Project Content

Technical Framework: ROS2 humble + GAZEBO environment YOLOv8: https://github.com/ultralytics/ultralytics
Work:

Branch base: Basic tasks, complete voice command recognition + visual target detection + SLAM + navigation planning + simple grasping; (Node structure diagram below)
Branch master: Multi-tasking, on the basis of basic tasks, can perform sequential tasks - go to multiple locations in sequence to perform grasping of different targets, added angle calibration, increased scene waypoints. (Node structure diagram below)

Project Effects

Go to the mybot_description folder to view

config: Configuration files generated by the moveit assistant for the robotic arm, used in conjunction with moveit to control the robotic arm;
launch: Launch files
- display_rviz2.launch: Start rviz2;
- gazebo.launch: Start gazebo, load map and robot, load joint states and their control;
- move_arm.launch: Start moveit;
- my_moveit_rviz.launch: Start moveit and rviz;
meshes: Store structural design;
mybot_description: Node files
- audio_classify: Language node, microphone input to text, then input to the large model for classification, output to the target detection node;
- image_detection: Target detection node, input RGB/depth camera module images to YOLO8 for recognition, get target position and distance, output data for specified category targets;
- move_arm: Node to control the robotic arm to grasp objects;
- move_claw: Node for debugging grasping actions only;
- moveit_move_arm: Node to control the robotic arm using moveit's motion group planning;
- rotate_wheel: Node to control the movement of the cart's wheels;
- navigate: Given a target point and final direction, the robot navigates to the destination and turns to that direction;
urdf: Store robot design
- my_car: Cart without a robotic arm, used for SLAM mapping;
- six_arm: Six-axis robotic arm mobile robot;
weights: Store YOLOv8 model weights;
world: Store maps.