🚔 Two Cops

Rokey Bootcamp Turtlebot Project | 🗓️: 2025.11.10 ~ 2025.11.21
Original Repository: Rokey-C1/turtlebot4_ws & Rokey-C1/twocops_monitoring

🎬 Demo Video

Click the image above to watch the full demonstration video on YouTube.

📝 Overview

Two Cops is a multi-robot tracking project. It features two TurtleBot4 robots equipped with depth cameras (OAK-D) and LIDAR sensors that collaboratively track a moving RC car in an indoor environment.

Each TurtleBot independently performs:

• 🔍 Object detection: Using a YOLO-based model.
• 📍 3D position estimation: Calculating the RC car's position using RGB–Depth data.
• 🚀 Navigation and motion control: Powered by the ROS 2 Nav2 stack.

The system is designed with independent ROS 2 namespaces (/robot1, /robot2), allowing for scalable multi-robot deployment without topic or TF conflicts.

---

🏗️ System Architecture

💻 Hardware

• TurtleBot4 × 2
• Onboard OAK-D (RGB + Depth camera)
• Red RC-car (Target object)

⚙️ Software

• ROS2 Humble
• Nav2 (Localization + Navigation)
• YOLO-based object detection
• Custom tracking & motion node
• RViz2 for visualization

🧠 Key Concepts

• Namespaced robots (/robot<n>)
• Map-based localization (AMCL)
• Goal-based navigation (NavToPose)
• Vision-driven tracking loop

---

🚀 Get Started

📋 Prerequisites

Before running this project, ensure the following:

• ROS2 Humble is installed.
• The TurtleBot4 workspace is properly built.
• Maps and parameter files exist under: ~/turtlebot4_ws/maps/.
• The YOLO model (best.pt) is available.

---

📍 NavToPose (Manual)

This section describes manual navigation testing using Nav2.

1. Launch localization

ros2 launch turtlebot4_navigation localization.launch.py namespace:=/robot<n> map:=$HOME/turtlebot4_ws/maps/key_map.yaml params_file:=$HOME/turtlebot4_ws/maps/local2.yaml

This starts AMCL localization, the Map server, and the TF tree under /robot<n>.

2. Set initial pose

Launch RViz and use the “2D Pose Estimate” tool to align the robot with the map.

ros2 launch turtlebot4_viz view_robot.launch.py namespace:=/robot<n>

3. Launch navigation

ros2 launch turtlebot4_navigation nav2.launch.py namespace:=/robot<n> params_file:=$HOME/turtlebot4_ws/maps/nav2_net2.yaml

4. Send goal position

ros2 action send_goal /robot<n>/navigate_to_pose nav2_msgs/action/NavigateToPose "{pose: {header: {frame_id: map}, pose: {position: {x: 2.3, y: 2.3, z: 0.0}, orientation: {w: 1.0}}}}"

---

🤖 NavToPose (Script)

For convenience, the full Nav2 pipeline can be launched via a single integrated script.

1. Run preparation checker

./prep_checker.sh --n <n>

2. Run integrated launch file

ros2 launch turtlebot4_nav2pose nav2pose.launch.py \
  namespace:=/robot<n> \
  map:=$HOME/turtlebot4_ws/maps/key_map.yaml \
  params_file:=$HOME/turtlebot4_ws/maps/local2.yaml \
  nav2_params_file:=$HOME/turtlebot4_ws/maps/nav2_net2.yaml \
  initial_x:=0.09 initial_y:=0.67 initial_yaw:=1.57

---

🎯 Detection & Tracking

This runs the core perception and tracking pipeline.

ros2 launch twocops_bringup twocops_bringup.launch.py robot_id:=3 \
  initial_pose_file:=$HOME/turtlebot4_ws/config/robot3_initial_pose.txt \
  map:=$HOME/turtlebot4_ws/maps/key_map.yaml params_file:=$HOME/turtlebot4_ws/maps/local2.yaml \
  nav2_params_file:=$HOME/turtlebot4_ws/maps/nav2_net2.yaml \
  robot_id:=3 partner_robot_id:=1

Key functions of this launch:

• Performs YOLO-based object detection on RGB images.
• Computes the 3D position of the RC car using depth data.
• Converts target position into navigation or velocity commands.
• Continuously updates robot motion to follow the target.

---

👥 Multi-Robot Operation

To run two robots simultaneously:

• Use different namespaces (robot1, robot2).
• Launch all nodes separately per robot.
• Ensure each robot has its own TF tree and Nav2 stack.

Both robots can track the same RC car independently, enabling redundant perception and comparative tracking behavior.

---

⚠️ Notes & Limitations

• Tracking is currently independent per robot with no inter-robot communication.
• Target loss handling depends on YOLO detection confidence.
• Depth accuracy is affected by lighting and surface reflectivity.
• Dynamic obstacle avoidance relies on your specific Nav2 configuration.