Racing Robot ROS 2 Workspace

Field	Content
Team Name	粉壳橙芯队 (Pink Shell Orange Core)
School	安徽建筑大学 (Anhui Jianzhu University)
Captain	李明濠
RDK Model	RDK X5
Competition Mode	Fully Autonomous

---

1. System Architecture Overview

This repository is a ROS 2 Humble colcon workspace for a three-stage racing-robot competition running on an RDK X5. The system is divided into three stages, managed by a central launch entry:

competition_total.launch.py
├── Stage 1: QR scan + corridor navigation
├── Stage 2: Inertial race track (clockwise / counterclockwise)
└── Stage 3: Visual search + P-marker detection + stop

• Stage 1 (racing_stage1): Blind drive, QR code scanning, path reversal, corridor navigation.
• Stage 2 (racing_stage2): Inertial navigation along a rectangular track. QR result determines clockwise or counterclockwise direction.
• Stage 3 (racing_stage3): Return to the visual-search goal, detect the P marker, and stop using marker-region depth.

Full production behavior, state transitions, coordinate conventions, and safety rules are documented in RACING_OVERVIEW.md.

---

2. Hardware Selection & Connections

Component	Model / Description
Main Board	RDK X5 (Horizon Robotics)
Lidar	LSLIDAR N10
Camera	Depth camera
Chassis	Origincar base with differential drive
Voice Module	Voice broadcast module

Connection overview:

• Lidar: USB serial → /scan topic
• IMU: I2C → /imu/data topic
• Camera: USB Video → vision processing pipeline
• Voice: Serial → audio feedback during competition

---

3. Software System Design

3.1 Framework

• ROS 2 Humble with colcon build system
• Python nodes for all competition logic

3.2 Key Principles

• Odometry (/odom) provides XY position and distance only.
• IMU (/imu/data) is the sole source for yaw heading and turn calculations.
• /cmd_vel is the final base command, published by the Stage 1 supervisor.
• Stage 2 and Stage 3 publish candidate commands to /stage2_cmd_vel and /stage3_cmd_vel respectively.

3.3 Core Topics

Topic	Role
/cmd_vel	Final base command
/stage2_cmd_vel, /stage3_cmd_vel	Candidate commands from later stages
/imu/data	Yaw source for all navigation
/odom_combined	XY position and distance source
/scan	Lidar input for obstacle avoidance
competition_phase	Phase signal: 1 -> 2 -> 3
competition_qr_task	QR direction: clockwise or counterclockwise

---

4. Key Task Implementation Strategies

4.1 QR Code Scanning (Stage 1)

• Blind drive forward until QR is detected.
• Parse QR content to determine direction (clockwise / counterclockwise).
• Reverse along recorded path, then navigate the corridor to the Stage 2 entry point.

4.2 Inertial Navigation (Stage 2)

• Use IMU yaw for straight-line driving and 90-degree turns.
• Waypoints defined in field_track_*.yaml (CW / CCW).
• Segment completion based on distance to target end point.
• Local obstacle avoidance via lidar (/scan) during straight segments.

4.3 Visual Search & Stop (Stage 3)

• Return to the visual-search goal area.
• Detect the P marker using vision model.
• Stop based on depth estimation from the P-marker bounding box region.

4.4 Voice Broadcast

• QR code and image-to-text reading.
• Chinese TTS support.

---

5. Competition Rule Adaptation

Rule Requirement	Implementation
QR task determines direction	competition_qr_task → parse_direction() → selects CW/CCW YAML
Fully autonomous operation	No remote control; all stages run via launch pipeline
Stage transition	competition_phase topic coordinates handover
Obstacle avoidance	Lidar-based local avoidance during straight segments
Return to start	Stage 3 visual search + P-marker detection

---

6. Build & Launch

source /opt/ros/humble/setup.bash
colcon build --symlink-install --packages-select racing_bringup
source install/setup.bash
ros2 launch racing_bringup competition_total.launch.py

Stage-specific launch commands:

ros2 launch racing_stage1 competition_stage1.launch.py
ros2 launch racing_stage2 competition_stage2.launch.py
ros2 launch racing_stage3 competition_stage3.launch.py
ros2 launch origincar_base origincar_bringup.launch.py