A Simulated Annealing Algorithm to Solve the Multi-objective Bike Routing Problem

Solving the multi‐objective bike routing problem by meta‐heuristic algorithms

“…Other meta-heuristics were implemented in routing context by Osaba et al (2018) to balance the trade-off between the length of the route and its safety level, and by Chang et al (2005) to find a set of nondominated paths in a dynamic network. Nunes et al (2020a) implemented a Multi-Objective Simulated Annealing (MOSA) approach to solve the MOBRP, based on the works developed by Suppapitatnarm et al (2000) and Sankararao and Yoo (2011). Despite being a common method employed to solve multiobjective problems, evolutionary approaches, such as genetic algorithms represent some challenges in MSP problems, because operators such as crossover and mutation should respect the nodes' connectivity in the graph.…”

“…The proposed approach is tested with two distinct real cycling networks, and compared with three other approaches, in terms of computational cost and the quality of the final set of solutions. The approaches tested are: the NSGA-II, a genetic-based approach, whose mutation operator is inspired in the work presented by Nunes et al (2020a); the genetic algorithm proposed by Nunes et al (2020b), that is a genetic approach in which a mutation operator was not applied; and the MOSA proposed by Nunes et al (2020a), which is a different meta-heuristic applied to the MO-BRP. Note that the proposed approach differs from the NSGA-II, since a new mechanism to select individuals and a semi-controlled population size schema during the creation of new populations, is applied.…”

“…Note that the proposed approach differs from the NSGA-II, since a new mechanism to select individuals and a semi-controlled population size schema during the creation of new populations, is applied. Moreover, to apply these algorithms, a mutation operator for the MOBRP, inspired in the work proposed by Nunes et al (2020a), is proposed.…”

“…This section presents the results of computational experiments performed to test the effectiveness of the NGA-MOBRP, by comparing the quality of the obtained solutions and computational time with the well-known NSGA-II and with the works proposed by Nunes et al (2020aNunes et al ( , 2020b. Subsection 5.1 presents the computational setup and the parameters adopted for each algorithm, as well as the test instances.…”

“…One of the most popular meta-heuristics for multi-objective problems is the Nondominated Sorting Genetic Algorithm II (NSGA-II) proposed by Deb et al (2000), however, when applied to the MOBRP, this approach revealed some limitations, which are mitigated by the new genetic approach (NGA-MOBRP) proposed in this work. The computational experiments compare these approaches with the works proposed by Nunes et al (2020aNunes et al ( , 2020b.…”