Monitoring the dynamics of traffic in major corridors can provide invaluable insight for traffic planning purposes. An important requirement for this monitoring is the availability of methods to automatically detect major traffic events and to annotate the abundance of travel data. This paper introduces a machine learning-based approach for reliable detection and characterization of highway traffic congestion events from hundreds of hours of traffic speed data. Indeed, the proposed approach is a generic approach for detection of changes in any given time series, which is the wireless traffic sensor data in the present study. The speed data is initially time-windowed by a 10 h-long sliding window and fed into three neural networks that are used to detect the existence and duration of congestion events (slowdowns) in each window. The sliding window captures each slowdown event multiple times and results in increased confidence in congestion detection. The training and parameter tuning are performed on 17,483 h of data that include 168 slowdown events. These data are collected and labeled as part of the ongoing probe data validation studies at the Center for Advanced Transportation Technologies at the University of Maryland. The neural networks are carefully trained to reduce the chances of over-fitting to the training data. The experimental results show that this approach is able to successfully detect most of the congestion events, while significantly outperforming a heuristic rule-based approach. Moreover, the proposed approach is shown to be more accurate in estimation of the start time and end time of the congestion events.

No abstract

In the school bus scheduling problem, the main contributing factor to the cost is the number of buses needed for the operations. However, when subcontracting the pupils' transportation, unbalanced tours can increase the costs significantly as the lengths of some tours can exceed the daily fixed driving goal and will result in over-hour charges. This paper proposes an MIP model and a matching-based heuristic algorithm to solve the "balanced" school bus scheduling problem with fixed start times in a multi-school setting. The heuristic solutions always have the minimum number of buses as it starts with a minimal number of tours and does not alter the number of tours during its balancing stage. The effectiveness of the heuristic is tested by comparing its solutions with results from solving the MIP using commercial solvers whenever solvers could find a good solution. To illustrate the performance of the MIP and the heuristic, 11 problems were examined with different numbers of trips which are all based on two real-world problems: a California case study with 54 trips and the Howard County Public School System with 994 trips. Our numerical results indicate the proposed heuristic algorithm can find reasonable solutions in a significantly shorter time. The balanced solutions of our algorithm can save up to 16% of school bus operation costs compared to the best solution found by solvers from optimizing the MIP model after 40 hours. The balancing stage of the heuristic decreases the Standard Deviation of the tour durations by up to 47%.

Taxi-sharing is a known solution for reducing congestion and is more beneficial when the traveled miles, imposed by detours while serving additional passengers, are minimized. This study proposes incorporation of alternative meeting points in taxi-sharing routes to boost the efficiency of the system by eliminating unnecessary detours and improving the chances of passengers being matched. Unlike most ridesharing systems, in the proposed approach, passengers are not necessarily picked up or dropped off at their original location, but at a walkable distance from their origin. The proposed framework deals with practical challenges of including meeting points in a real-world high-demand taxi-sharing system with hundreds of requests per minute. This includes efficient selection of alternative pick up locations and incorporation of these alternatives into a novel mixed integer linear programming (MILP) formulation to find the optimal schedule. Using the 2015 New York City (NYC) yellow cab dataset, first, the potential benefits of introducing meeting points in Manhattan road network are demonstrated. Given the Nondeterministic Polynomial-time Hard (NP-hard) nature of the associated optimization problem, the problem is then broken down into smaller-sized problems forming clusters of passengers with high potential of sharing a ride. Then, the proposed MILP model is used to find the optimal route for each cluster, while selecting the best pickup point for each passenger. Testing on a sample of the NYC dataset, it is shown that the proposed methodology improves the efficiency of the taxi-sharing system by reducing the wait times by about 50% while considerably reducing total travel times and the number of vehicles used.

No abstract