EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
1562. Deep Controlled Learning for the Delayed Time Window Assignment and Vehicle Routing Problem
Invited abstract in session MD-64: Vehicle Routing Under Uncertainty 1, stream VeRoLog - Vehicle Routing and Logistics.
Monday, 14:30-16:00Room: S16 (building: 101)
Authors (first author is the speaker)
| 1. | Layla Martin
|
| Operations, Planning, Accounting and Control, Eindhoven University of Technology | |
| 2. | Sifanur Celik
|
| Department of Industrial Engineering and Innovation Sciences, Eindhoven University of Technology | |
| 3. | Albert Schrotenboer
|
| Department of Industrial Engineering and Innovation Sciences, Eindhoven University of Technology | |
| 4. | Tom van Woensel
|
| Technische Universiteit Eindhoven |
Abstract
When delivering chilled/frozen, expensive, or odd-sized goods, customers must be present to receive the shipment. In current operations, this often results in failed delivery attempts or considerable inconvenience for customers who need to be present long, often the entire day. To alleviate this inconvenience, logistics service providers (LSPs) moved towards communicating time windows, either directly after the customer shipment is received or batched the day before. Longer waiting for a time window poses an inconvenience for customers. We suggest a scheme in which the LSP decides (i) whether to assign a time window to a newly arriving customer, and if the LSP assigns a time window, also if some previously postponed customers shall also receive a time window, (ii) the actual time window, and (iii) the routes satisfying the time windows. To solve the resulting semi-Markov decision process, we adapt three solution methodologies: Deep Controlled Learning, the Rollout Algorithm, and the Multi-Scenario Approach. We show that Deep Controlled Learning manages to balance solution quality and runtime. We further find that customers located very close or very far from the depot receive time windows quickly, while other customers have to wait longer. On the other hand, customer arrival time does not influence the waiting time substantially.
Keywords
- Vehicle Routing
Status: accepted
Back to the list of papers