THE SYNCHRONIZED MULTI-ASSIGNMENT ORIENTEERING PROBLEM

Abstract

We introduce the Synchronized Multi-Assignment Orienteering Problem (SMOP), a vehicle routing problem that requires jointly selecting a set of jobs while synchronizing the assignment and transportation of agents to roles to form ad-hoc teams at different job locations. Agents must be assigned only to roles for which they are qualified. Each job requires a certain number of agents in each role within a time window and contributes a reward score if selected. The task is to maximize the total reward attained. SMOP can model many real-world scenarios requiring coordinated transportation of resources and accommodates traditional depot-based workforces, depot workforces supplemented by ad-hoc workers, and fully ad-hoc workforces alike. The same problem formulation can be used for initial planning and mid-course replanning. We develop a mixed integer programming formulation (MIP) and an Adaptive Large Neighborhood Search algorithm (ALNS). In computational experiments covering a range of considerations, ALNS consistently found very near-optimal solutions on smaller problems and surpassed a commercial MIP solver substantially on larger problems. ALNS also found 24 new best solutions on a set of benchmark problems from the literature for the related Cooperative Orienteering Problem with Time Windows.