6-DOF Robotic Arm Simulation

This project demonstrates a complete robotic arm system using ROS 2 Humble, Gazebo, MoveIt 2, and YOLOv8. The system is built around a UR5 6-DOF robotic arm and includes simulation, motion planning, kinematics-based control, and vision-based object detection.

---

Features

• 6-DOF UR5 robotic arm simulation
• MoveIt 2 integration for trajectory planning
• Custom inverse and forward kinematics implementation
• Joint angle control via ROS 2 publishers
• Simulated RGB-D camera mounted on robot
• YOLOv8 object detection on camera feed

---

Demo Videos

Task	Description	Demo
Task 1	UR5 Simulation + MoveIt 2	Watch Video
Task 2	Kinematics Node (IK/FK) + Joint Control	Watch Video
Task 3	Object Detection using Simulated Camera + YOLOv8	Watch Video

---

Requirements / Dependencies

• Ubuntu 22.04
• ROS 2 Humble
• Gazebo (Fortress or compatible)
• Python 3.10+
• ultralytics (YOLOv8)
• OpenCV
• cv_bridge, sensor_msgs, trajectory_msgs
• IKPy

Install Python dependencies:

pip install -r requirements.txt

---

Installation Instructions

git clone --recurse-submodules <repo-url>
cd ros2_ur5_ws
rosdep install --from-paths src --ignore-src -r -y
colcon build --symlink-install
source install/setup.bash

---

TASK 1 – SIMULATION & MOVEIT 2

Description

Simulates the UR5 robotic arm in Gazebo and visualizes it in RViz. Integrates MoveIt 2 to plan and execute trajectories.

Run Instructions

ros2 launch ur5_sim spawn_ur5_moveit.launch.py

Topics Used

• /joint_states (Published)
• /tf (Published)
• /move_group action interface (MoveIt)

Notes

• You can use the interactive marker in RViz to set a new pose and click “Plan” and then “Execute”.
• TF and JointStatePublisher work in sync.

---

TASK 2 – KINEMATICS CONTROL (FK/IK)

Description

Implements forward and inverse kinematics using IKPy. Accepts 3D target position as input, computes joint angles, and publishes them to move the robot in Gazebo.

Run Instructions

# Compute FK and IK
ros2 run ur5_kinematics ik_command_node --ros-args -p x:=0.5 -p y:=0.4 -p z:=0.8

Topics Used

• /joint_trajectory_controller/joint_trajectory (Published)
• /joint_states (Subscribed)

Notes

• The IK node loads the robot’s URDF dynamically.
• Uses 6 active links including wrist and base.
• You can send different XYZ goals for testing.

---

TASK 3 – OBJECT DETECTION WITH YOLOv8

Description

A simulated RGB camera is mounted on the robot’s end-effector in Gazebo. Captures live feed, runs YOLOv8 detection, and shows real-time bounding boxes using OpenCV.

Run Instructions

ros2 run ur5_yolov8_detector yolo_node

Topics Used

• /camera/image_raw (Subscribed)
• cv2.imshow() (Pop-up detection window)
• Terminal logs for object classes and confidence scores

Notes

• Uses ultralytics YOLOv8 model (e.g., yolov8n.pt)
• Simulated camera is placed to view the objects on the table.
• You may add custom textures or PNG images to improve detection.

---

Launch File

File	Description
spawn_ur5_moveit.launch.py	Spawns robot in Gazebo + RViz + Launches MoveIt 2

---

Project Directory Structure (Simplified)

ros2_ur5_ws/
├── src/
│   ├── ur5_sim/
│   ├── ur5_kinematics/
│   └── yolov8_detector/
├── install/
├── build/
├── log/
├── README.md
└── requirements.txt

---

Usage Notes

• Use XYZ target values that are reachable by UR5.
• Make sure to position objects where the camera can see them.
• You can add more objects using SDF or include URDF models.

---

Troubleshooting

• No detection output?

  • Ensure camera is publishing /image_raw
  • Add larger, common YOLOv8-detectable objects (bottle, cup)

• CV Bridge error?

  • Install with: sudo apt install ros-humble-cv-bridge

---

Credits / Acknowledgements

• Universal Robots UR5 model
• MoveIt 2
• Gazebo Sim
• IKPy library
• YOLOv8 (Ultralytics)

---

Contact

Azhar Pathan azarpathan888@gmail.com