Author: Leandro Do Carmo Martins
Programme: Doctoral Programme in Network and Information Technologies
Language: English
Supervisors: Dr Angel A. Juan Pérez i Dr Helena Ramalhinho
Faculty / Institute: Doctoral School UOC
Subjects: Computer Science
Key words: agile optimization, biased-randomized heuristics, simheuristics, metaheuristics, real-time optimization

Area of knowledge: Network and Information Technologies

Summary

Transportation and logistics (T&L) activities play a vital role in the development of many businesses from different industries. With the increasing number of people living in urban areas, the expansion of on-demand economy and e-commerce activities, the number of services from transportation and delivery has considerably increased. Consequently, several urban problems have been potentialized, such as traffic congestion and pollution. Several related problems can be formulated as a combinatorial optimization problem (COP). Since most of them are NP-Hard, the finding of optimal solutions through exact solution methods is often impractical in a reasonable amount of time. In realistic settings, the increasing need for 'instant' decision-making further refutes their use in real life. Under these circumstances, this thesis aims at: (i) identifying realistic COPs from different industries; (ii) developing different classes of approximate solution approaches to solve the identified T&L problems; (iii) conducting a series of computational experiments to validate and measure the performance of the developed approaches. The novel concept of 'agile optimization' is introduced, which refers to the combination of biased-randomized heuristics with parallel computing to deal with real-time decision-making.