AIMS: Artificial Intelligence for the Management of Shifts

Research Overview

The AIMS project (Artificial Intelligence for the Management of Shifts) was a research initiative funded by the Tuscany Region (POR CreO 2014-2020). The project focused on developing advanced optimisation tools for personnel shift scheduling based on real-world activity demands.

Our research unit, the Global Optimization Laboratory (GOL) at the University of Florence, in collaboration with FirLab s.r.l., spearheaded the development of mathematical models to solve the Integrated Task Scheduling and Personnel Rostering (TSPR) problem.

Our Scientific Contribution: Integrated TSPR

Historically, personnel management issues—staff rostering (defining shifts) and activity assignment (scheduling tasks)—have been treated separately and sequentially. Our research demonstrates that treating these as a single, integrated problem significantly increases solution quality.

Key Innovation: Mixed-Integer Linear Programming (MILP)

We formalised the TSPR problem into a Mixed-Integer Linear Programming (MILP) model. This model simultaneously decides:

  • The activation of employees.
  • The definition of optimal shifts.
  • The assignment of specific tasks (e.g., check-in, cleaning, boarding) to those shifts.

Mathematical Advancements for Tractability

To handle large-scale data from international airports, our unit introduced clique-based valid inequalities.

  • Clique Inequalities: These mathematical constraints strengthen the model by cutting off “unfeasible” fractional solutions during computation, allowing for near-optimal solutions in reasonable timeframes.
  • Constraint Management: The model incorporates complex real-world requirements, including heterogeneous workforce skills, contractual limits on weekly hours, and mandatory 11-hour resting periods between shifts.

Validation: The Airport Case Study

The research was validated using real-world data from the Bergamo Orio al Serio Airport. Airport environments are uniquely challenging due to 24/7 operations and highly time-varying demand.

Results showed that our integrated model:

  • Significantly reduced idle time compared to traditional sequential methods.
  • Lowered the total number of active workers needed to cover 100% of the demand.
  • Outperformed the meta-heuristic models currently used in production systems by ensuring higher efficiency and strict adherence to legal constraints.

Official Project Partners

References

  • Cappanera, P., Di Gangi, L., Lapucci, M., Pellegrini, G., Roma, M., Schoen, F., and Sortino, A. (2024). Integrated task scheduling and personnel rostering of airports ground staff: A case study. Expert Systems with Applications, 238, 121953. DOI: 10.1016/j.eswa.2023.121953.

Project Funding and Logos

This project was supported by the POR CreO FESR 2014-2020 funds of the Tuscany Region.

     
POR CreO Toscana - Unione Europea - Repubblica Italiana - Regione Toscana Università di Firenze FirLab