Answers are grounded in this paper and cite their sources.
This paper addresses a stochastic variant of the car patrolling problem, in which unexpected alarms introduce uncertainty into route planning for private security services provided by a large Italian company. Each customer has different weekly demands (i.e., services), which can be optional or mandatory. Some services (i.e., alarms) are not known in advance and are thus treated as stochastic information. The company is concerned about providing an immediate response to alarms triggered during a patrol. The objective is to determine a set of patrol routes, one per day, that comply with several operational constraints, including hard time windows, maximum riding time, the minimum number of services performed, and the uncertainty regarding alarm triggers. Since our interest is in solving real‐world instances, we propose a scenario‐based iterated local search heuristic. It incorporates diverse operators, along with a local search and recourse function, to guide the search toward robust solutions. Computational experiments demonstrate that solutions that ignore unexpected alarms frequently deteriorate under stochastic evaluation, with losses exceeding 16% in some cases. In contrast, the proposed heuristic improves both robustness and solution quality, resulting in gains of up to 7% and reducing service cancellations. Further analysis demonstrates the sensitivity of solution performance to scenario diversity and sample size, highlighting the importance of carefully selecting scenarios that reflect alarm occurrences.
No takes yet. Share an insight, caveat, or question.
Corrêa et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: