Using Deep Q-Learning (Alpha-Go algorithm) to solve routing problems in warehouse logistics

From ISLAB/CAISR
Revision as of 17:00, 19 October 2018 by Jendav (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Title Using Deep Q-Learning (Alpha-Go algorithm) to solve routing problems in warehouse logistics
Summary Solving the famous vehicle-routing problem using Deep Q-Learning
Keywords vehicle routing problem, Reinforcement learning
TimeFrame
References
Prerequisites
Author
Supervisor Jennifer David, Thorsteinn Rögnvaldsson
Level Master
Status Open


For a group of robots or forklifts involved in warehouses, Sony Mobile in Lund has a Masters Thesis project. The aim is to generate paths/routes for all vehicles according to a pre-determined schedule without any path-conflicts and schedule conflicts. This is an NP-hard problem termed as vehicle routing problem and has been solved using heuristic approaches. Here, we try to use Deep Q learning to solve it.