The objective of this repo is to solve the Vehicle Routing Problem using established deep Reinforcement Learning techniques, primarily (Nazari et. al 2018).
The stretch objective is to improve on this deep RL algorithm.
Task 1: Replicate results from (Nazari et. al 2018)
Task 1.1: (Stretch) Replicate results from (Bello et. al 2017; Vinyals et. al 2017).
Task 2: (Stretch) Develop a stronger RL algorithm than Task 1
Dependencies:
Gym, TensorFlow, Pandas, NumPy, PyTest, flake8, mypy
Go to src and run pip install -e . which will identify the relevant files.
VRPEnv5 is the CVRP problem with 5 destinations, 1 depot and 1 (capacitated vehicle).
This repo was developed by Stephen Mak. Contact me if you would like to learn more! :)