Getting Started
Note
This page is currently under construction
Currently the best way of getting started is to:
- Look at the examples from the main repository
- See below for suggestions on which examples to focus on depending on your background
- While going through examples, reference the following two pages
- Python API for building models
- MOOS-IvP Reference for configuring new missions
If you are coming from MOOS-IvP
I would recommend looking at the "ManagerExample" and look at the configuration blocks for the BHV_Agent. This behavior is one that does the actual connection to python land through a TCP socket.
Then, focus on the python side of the more complex "QTable" mission below.
If you said "What is MOOS-IvP"
Look at the "QTable" example. This implements a reinforcement learning q-table. This example is trained on the Project Aquaticus scenario. It's goal is to grab a flag in virtual game of capture flag.
If you came from the MOOS-IvP section:
The associated MOOS-IvP mission files which are in that directory are based on the moos-ivp-agent's AgentAquaticus which allows for spawning of multiple vehicles in one MOOS-IvP simulation during training time. A good config block to look at is that for pEpisodeManager which manages the resetting of vehicles to form episodes of training.
