An adaptive and diversified vehicle routing approach to reducing the security risk of cash‐in‐transit operations

Abstract

We consider the route optimization problem of transporting valuables in cash-in-transit (CIT) operations. The problem arises as a rich variant of the capacitated vehicle routing problem (CVRP) with time windows and pickup and deliveries. Due to the high-risk nature of this operation (e.g., robberies) we consider a bi-objective function where we attempt to minimize the total transportation cost and the security risk of transporting valuables along the designed routes. For risk minimization, we propose a composite risk measure that is a weighted sum of two risk components: (i) following the same or very similar routes, and (ii) visiting neighborhoods with low socioeconomic status along the routes. We also consider vehicle capacities in terms of monetary value carried as per insurance regulations. We develop an adaptive randomized bi-objective path selection algorithm that uses the composite risk measure in choosing alternative paths between origin-destination pairs over a sequence of days. We solve the rich CVRP approximately for each day with updated costs. We test our solution approach on a data set from a CIT delivery service provider and provide insights on how the routes diversify daily. Our approach generates a spectrum of solutions with costrisk trade-off to support decision making.

A concise guide to existing and emerging vehicle routing problem variants

“…Finally, inconsistency can be desired in some applications. For cash-in-transit operations, the routes should be unpredictable from day to day to reduce the risk of robberies (Bozkaya et al 2017, Constantino et al 2017. For the transportation of hazardous materials, safe backup routes should be available in case of adverse weather conditions or to spread the accident risk geographically (Akgün et al 2000).…”

A Comprehensive Mathematical Model for a Location-routing-inventory Problem under Uncertain Demand: a Numerical Illustration in Cash-in-transit Sector

“…For instance, the goal in CIT sector is to minimize transportation cost while boost safe and efficient routes. However, to the best of our knowledge and according to the published scientific works, the only multi-objective papers in CIT sectors can found only a few articles [8,[23][24][25], and there is not any multi-objective problem in LRIP in CIT sector.…”

“…Time window constraint and maximum number of selected facility centers are shown in constraints (22) and (23), respectively. Finally, constraints (24) and (25) define binary and non-negativity conditions on the variables. The formulation is nonlinear due to constraint (13).…”

A cash transportation vehicle routing problem with combinations of different cash denominations

“…Bozkaya et al. () considered two robbery risk components: (i) following the same or very similar routes arranged before and (ii) visiting neighborhoods with low socioeconomic status along the routes. A risk‐based model was proposed to generate alternative routing solutions that make the routes unpredictable.…”

“…To solve the proposed problem, we may consider different solution methods, including exact methods (e.g., Yan et al., ; Van Anholt et al., 2016) and heuristics. The latter includes ant colony optimization with large neighborhood search (e.g., Talarico et al., ), an adaptive and diversified vehicle routing approach (Bozkaya et al., ), a progressive multiobjective optimization with iterative local search (e.g., Talarico et al., ), a variable mixed integer programming neighborhood search (e.g., Larrain et al., ), the multistart and perturb‐and‐improve metaheuristic (e.g., Talarico et al., ), the multistart iterated local search (e.g., Michallet et al., ), an improved ant colony algorithm (e.g., Dai and Liu, ), nearest neighbor algorithm (e.g., Boonsam et al., ), group sweep algorithm (e.g., Boonsam et al., ), and an adaptive memory‐based metaheuristic (e.g., Tarantilis and Kiranoudis, ). The proposed problem is NP‐hard since it is a variant of VRP, exact methods may only obtain optimal solutions in quite small instances in a reasonable time.…”