Assessing the benefits of more complex commitment strategies in dynamic assignment in the context of crowdsourced deliveries
Files
Innocente_36661600_2021.pdf
Open access - Adobe PDF
- 2.02 MB
Details
- Supervisors
- Faculty
- Degree label
- Abstract
- Crowdsourcing has the potential to reduce the negative side effects – whether environmental or economic or even related to the quality of the service – of freight transport but requires fast decision making as drivers are available for a short time to fulfill a task. Besides, the stochasticity and dynamics in this assignment problem requires assignment decisions to be taken before the availability of future resources and/or tasks gets known. Efficient optimization techniques are thus needed to provide optimal solutions with a short computation time. We therefore develop a dynamic matching algorithm along with a commitment strategy in order to capture real-time changes in the system. In this master thesis, we aim at demonstrating the value of implementing more complex commitment strategies since Agatz et al. (2011) suggest that using sophisticated optimization methods instead of simple greedy matching rules can substantially improve the performance of the system. Yet, existing papers are limited to implementing rather simplistic strategies. Instead of simply deciding to confirm the first match found with the risk of missing a better solution in the future or waiting until the last minute with the risk of losing an already good solution, we tackle these risks by evaluating the detour we can expect in the future using a probability distribution. This provides an objective criterion allowing to know with more certainty what the future holds in terms of detour and upon which to decide if it is reasonable to confirm tentative matches to drivers at time t or if, on the contrary, it is interesting to wait because we can expect a better solution in the remaining delivery time. The results appear to be conclusive as this strategy leads to a reduction between 13,88 and 43,88% of the detour to deliver a package via crowdshipping as well as of the cost of crowdshipping between 8,97 and 28,56%, compared to the four main strategies in the existing literature, which confirms the assumption by Agatz et al. (2011). It also appears in the results that the commitment strategy designed in this master thesis provide better results than any other strategy of the existing literature, whatever the values of the parameters of the model.
Citations
Innocente, E. (2021). Assessing the benefits of more complex commitment strategies in dynamic assignment in the context of crowdsourced deliveries. https://dial-mem.test.bib.ucl.ac.be/handle/2078.2/24461