Arm Sample

As one of the projects in the Robotics Software Engineer Nanodegree program, you will explore how to train a robotic arm to touch objects without needing explicit inverse kinematics.

The repository contains a gazebo-arm.world file that defines the environment with three main components:

• The robotic arm with a gripper attached to it.
• A camera sensor, to capture images to feed into the DQN.
• A cylindrical object or prop.

## Running the Rover

To get started with the arm sample, open the desktop in the Udacity Workspace, open a terminal, and once again navigate to the folder containing the samples with:

$ cd /home/workspace/jetson-reinforcement/build/x86_64/bin

Launch the executable from the terminal:

$ ./gazebo-arm.sh

Once the gazebo environment loads up, you will observe the robotic arm, a camera sensor, and an object in the environment. The gazebo arm will fall to the ground after a short while, and the terminal will continuously display the following message:

ArmPlugin - failed to create DQN agent

Since no DQN agent is currently defined, the arm isn't learning anything and has no input to control it. For this sample, it's up to you to implement your own DQN agent!

If you'd like to learn how to train the robotic arm from the robotics experts at Udacity, you're encouraged to explore the Robotics Software Engineer Nanodegree program!