Research

We conduct an active research program. Internally and with ClickSoftware, we explore practical optimisation methods applicable to field service optimisation and related fields. The theoretical foundation behind Xoom in its more generic form is subject of a separate research effort. Most of our research is intended for internal development purposes, but some of it has been published in academic literature. Below is the list of such publications.

Optimization Strategies for Restricted Candidate Lists in Field Service Scheduling

Marko Žerdin, Alexander Gibrekhterman, Uzi Zahavi and Dovi Yellin

Abstract
Field service scheduling (FSS) is a large class of practical optimization problems combining features of the vehicle routing problem (VRP), scheduling problems and the general assignment problem (GAP). In some cases the problem reduces to well known variants of VRP, while other, more common circumstances give rise to a distinct set of optimization problems that have so far received very little attention in the literature. In this chapter we show how strategies for restricted candidate lists (RCL) – methods for pre-calculating and contextualizing the candidate list reduction procedures within a context of a generic optimization framework, can be used to efficiently solve a wide spectrum of FSS instances in a real-life industrial environment. A comparison of results obtained using a greedy randomized adaptive search procedure (GRASP) meta-heuristic with and without the use of certain RCL strategies is presented as it applies to specific variants of the problem.

When the Rubber Meets the Road: Bio-inspired Field Service Scheduling in the Real World

Israel Beniaminy, Dovi Yellin, Uzi Zahavi, and Marko Žerdin

Abstract
We discuss a class of large-scale real-world field service optimization problems which may be described as generalizations of the Vehicle Routing Problem with Time Windows (VRPTW). We describe our experience in the real-world issues concerned with describing and solving instances of such problems, and adapting the solution to the needs of service organizations using a ”universal framework” for bringing together various problem representations and experimenting with different algorithms. Implementations and results of several bio-inspired approaches are discussed: Genetic Algorithm (GA), Ant Colony Optimization (ACO), and a hybrid of ACO with GRASP (Greedy Randomized Adaptive Search Procedure). We conclude by discussing generation of ”human-friendly” solutions, through introduction of local considerations into the global optimization process.