AUTOMATION AND CONTINUOUS IMPROVEMENT IN LAST-MILE LOGISTICS: THE AMAZON DELIVERY STATION CASE STUDY

This thesis is the result of an internship experience carried out at an Amazon Delivery Station located in the Venice area, an operational environment characterized by high parcel volumes, process variability, and an increasing level of automation. The research focuses on the analysis and improvement of the ADTA (Auto Divert to Aisle) system, a key component of the parcel sorting process within last-mile logistics. The objective of the study is to identify sources of inefficiency and process instability and to develop targeted actions aimed at enhancing the reliability and overall performance of the system. To achieve this goal, the work adopts a Lean oriented methodology structured around the PDCA cycle, integrating direct process observation, root-cause analysis, and data-driven evaluation. This approach enabled the identification of several operational challenges affecting parcel flow and machine stability, and supported the design and implementation of improvement actions, complemented by verification plans for initiatives still pending execution. The analysis carried out during the operational stabilisation period showed an overall improvement in ADTA Quality, a reduction in key categories of Technical Failures and Process Errors, and a more stable and predictable parcel flow. These results demonstrate how the combination of automation, data analysis, and continuous improvement methodologies represents a fundamental driver for optimising last-mile logistics processes and provides a replicable framework for future enhancement initiatives in highly automated operational contexts.