[Feat] Adding Multi-depot Capacity Pickup and Delivery Problem (MDCPDP) Environment

[cbhua]

Description

The multi-depot capacity pickup and delivery problem (MDCPDP) environment has the following definition:

• The environment is made of num_loc + num_depots locations (cities):
  • num_depot depot;
  • num_loc / 2 pickup locations;
  • num_loc / 2 delivery locations;
• The goal is to visit all the pickup and delivery locations in the shortest path possible starting from the depot;
• The conditions is that the agent must visit a pickup location before visiting its corresponding delivery location;
• The capacity is the maximum number of pickups that the vehicle can carry at the same time;

Motivation and Context

Aside from previous simple pickup and delivery (PDP) environment, this new environment supports the following new features:

1. Multi depots: there are more than one depot and each depot will have a vehicle, which makes the problem more flexible and complex;
2. Lateness reward: we support the reward calculated as the lateness, i.e. cumulate sum of the arrival time to delivery nodes;
3. Multi distance calculation modes: we support L1 and L2 distance calculation methods for more flexible reward calculation;
4. Multi reward calculation modes: we support the lateness, lateness-square, min-max and min-sum reward calculation methods as objectives;
5. Multi start modes: you could choose which depot starts firstly or let them randomly start;

Types of changes

• [x] New feature (non-breaking change which adds core functionality)
• [x] Example (update in the folder of examples)

Checklist

• [x] My change requires a change to the documentation.
• [x] I have updated the tests accordingly (required for a bug fix or a new feature).
• [TODO]: I have updated the documentation accordingly.

[fedebotu]

LGTM