Vehicle pick-up and drop-off schedule optimization in a university setting

Teo, Chun Kit (2024) Vehicle pick-up and drop-off schedule optimization in a university setting. Final Year Project, UTAR.

Preview

PDF Download (2954Kb) | Preview

Abstract

This project aims to enhance the convenience and safety of university students and staff by developing an optimized vehicle pick-up and drop-off scheduling system, integrating aspects of the Dial-a-Ride Problem (DARP) and the Carpooling Problem (CPP) to better suit the university setting. The formulated problem is a multi-objective, many-to-many, one-day scheduling problem with both static and dynamic components. Uniquely, participants can serve as both drivers and passengers, with fairness constraints applied equally to both roles. The primary objectives include minimizing earliness waiting times, reducing DARP-like cases, and lowering total expenses, with penalties imposed for unserved requests. Lateness will be removed using a lateness waiting time rollback mechanism. A simulated annealing-based multi-directional iterative local search algorithm is employed for solution optimization. The initial solution is generated by distributing requests across vehicles, and local searches are performed through request swapping and movement. Simulated annealing explores the solution space in multiple directions to avoid convergence to suboptimal solutions, with iterative loops preventing premature convergence. For dynamic requests, a handler evaluates acceptance based on time constraints, and schedule re-optimization is triggered as necessary, using the same methods as in the static case. Extensive experiments validate the algorithm’s effectiveness, optimize parameters, and demonstrate the dynamic handler's ability to manage real-time requests accurately. The results confirm the efficiency and robustness of the proposed approach in both static and dynamic scenarios.

Actions (login required)