Deep Reinforcement Learning with a Robotic Arm

Udacity Robotics  ​​
Utilization of Deep Q Learning to Train a Robotic Arm to Touch an Object
​Full Implementation Using Gazebo, Rviz, and ROS

  1.  Used Deep Q Learning to ensure the gripper of a robotic arm touched an object
  2.  Created a reward function to ensure that training could be accomplished in a low number of iterations
  3. Utilized LSTM and parameters were tuned to ensure high performance

More details about this project can be seen in the paper attached below.

Project ReportDownload