An Effective Large Neighborhood Search for the Team Orienteering Problem with Time Windows.

We propose an effective metaheuristic for the Team Orienteering Problem with Time Windows. The metaheuristic is based on the principle of Large Neighborhood Search and can outperform the performance of algorithms available in the literature. We provide computational experiments for well known benchmark instances and are able to compute new best solutions for 17 of these instances. On average, the gap between our results and best known solutions so far is below 1%, and our solution approach yields 70% of the best known solutions available in the literature. The new results can serve as benchmarks for future computational studies.