale physical operations with cyber components for planning and control under uncertainty. Despite its scale and societal importance, the system remains vulnerable to operational disruptions such as vehicle breakdowns, road closures, traffic congestion, and driver absences. This work demonstrates how data-driven optimization can enhance operational resilience in a real-world school transit context. To advance research in this domain, we introduce SchoolRide, a platform developed in close collaboration with a school district in the southern United States. SchoolRide serves as a comprehensive testbed for studying and evaluating robust operational policies for disruption management, enabling systematic investigation of strategies under realistic data and operational constraints. We design an integrated pipeline for dynamic bus status collection and formulate the School Bus Disruption Management (SBDM) problem as a combinatorial optimization task that replans routes based on predefined schedules, real-time status, and disruption events. The framework balances student service quality (e.g., waiting time and school delays) with operational efficiency (e.g., route adjustments and driver workload). We explore heuristic and optimization-based approaches that leverage historical disruption logs from the partner district to proactively replan routes and evaluate their performance using synthetic data generated from real-world operational records to protect privacy. The generated synthetic datasets will be released to facilitate future research in this domain. Our approach outperforms current operational policies, effectively preserving service quality while reducing disruptions and workload.
@inproceedings{paviaIJCAI24AISG,
author = {Pavia, Sophie and Rogers, David and Sivagnanam, Amutheezan and Wilbur, Michael and Edirimanna, Danushka and Kim, Youngseok and Pugliese, Philip and Samaranayake, Samitha and Laszka, Aron and Mukhopadhyay, Ayano and Dubey, Abhishek},
booktitle = {Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence},
title = {Deploying mobility-on-demand for all by optimizing paratransit services},
year = {2024},
series = {IJCAI '24},
articleno = {822},
contribution = {lead},
acceptance = {15},
doi = {10.24963/ijcai.2024/822},
isbn = {978-1-956792-04-1},
location = {Jeju, Korea},
numpages = {8},
url = {https://doi.org/10.24963/ijcai.2024/822},
what = {This work develops a software framework and routing application for paratransit and microtransit services operating in urban environments. The SmartTransit.AI system integrates multiple ridesharing algorithms including both day-ahead optimization for planned trips and real-time dynamic vehicle routing problem solvers. The framework provides operational interfaces for dispatchers, a vehicle operator mobile application, and a user-facing booking interface. The system incorporates state-of-the-art algorithms while addressing practical constraints like time windows, vehicle capacity limitations, and service accessibility requirements.},
why = {Paratransit services are critical for accessibility but face significant operational challenges due to complex constraints not present in traditional ridesharing. Existing commercial systems are often inflexible and fail to adapt to real-world conditions, while research algorithms are difficult to deploy in practice. This work is innovative because it bridges this gap by creating a modular software system that can accommodate different algorithmic approaches and constraints specific to transit agencies, while also providing human operators the ability to override and validate system recommendations.},
results = {The deployed system demonstrates substantial operational improvements when tested with real paratransit data, showing significantly higher shared ride rates and reduced vehicle miles compared to baseline approaches. Pilot testing in Chattanooga, Tennessee with the Chattanooga Area Regional Transportation Authority validates the system's ability to improve both efficiency and service quality in a real operational environment. The results show clear benefits in reducing operational costs while maintaining service accessibility.},
keywords = {paratransit optimization, microtransit, vehicle routing, shared mobility, transportation dispatch, accessibility, real-time optimization, mobility-on-demand},
project_tags = {transit, energy, planning, scalable AI, middleware}
}
While on-demand ride-sharing services have become popular in recent years, traditional on-demand transit services cannot be used by everyone, e.g., people who use wheelchairs. Paratransit services, operated by public transit agencies, are a critical infrastructure that offers door-to-door transportation assistance for individuals who face challenges in using standard transit routes. However, with declining ridership and mounting financial pressure, public transit agencies in the USA struggle to operate existing services. We collaborate with a public transit agency from the southern USA, highlight the specific nuances of paratransit optimization, and present a vehicle routing problem formulation for optimizing paratransit. We validate our approach using real-world data from the transit agency, present results from an actual pilot deployment of the proposed approach in the city, and show how the proposed approach comprehensively outperforms existing approaches used by the transit agency. To the best of our knowledge, this work presents one of the first examples of using open-source algorithmic approaches for paratransit optimization.
@inproceedings{wilbur2023mobility,
author = {Wilbur, Michael and Coursey, Maxime and Koirala, Pravesh and Al-Quran, Zakariyya and Pugliese, Philip and Dubey, Abhishek},
booktitle = {Proceedings of the ACM/IEEE 14th International Conference on Cyber-Physical Systems (with CPS-IoT Week 2023)},
title = {Mobility-On-Demand Transportation: A System for Microtransit and Paratransit Operations},
year = {2023},
address = {New York, NY, USA},
note = {demonstration},
pages = {260--261},
publisher = {Association for Computing Machinery},
series = {ICCPS '23},
contribution = {lead},
doi = {10.1145/3576841.3589625},
isbn = {9798400700361},
keywords = {mobility-on-demand, software systems, microtransit, paratransit, operational software, deployment, system integration, transportation technology},
location = {San Antonio, TX, USA},
numpages = {2},
url = {https://doi.org/10.1145/3576841.3589625},
what = {This paper presents a comprehensive software system for managing mobility-on-demand services including microtransit and paratransit operations. The SmartTransit.AI system provides web-based interfaces for operational management, mobile applications for drivers and users, and modular optimization components that can accommodate different algorithms and constraints. The paper describes the architecture, implementation challenges, and deployment experiences from real-world testing with transit agencies.},
why = {Despite advances in optimization algorithms, deploying ridesharing systems in practice requires solving numerous challenges beyond pure algorithmic optimization including user interfaces, real-time data integration, and operational constraints. This work is valuable because it demonstrates how research algorithms can be integrated into functional systems that transit agencies can actually deploy. The modular architecture enables different agencies to adopt the system while customizing it to their specific operational needs.},
results = {The SmartTransit.AI system successfully demonstrates the feasibility of deploying advanced optimization algorithms in real transit operations. The integrated software system handles both offline planning and real-time optimization for shared mobility services. Real-world deployment results show the system's ability to improve operational efficiency while maintaining usability for operators and accessibility for passengers.},
project_tags = {transit, middleware}
}
New rideshare and shared-mobility services have transformed urban mobility in recent years. Therefore, transit agencies are looking for ways to adapt to this rapidly changing environment. In this space, ridepooling has the potential to improve efficiency and reduce costs by allowing users to share rides in high-capacity vehicles and vans. Most transit agencies already operate various ridepooling services including microtransit and paratransit. However, the objectives and constraints for implementing these services vary greatly between agencies. This brings multiple challenges. First, off-the-shelf ridepooling formulations must be adapted for real-world conditions and constraints. Second, the lack of modular and reusable software makes it hard to implement and evaluate new ridepooling algorithms and approaches in real-world settings. Therefore, we propose an on-demand transportation scheduling software for microtransit and paratransit services. This software is aimed at transit agencies looking to incorporate state-of-the-art rideshare and ridepooling algorithms in their everyday operations. We provide management software for dispatchers and mobile applications for drivers and users. Lastly, we discuss the challenges in adapting state-of-the-art methods to real-world operations.
@inproceedings{sivagnanam2022offline,
author = {Sivagnanam, Amutheezan and Kadir, Salah Uddin and Mukhopadhyay, Ayan and Pugliese, Philip and Dubey, Abhishek and Samaranayake, Samitha and Laszka, Aron},
booktitle = {31st International Joint Conference on Artificial Intelligence (IJCAI)},
title = {Offline Vehicle Routing Problem with Online Bookings: A Novel Problem Formulation with Applications to Paratransit},
year = {2022},
acceptance = {15},
month = jul,
contribution = {colab},
preprint = {https://arxiv.org/abs/2204.11992},
what = {This work addresses the offline vehicle routing problem with online bookings for paratransit services, where pickup windows are selected at the time of booking rather than predetermined. The authors propose a formulation combining an offline vehicle routing model with an online bookings model, and present computational approaches including an anytime algorithm with reinforcement learning and a Markov decision process formulation.},
why = {Paratransit services for elderly and disabled passengers require high flexibility in response to real-time requests while maintaining operational efficiency. This work is novel because it bridges the gap between offline and online routing problems with practical constraints on pickup windows. The combination of optimization and learning approaches enables the system to adapt to dynamic demand while respecting the transportation agency's operational requirements.},
results = {The proposed methods were evaluated using real-world paratransit data from Chattanooga, showing that the anytime algorithm with learning outperforms baseline approaches. The reinforcement learning approach effectively learns policies that balance responsiveness to immediate requests with long-term efficiency considerations. The experimental results demonstrate significant improvements in cost reduction and robustness when environmental conditions change dynamically.},
keywords = {vehicle routing, online optimization, paratransit services, reinforcement learning, demand-responsive transport},
project_tags = {transit, planning, scalable AI, POMDP}
}
Vehicle routing problems (VRPs) can be divided into two major categories: offline VRPs, which consider a given set of trip requests to be served, and online VRPs, which consider requests as they arrive in real-time. Based on discussions with public transit agencies, we identify a real-world problem that is not addressed by existing formulations: booking trips with flexible pickup windows (e.g., 3 hours) in advance (e.g., the day before) and confirming tight pickup windows (e.g., 30 minutes) at the time of booking. Such a service model is often required in paratransit service settings, where passengers typically book trips for the next day over the phone. To address this gap between offline and online problems, we introduce a novel formulation, the offline vehicle routing problem with online bookings. This problem is very challenging computationally since it faces the complexity of considering large sets of requests—similar to offline VRPs—but must abide by strict constraints on running time—similar to online VRPs. To solve this problem, we propose a novel computational approach, which combines an anytime algorithm with a learning-based policy for real-time decisions. Based on a paratransit dataset obtained from our partner transit agency, we demonstrate that our novel formulation and computational approach lead to significantly better outcomes in this service setting than existing algorithms.
@article{wilbur2022_trr,
author = {Wilbur, Michael and Ayman, Afiya and Sivagnanam, Amutheezan and Ouyang, Anna and Poon, Vincent and Kabir, Riyan and Vadali, Abhiram and Pugliese, Philip and Freudberg, Daniel and Laszka, Aron and Dubey, Abhishek},
journal = {Transportation Research Record},
title = {Impact of COVID-19 on Public Transit Accessibility and Ridership},
year = {2023},
number = {4},
pages = {531--546},
volume = {2677},
contribution = {minor},
doi = {10.1177/03611981231160531},
eprint = {https://doi.org/10.1177/03611981231160531},
url = {https://doi.org/10.1177/03611981231160531},
what = {This paper investigates how COVID-19 changed public transit ridership patterns across different socioeconomic groups in Nashville and Chattanooga, Tennessee. The work analyzes boarding data, paratransit demand, and cellular mobility data to understand changes in transit usage before, during, and after pandemic restrictions. The study examines whether ridership changes were distributed equitably across populations or if COVID-19 disproportionately affected certain groups including lower-income residents and mobility-impaired transit users.},
why = {COVID-19 caused dramatic shifts in travel patterns and transit demand, but with potentially inequitable impacts across populations. This work is important because it documents how the pandemic affected different demographic groups differently, providing evidence about the resilience and vulnerability of different user populations. The analysis contributes to understanding how external shocks affect transit systems and which populations are most vulnerable to disruptions.},
results = {The analysis shows that COVID-19 caused significant initial declines in ridership that gradually recovered, with ridership declining more in higher-income areas initially but recovery being faster there. The distribution of changes across socioeconomic groups and mobility-impaired users varied, suggesting that the pandemic's impact on transit access was not uniform. The findings highlight the importance of understanding how external disruptions affect different populations and the need to protect transit access for vulnerable groups.},
keywords = {COVID-19 pandemic, transit ridership, equity, socioeconomic disparities, mobility-impaired users, pandemic impacts, transportation resilience, urban mobility},
project_tags = {transit, emergency}
}
COVID-19 has radically transformed urban travel behavior throughout the world. Agencies have had to provide adequate service while navigating a rapidly changing environment with reduced revenue. As COVID-19-related restrictions are lifted, transit agencies are concerned about their ability to adapt to changes in ridership behavior and public transit usage. To aid their becoming more adaptive to sudden or persistent shifts in ridership, we addressed three questions: To what degree has COVID-19 affected fixed-line public transit ridership and what is the relationship between reduced demand and -vehicle trips? How has COVID-19 changed ridership patterns and are they expected to persist after restrictions are lifted? Are there disparities in ridership changes across socioeconomic groups and mobility-impaired riders? Focusing on Nashville and Chattanooga, TN, ridership demand and vehicle trips were compared with anonymized mobile location data to study the relationship between mobility patterns and transit usage. Correlation analysis and multiple linear regression were used to investigate the relationship between socioeconomic indicators and changes in transit ridership, and an analysis of changes in paratransit demand before and during COVID-19. Ridership initially dropped by 66% and 65% over the first month of the pandemic for Nashville and Chattanooga, respectively. Cellular mobility patterns in Chattanooga indicated that foot traffic recovered to a greater degree than transit ridership between mid-April and the last week in June, 2020. Education-level had a statistically significant impact on changes in fixed-line bus transit, and the distribution of changes in demand for paratransit services were similar to those of fixed-line bus transit.
@article{aymantoit2020,
author = {Ayman, Afiya and Sivagnanam, Amutheezan and Wilbur, Michael and Pugliese, Philip and Dubey, Abhishek and Laszka, Aron},
journal = {ACM Transations of Internet Technology},
title = {Data-Driven Prediction and Optimization of Energy Use for Transit Fleets of Electric and ICE Vehicles},
year = {2020},
contribution = {colab},
tag = {ai4cps,transit},
what = {This paper presents a comprehensive framework for data-driven prediction and optimization of energy consumption in transit fleets. The work integrates vehicle telemetry data, elevation information, weather conditions, and traffic data to build machine learning models for predicting energy consumption at the trip level. The framework includes algorithms for data cleaning, feature engineering, and optimization of vehicle-to-route assignments to minimize energy costs while meeting service constraints.},
why = {Public transit agencies operating mixed fleets of electric and internal combustion vehicles face significant challenges in optimizing operations while reducing environmental impact and operating costs. Accurate prediction of energy consumption is essential for effective vehicle scheduling and fleet management. This work is innovative because it provides an end-to-end framework that integrates real-world data collection, processing, and optimization to support practical decisions about vehicle assignments and fleet operation.},
results = {The framework successfully demonstrates energy consumption prediction for mixed transit fleets using real data from the Chattanooga Area Regional Transportation Authority. Machine learning models including neural networks and decision trees achieve accurate energy predictions that outperform simpler baselines. Results show that the approach can support optimization of vehicle assignments to minimize energy costs while maintaining service levels, providing practical benefits for transit agencies operating electric vehicles.},
keywords = {electric vehicles, energy consumption prediction, transit optimization, machine learning, vehicle scheduling, data-driven optimization},
project_tags = {energy, transit, ML for CPS}
}
Due to the high upfront cost of electric vehicles, many public transit agencies can afford only mixed fleets of internal-combustion and electric vehicles. Optimizing the operation of such mixed fleets is challenging because it requires accurate trip-level predictions of electricity and fuel use as well as efficient algorithms for assigning vehicles to transit routes. We present a novel framework for the data-driven prediction of trip-level energy use for mixed-vehicle transit fleets and for the optimization of vehicle assignments, which we evaluate using data collected from the bus fleet of CARTA, the public transit agency of Chattanooga, TN. We first introduce a data collection, storage, and processing framework for system-level and high-frequency vehicle-level transit data, including domain-specific data cleansing methods. We train and evaluate machine learning models for energy prediction, demonstrating that deep neural networks attain the highest accuracy. Based on these predictions, we formulate the problem of minimizing energy use through assigning vehicles to fixed-route transit trips. We propose an optimal integer program as well as efficient heuristic and meta-heuristic algorithms, demonstrating the scalability and performance of these algorithms numerically using the transit network of CARTA.
@article{Sun2019,
author = {Sun, Fangzhou and Dubey, Abhishek and White, Jules and Gokhale, Aniruddha},
journal = {Cluster Computing},
title = {Transit-hub: a smart public transportation decision support system with multi-timescale analytical services},
year = {2019},
month = jan,
number = {Suppl 1},
pages = {2239--2254},
volume = {22},
bibsource = {dblp computer science bibliography, https://dblp.org},
biburl = {https://dblp.org/rec/bib/journals/cluster/SunDWG19},
contribution = {lead},
doi = {10.1007/s10586-018-1708-z},
file = {:Sun2019-Transit-hub_a_smart_public_transportation_decision_support_system_with_multi-timescale_analytical_services.pdf:PDF},
keywords = {public transit, decision support, real-time prediction, schedule optimization, data integration, Kalman filtering},
project = {smart-cities,smart-transit},
tag = {transit},
timestamp = {Wed, 21 Aug 2019 01:00:00 +0200},
url = {https://doi.org/10.1007/s10586-018-1708-z},
what = {This paper presents Transit-Hub, a decision support system for public transportation that integrates multi-timescale analytical services for real-time bus arrival prediction and schedule optimization. The system combines historical and real-time transit data from multiple sources including GTFS data feeds and live vehicle location tracking. Advanced analytics including SVM-Kalman models enable both short-term and long-term delay prediction.},
why = {Public transit agencies struggle with providing accurate real-time information and optimizing schedules based on actual demand patterns, particularly with heterogeneous data quality. Transit-Hub is innovative because it integrates data cleaning and management with advanced analytical models that address data quality issues while providing decision support to transit authorities. The multi-timescale approach enables both immediate customer-facing predictions and longer-term operational optimization.},
results = {The system reduced root-mean-square deviation in travel time prediction by 30-65% when predicting 15 minutes ahead using the integrated short-term prediction model. The approach demonstrated ability to integrate and manage heterogeneous data sources from multiple transit systems while providing reliable transit information. The system achieved significant improvements over basic average models in delay prediction accuracy.},
project_tags = {transit, planning}
}
Public transit is a critical component of a smart and connected community. As such, citizens expect and require accurate information about real-time arrival/departures of transportation assets. As transit agencies enable large-scale integration of real-time sensors and support back-end data-driven decision support systems, the dynamic data-driven applications systems (DDDAS) paradigm becomes a promising approach to make the system smarter by providing online model learning and multi-time scale analytics as part of the decision support system that is used in the DDDAS feedback loop. In this paper, we describe a system in use in Nashville and illustrate the analytic methods developed by our team. These methods use both historical as well as real-time streaming data for online bus arrival prediction. The historical data is used to build classifiers that enable us to create expected performance models as well as identify anomalies. These classifiers can be used to provide schedule adjustment feedback to the metro transit authority. We also show how these analytics services can be packaged into modular, distributed and resilient micro-services that can be deployed on both cloud back ends as well as edge computing resources.
@inproceedings{samir2024smartcomp,
author = {Gupta, Samir and Khanna, Agrima and Talusan, Jose Paolo and Said, Anwar and Freudberg, Dan and Mukhopadhyay, Ayan and Dubey, Abhishek},
booktitle = {2024 IEEE International Conference on Smart Computing (SMARTCOMP)},
title = {A Graph Neural Network Framework for Imbalanced Bus Ridership Forecasting},
year = {2024},
acceptance = {32.9},
month = jun,
contribution = {lead},
what = {This paper proposes a Graph Convolutional Network framework for bus ridership forecasting that addresses data sparsity and imbalance issues in public transit occupancy prediction. The approach combines graph neural networks to capture spatial-temporal dependencies with data augmentation and focal loss to handle the heavy-tail occupancy distribution. GCNs model bus networks as graphs where stops and routes capture the transit network structure, enabling the model to learn patterns specific to route dynamics.},
why = {Public transit systems require accurate occupancy forecasting for operational planning, but many routes exhibit sparse data with imbalanced occupancy distributions (most trips have low occupancy, few have high occupancy). GCN-based methods are innovative because they leverage the underlying graph structure of transit networks to learn more expressive representations while handling data sparsity through inductive learning across stops and routes, improving generalization.},
results = {Evaluation on real WEGo Public Transit data from Nashville demonstrates that the GCN approach significantly outperforms traditional baselines including random forest and XGBoost methods, with particular improvements in predicting high-occupancy events that are critical for preventing overcrowding and ensuring service quality.},
keywords = {ridership forecasting, graph neural networks, public transit, occupancy prediction, data imbalance, spatio-temporal modeling},
project_tags = {transit, ML for CPS}
}
Public transit systems are paramount in lowering carbon emissions and reducing urban congestion for environmental sustainability. However, overcrowding has adverse effects on the quality of service, passenger experience, and overall efficiency of public transit causing a decline in the usage of public transit systems. Therefore, it is crucial to identify and forecast potential windows of overcrowding to improve passenger experience and encourage higher ridership. Predicting ridership is a complex task, due to the inherent noise of collected data and the sparsity of overcrowding events. Existing studies in predicting public transit ridership consider only a static depiction of bus networks. We address these issues by first applying a data processing pipeline that cleans noisy data and engineers several features for training. Then, we address sparsity by converting the network to a dynamic graph and using a graph convolutional network, incorporating temporal, spatial, and auto-regressive features, to learn generalizable patterns for each route. Finally, since conventional loss functions like categorical cross-entropy have limitations in addressing class imbalance inherent in ridership data, our proposed approach uses focal loss to refine the prediction focus on less frequent yet task-critical overcrowding instances. Our experiments, using real-world data from our partner agency, show that the proposed approach outperforms existing state-of-the-art baselines in terms of accuracy and robustness.
@inproceedings{talusan2022apc,
author = {Talusan, Jose Paolo and Mukhopadhyay, Ayan and Freudberg, Dan and Dubey, Abhishek},
booktitle = {2022 IEEE International Conference on Big Data (Big Data)},
title = {On Designing Day Ahead and Same Day Ridership Level Prediction Models for City-Scale Transit Networks Using Noisy APC Data},
year = {2022},
address = {Los Alamitos, CA, USA},
month = dec,
pages = {5598-5606},
publisher = {IEEE Computer Society},
contribution = {lead},
doi = {10.1109/BigData55660.2022.10020390},
keywords = {transit prediction, occupancy forecasting, delay prediction, automated passenger counting, machine learning, operational planning, transit optimization, real-time information},
url = {https://doi.ieeecomputersociety.org/10.1109/BigData55660.2022.10020390},
what = {This paper presents methods for predicting transit occupancy and delay at both trip and stop levels despite sparse automated passenger counter data. The approach combines data from multiple sources including GTFS schedules, weather data, and historical patterns to develop separate prediction models for different problem formulations. The work demonstrates how to handle data sparsity and noise through careful feature engineering and data aggregation strategies.},
why = {Accurate occupancy and delay predictions are essential for transit agencies to optimize operations and improve passenger information, but predictions are challenging due to sparse sensor data and the complexity of transit dynamics. This work addresses the practical challenge of developing predictive models despite data quality issues that plague real-world transit systems. The end-to-end framework demonstrates how to process raw sensor data into actionable predictions.},
results = {The prediction models achieve reasonable accuracy for occupancy and delay forecasting on real transit data from Nashville. The approach demonstrates how different aggregation strategies and feature engineering choices affect prediction performance. Results show that treating occupancy and delay as related prediction problems improves accuracy compared to separate approaches, providing transit agencies with tools for operational planning.},
project_tags = {transit, ML for CPS}
}
The ability to accurately predict public transit ridership demand benefits passengers and transit agencies. Agencies will be able to reallocate buses to handle under or over-utilized bus routes, improving resource utilization, and passengers will be able to adjust and plan their schedules to avoid overcrowded buses and maintain a certain level of comfort. However, accurately predicting occupancy is a non-trivial task. Various reasons such as heterogeneity, evolving ridership patterns, exogenous events like weather, and other stochastic variables, make the task much more challenging. With the progress of big data, transit authorities now have access to real-time passenger occupancy information for their vehicles. The amount of data generated is staggering. While there is no shortage in data, it must still be cleaned, processed, augmented, and merged before any useful information can be generated. In this paper, we propose the use and fusion of data from multiple sources, cleaned, processed, and merged together, for use in training machine learning models to predict transit ridership. We use data that spans a 2-year period (2020-2022) incorporating transit, weather, traffic, and calendar data. The resulting data, which equates to 17 million observations, is used to train separate models for the trip and stop level prediction. We evaluate our approach on real-world transit data provided by the public transit agency of Nashville, TN. We demonstrate that the trip level model based on Xgboost and the stop level model based on LSTM outperform the baseline statistical model across the entire transit service day.
@inproceedings{sun2021transitgym,
author = {Sun, Ruixiao and Gui, Rongze and Neema, Himanshu and Chen, Yuche and Ugirumurera, Juliette and Severino, Joseph and Pugliese, Philip and Laszka, Aron and Dubey, Abhishek},
booktitle = {2021 IEEE International Conference on Smart Computing (SMARTCOMP)},
title = {TRANSIT-GYM: A Simulation and Evaluation Engine for Analysis of Bus Transit Systems},
year = {2021},
month = aug,
pages = {69-76},
contribution = {colab},
acceptance = {31.7},
doi = {10.1109/SMARTCOMP52413.2021.00030},
issn = {2693-8340},
keywords = {Training;Analytical models;Uncertainty;Computational modeling;Microscopy;Vehicle routing;Urban areas;Transit simulation;domain-specific modeling language;traffic simulation;micro-simulation;regional transportation system;transportation planning;data-driven optimization},
tag = {transit}
}
Public-transit systems face a number of operational challenges: (a) changing ridership patterns requiring optimization of fixed line services, (b) optimizing vehicle-to-trip assignments to reduce maintenance and operation codes, and (c) ensuring equitable and fair coverage to areas with low ridership. Optimizing these objectives presents a hard computational problem due to the size and complexity of the decision space. State-of-the-art methods formulate these problems as variants of the vehicle routing problem and use data-driven heuristics for optimizing the procedures. However, the evaluation and training of these algorithms require large datasets that provide realistic coverage of various operational uncertainties. This paper presents a dynamic simulation platform, called TRANSIT-GYM, that can bridge this gap by providing the ability to simulate scenarios, focusing on variation of demand models, variations of route networks, and variations of vehicle-to-trip assignments. The central contribution of this work is a domain-specific language and associated experimentation tool-chain and infrastructure to enable subject-matter experts to intuitively specify, simulate, and analyze large-scale transit scenarios and their parametric variations. Of particular significance is an integrated microscopic energy consumption model that also helps to analyze the energy cost of various transit decisions made by the transportation agency of a city.
@inproceedings{sen2022,
author = {Sen, Rishav and Tran, Toan and Khaleghian, Seyedmehdi and Pugliese, Philip and Sartipi, Mina and Neema, Himanshu and Dubey, Abhishek},
booktitle = { 2022 IEEE International Conference on Big Data (Big Data) },
title = {{ BTE-Sim: Fast Simulation Environment For Public Transportation }},
year = {2022},
address = {Los Alamitos, CA, USA},
month = dec,
pages = {2886-2894},
publisher = {IEEE Computer Society},
contribution = {lead},
doi = {10.1109/BigData55660.2022.10020973},
keywords = {transit simulation, traffic modeling, computational efficiency, simulation environment, transit planning tools, operational evaluation, vehicle routing, transportation networks},
url = {https://doi.ieeecomputersociety.org/10.1109/BigData55660.2022.10020973},
what = {This paper presents BTE-Sim, a fast simulation environment for public transit systems that enables rapid evaluation of transit designs and operational strategies. The system combines a background traffic elimination module that speeds simulation by efficiently modeling traffic flow, with detailed transit simulation for buses and passengers. The simulator is built on SUMO and includes capabilities for modeling multiple transit service types including fixed-route and demand-responsive services.},
why = {Traditional transit simulation tools are computationally expensive, limiting their practical use for operational planning and optimization. Transit planners need tools that can rapidly evaluate different service designs and operational strategies without simulating every second of a full day. This work is important because it demonstrates how background traffic can be efficiently modeled to dramatically speed simulation while maintaining accuracy, enabling practical use of simulation tools for transit planning.},
results = {The BTE-Sim simulator achieves approximately 13x speedup compared to conventional simulation approaches while maintaining accuracy comparable to detailed simulations. The system successfully simulates complete transit networks including multiple service types and evaluates operational performance metrics. The simulator's efficiency enables practical use in transit planning for evaluating different route designs and service configurations.},
project_tags = {transit, planning, scalable AI}
}
The public commute is essential to all urban centers and is an efficient and environment-friendly way to travel. Transit systems must become more accessible and user-friendly. Since public transit is majorly designed statically, with very few improvements coming over time, it can get stagnated, unable to update itself with changing population trends. To better understand transportation demands and make them more usable, efficient, and demographic-focused, we propose a fast, multi-layered transit simulation that primarily focuses on public transit simulation (BTE-Sim). BTE-Sim is designed based on the population demand, existing traffic conditions, and the road networks that exist in a region. The system is versatile, with the ability to run different configurations of the existing transit routes, or inculcate any new changes that may seem necessary, or even in extreme cases, new transit network design as well. In all situations, it can compare multiple transit networks and provide evaluation metrics for them. It provides detailed data on each transit vehicle, the trips it performs, its on-time performance and other necessary factors. Its highlighting feature is the considerably low computation time it requires to perform all these tasks and provide consistently reliable results.
@inproceedings{khanna2025driverbreaks,
author = {Khanna, Agrima and Liu, Fangqi and Gupta, Samir and Pavia, Sophie and Mukhopadhyay, Ayan and Dubey, Abhishek},
booktitle = {Proceedings of the 26th International Conference on Distributed Computing and Networking},
title = {PDPTW-DB: MILP-Based Offline Route Planning for PDPTW with Driver Breaks},
year = {2025},
address = {New York, NY, USA},
pages = {73--83},
acceptance = {32},
publisher = {Association for Computing Machinery},
series = {ICDCN '25},
category = {other},
contribution = {lead},
doi = {10.1145/3700838.3700854},
isbn = {9798400710629},
keywords = {vehicle routing, pickup-delivery problems, driver breaks, hours-of-service, mixed-integer programming, logistics optimization, microtransit},
numpages = {11},
url = {https://doi.org/10.1145/3700838.3700854},
what = {PDPTW-DB presents a mixed-integer linear programming formulation for pickup-delivery problems with time windows that integrates periodic driver break requirements. The work addresses the practical challenge of incorporating mandatory driver rest periods into route planning while maintaining service feasibility. The formulation enables optimization of vehicle routing and break scheduling simultaneously, accounting for realistic constraints like hours-of-service regulations, service time windows, and vehicle capacity limitations.},
why = {Existing vehicle routing formulations often overlook mandatory driver break requirements or handle them as post-hoc constraints, leading to infeasible or inefficient solutions in practice. PDPTW-DB is innovative because it integrates break scheduling directly into the optimization formulation, enabling principled trade-offs between vehicle utilization, travel distance, and driver compliance with service regulations. This bridge between operational planning and human factors considerations addresses a critical real-world constraint.},
results = {Implementation and evaluation using real Microtransit delivery data demonstrates the formulation produces cost-effective solutions while ensuring full regulatory compliance. Experiments validate both computational efficiency of the mixed-integer approach and the quality of solutions achievable when driver breaks are explicitly modeled.},
project_tags = {transit, planning}
}
The Pickup and Delivery Problem with Time Windows (PDPTW) involves optimizing routes for vehicles to meet pickup and delivery requests within specific time constraints, a challenge commonly faced in logistics and transportation. Microtransit, a flexible and demand-responsive service using smaller vehicles within defined zones, can be effectively modeled as a PDPTW. Yet, the need for driver breaks—a key human constraint—is frequently overlooked in PDPTW solutions, despite being necessary for regulatory compliance. This study presents a novel mixed-integer linear programming formulation for the Pickup and Delivery Problem with Time Windows and Driver Breaks (PDPTW-DB). To the best of our knowledge this formulation is the first to consider mandatory periodic driver breaks within optimized Microtransit routes. The proposed model incorporates regulatory compliant break scheduling directly within the vehicle routing optimization framework. By considering driver break requirements as an integral component of the optimization process, rather than as a post-processing step, the model enables the generation of routes that respect hours of service regulations while minimizing operational costs. This integrated approach facilitates the generation of schedules that are operationally efficient and prioritize driver welfare through driver breaks. We work with a public transit agency from the southern USA, and highlight the specific nuances of driver break optimization, and present a Pickup and Delivery Problem with Time Windows formulation for optimizing Microtransit operations and scheduling driver breaks. We validate our approach using real-world data from the transit agency. Our results validate our formulation in producing cost-effective, and regulation-compliant solutions.
@inproceedings{iccps2026_moveod,
author = {Sen, Rishav and Talusan, Jose Paolo and Dubey, Abhishek and Mukhopadhyay, Ayan and Samaranayake, Samitha and Laszka, Aron},
title = {MoveOD: Synthesizing Fine-Grained Origin--Destination Data for Transportation CPS},
year = {2026},
booktitle = {Proceedings of the HSCC/ICCPS 2026: 29th ACM International Conference on Hybrid Systems: Computation and Control and 17th ACM/IEEE International Conference on Cyber-Physical Systems},
location = {Saint Malo, France},
keywords = {origin-destination synthesis, travel demand, transportation planning, data fusion, Bayesian methods, public datasets, traffic simulation},
note = {Acceptance rate: 28\%; Short Paper; Track: Systems and Applications},
series = {HSCC/ICCPS '26},
what = {MoveOD presents a framework for synthesizing fine-grained origin-destination commute patterns from publicly available datasets by integrating census data, employment records, and road networks. The approach uses Bayesian decomposition to generate minute-level commute trip distributions while preserving spatial and temporal coherence with observed commuting patterns. The framework leverages public data sources including US Census Community Survey, Longitudinal Employer-Household Dynamics, and OpenStreetMap to generate realistic synthetic commute data.},
why = {High-resolution origin-destination data is essential for transportation planning and traffic management, yet collecting such data through surveys or GPS tracking is expensive and privacy-invasive. Existing synthetic approaches fail to capture temporal and spatial granularity needed for realistic simulation. MoveOD is innovative because it demonstrates how publicly available marginal data can be combined through principled statistical methods to generate detailed, temporally-resolved commute patterns that preserve observed macro-level statistics while enabling microscopic simulations.},
results = {Validation on Hamilton County, Tennessee data demonstrates that the calibrated MoveOD approach accurately reproduces observed census commute patterns while generating realistic minute-level departure time distributions. The framework achieves alignment with ACS travel time margins through careful calibration, enabling fast synthetic data generation suitable for any US county and providing a reusable tool for transportation research.},
project_tags = {transit, planning}
}
High-resolution origin–destination (OD) tables are critical to cyber-physical transportation systems, enabling realistic digital twins, adaptive routing strategies, signal timing optimization, and demand-responsive mobility services. However, such OD data is rarely available outside a small number of data-rich metropolitan regions. We introduce MoveOD, an open-source pipeline that synthesizes publicly available datasets to generate fine-grained commuter OD flows with spatial and temporal departure distributions for any U.S. county. MoveOD fuses American Community Survey travel-time and departure distributions, Longitudinal Employer–Household Dynamics (LODES) residence–workplace flows, OpenStreetMap (OSM) road networks, and building footprint data. Our approach ensures consistency with observed commuter totals, workplace employment distributions, and reported travel durations. MoveOD is integrated with a transportation digital twin, enabling end-to-end CPS experimentation. We demonstrate the system in Hamilton County, Tennessee, generating approximately 150,000 synthetic daily trips and evaluating routing algorithms in a live dashboard.