The agent is a double-jointed arm. The goal is to move to a target location.
- States: 33
- Actions: 4 (torque applicable to two joints)
- Rewards: +0.04 for each step that the agent's hand moves toward the goal location
Open ContinuousControl.ipynb Jupyter notebook, it also contains a section on how to install the necessary dependencies.
Download the Reacher environment (binary file) for your operating system and configure file_name variable (in ContinuousControl.ipynb Jupyter notebook) to point to the location of the binary file.
- Linux: click here
- Mac OSX: click here
- Windows (32-bit): click here
- Windows (64-bit): click here
Open ContinuousControl.ipynb Jupyter notebook and execute each code block to train the agent using DDPG.