SUSTech EE368 Robotic Motion and Control Project

Introduction

This project is a robotic system that can play Gomoku with a human player.

We use Intel D435 camera and Kinova Gen3 Lite robot arm to build the system. ROS1 neotic, OpenCV, and AlphaZero are used in the project.

Usage

First build the workspace:

$ catkin_make
$ source devel/setup.bash

For the completed function, you need to run the following three parts:

Gomoku Vision

$ roscore
$ rosrun gomoku_vision vision_node.py

If you can not connect to the camera (that's very common), you can use the planc to simulate the camera function:

$ rosrun gomoku_vision planc.py

Gomoku AI

$ rosrun gomoku_ai ai_node.py

Gomoku Control

$ roslaunch kortex_driver kortex_driver.launch
$ roslaunch gomoku_control pick_and_place_example.launch

Test example: $ rostopic pub /my_gen3_lite/arm_point geometry_msgs/Point '{x: 0.427, y: 0.07, z: 0.1}'

Development

Software Architecture

Below is the system architecture:

Framework: ROS1 Neotic (Ubuntu 20.04)

Nodes:

• gomoku_vision
• gomoku_ai
• gomoku_control

Messages:

• /arm_pose: geometry_msgs/Point

  The desired position of the piece to be placed. From gomoku_vision to gomoku_control

• /piece_position: std_msgs/UInt8MultiArray

  from gomoku_vision to gomoku_ai

• /next_move: std_msgs/UInt8MultiArray

  from gomoku_ai to gomoku_vision

• /victory: std_msgs/Int8

  from gomoku_vision to gomoku_control

Hardware Architecture

• Kinova Gen3 Lite: 6DOF Robot Arm
• Intel RealSense D435: RGBD Camera
• Lego 2x2 cylinder bricks: Gomoku pieces

Commits

Follow the format proposed in the Semantic Commits

Contributors

This project is developed by the group 1 of EE368 Robotic Motion and Control course in SUSTech.

License

Apache License 2.0