Project 2: Continuous Control

Introduction

For this project, the Reacher environment is used.

In this environment, a double-jointed arm can move to target locations. A reward of +0.1 is provided for each step that the agent's hand is in the goal location. Thus, the goal of your agent is to maintain its position at the target location for as many time steps as possible.

The observation space consists of 33 variables corresponding to position, rotation, velocity, and angular velocities of the arm. Each action is a vector with four numbers, corresponding to torque applicable to two joints. Every entry in the action vector should be a number between -1 and 1.

A single agent is used in this environment.

Solving the Environment

The task is episodic, and in order to solve the environment, the agent must get an average score of +30 over 100 consecutive episodes.

Getting Started

1. Download the environment from one of the links below. You need only select the environment that matches your operating system:

   • Version 1: One (1) Agent
     • Linux: click here
     • Mac OSX: click here
     • Windows (32-bit): click here
     • Windows (64-bit): click here

   (For Windows users) Check out this link if you need help with determining if your computer is running a 32-bit version or 64-bit version of the Windows operating system.

   (For AWS) If you'd like to train the agent on AWS (and have not enabled a virtual screen), then please use this link (version 1) or this link (version 2) to obtain the "headless" version of the environment. You will not be able to watch the agent without enabling a virtual screen, but you will be able to train the agent. (To watch the agent, you should follow the instructions to enable a virtual screen, and then download the environment for the Linux operating system above.)

2. Place the file in the root folder, and unzip (or decompress) the file.
   
   

Instructions

Follow the instructions in Continuous_Control.py to get started with training your own agent. In particular, everything is arranged so that you provide a folder name where the results will be saved and that's it. Progress is shown every two seconds by default and the agent's progress is saved (by default) every minute - this includes performance curves to make it easy to compare different models.

In misc.py you'll find useful functions for loading data and plotting it. Examples can be found in Report_Helper.ipynb.

A smart agent can be watched by running Watch_Pretrained_Agent.py.

Room to improve:

• Currently the code relies on defaults set in ddpg_agent.py for things like learning rates, network architecture, and actually most things. Choosing between regular and batchnorm'ed networks has to be done by commenting and un-commenting code in model.py. My usage was to include as many of these parameters as possible in the folder name.