Application Framework for Conveyor Belt-based Pick-and-Place Systems

---

Requirement

1. OS: Linux Ubuntu 20.04LTS
2. Unity Hub: v.2020.2.2f1, ML-Agent 1.7.2
3. ROS Noetic
4. Install python3, pytorch

Run ROS service

1. Move to ROS service folder: cd ros-app
2. Source ROS command: source /opt/ros/noetic/setup.bash
3. Build ROS service: catkin_make
4. Source devel: source devel/setup.bash
5. Run ROS service: roslaunch niryo_moveit bin_picking.launch

Working with Unity App

1. Download Unity Hub (v.2020.2.2f1)
2. Open Unity project located at folder unity-app
3. Build Unity Simulator (I already build a simulator called simulator.x86_64 that existed in folder unity-app)
4. Execute simulator app (Default algorithm is FSFP (a derived version of FIFO))

Training Agent

1. Run environment only: python3 robot_env.py
2. Training agent with default parameters (given in trainning.py file): python3 training.py