BREAKERS Robot — Robotics Dojo Challenge 2025

The BREAKERS Robot was developed for the Robotics Dojo 2025 Robotics Challenge under the mentorship of Robotics Dojo and Dr. Shohei Aoki.

The challenge focused on creating robots that assist in agriculture, addressing tasks such as:

Navigation in unstructured environments
Crop monitoring
Autonomous field operations

Our solution — BREAKERS — is a fully modular ROS 2 Humble–based robotic system built for mapping, navigation, and AI-assisted perception. It operates as a distributed system across:

Raspberry Pi (robot control + sensors)
Development PC (AI, mapping, navigation)

Communication between them uses ROS 2 DDS.

Repository Branch Overview

Branch	Purpose	Link
pi (Default)	Robot-side code for Raspberry Pi 4/5 running Ubuntu 22.04 Server + ROS 2 Humble	🔗 pi branch
breakerpc	Development machine branch for AI, mapping, and navigation	🔗 breakerpc branch
ros_bridge_with_imu	Arduino firmware + ROS bridge (for Arduino UNO)	🔗 ros_bridge_with_imu branch

System Overview

Component	Platform	Description
Raspberry Pi 4/5 (Robot)	Ubuntu 22.04 Server + ROS 2 Humble	Handles low-level hardware control, sensors, and camera
PC / Laptop (Base Station)	Ubuntu 22.04 Desktop + ROS 2 Humble	Handles SLAM, navigation, AI perception, and behavior trees

Hardware Architecture

Main MCU: Arduino UNO (via diffdrive_arduino)
Microprocessor: Raspberry Pi 4 / 5
Sensors: SLLIDAR A1 • MPU6050 IMU • Wheel Encoders
Actuators: Dual DC motors with encoders • Servo (pan/tilt)
Camera: Raspberry Pi Camera Module V1 (Low FPS; RViz disabled)

Inspirations & References

This project builds upon the work of:

Final working DiffDrive: Pi branch → diffdrive_arduino ROS–Arduino Bridge with IMU: ros_bridge_with_imu branch

Note: Some IMU axes return negative readings — stable but under refinement.

Node Launch Summary

On Raspberry Pi — Robot Side

Handles hardware, movement, sensors, and camera streaming.

#	Node	Command	Purpose
1️⃣	Robot Core	`ros2 launch dojo launch_robot.launch.py`	Motor control + odometry
2️⃣	LiDAR	`ros2 run sllidar_ros2 sllidar_node`	Publishes `/scan`
3️⃣	EKF Fusion	`ros2 launch dojo ekf_launch.py`	Fuses IMU + encoders → `/odom`
4️⃣	Camera Stack	`ros2 launch dojo robot_stack_launch.py`	Publishes camera stream
5️⃣	Servo Control	`ros2 run dojo_servo servo_node`	Controls pan/tilt

On PC — Base Station / AI & Autonomy

Handles mapping, localization, navigation, and AI perception.

#	Node	Command	Description
1️⃣	Mapping (SLAM Toolbox)	`ros2 launch slam_toolbox online_async_launch.py slam_params_file:=./src/dojo/config/mapper_params_online_async.yaml use_sim_time:=false`	Builds live map
2️⃣	Teleop Keyboard	`ros2 run teleop_twist_keyboard teleop_twist_keyboard --ros-args -r /cmd_vel:=/cmd_vel_key`	Manual control
3️⃣	RViz Visualization	`rviz2 -d ./src/dojo/rviz/default.rviz`	3D visualization
4️⃣	Map Saver	`ros2 run nav2_map_server map_saver_cli -f ~/my_map123`	Saves `.pgm + .yaml`
5️⃣	Localization	`ros2 launch dojo localization_launch.py map:=/home/nathan/my_map123.yaml`	AMCL localization
6️⃣	Navigation Stack	`ros2 launch dojo navigation_launch.py use_sim_time:=false`	Nav2 autonomy
7️⃣	Color Detection	`ros2 launch image_processor colour_detector.launch.py`	Detects color markers
8️⃣	Disease Detector (ML)	`ros2 launch image_processor disease_detector.launch.py`	Classifies plant health
9️⃣	Behavior Tree (PyTree)	`python3 app.py`	Controls mission flow

Sensor Fusion & EKF Configuration

IMU (MPU6050) integrated via Arduino bridge + robot_localization EKF for drift-free odometry. Key matrices tuned to eliminate wheel skid.

📄 ekf.yaml

odom0_config: [false, false, false,
               false, false, false,
               true, true, false,
               false, false, true,
               false, false, false]

imu0: /imu_broad/imu
imu0_config: [false, false, false,
              false, false, true,
              false, false, false,
              false, false, true,
              false, false, false]

ROS 2 Control & Arduino Interface

Modified DiffDriveArduino plugin for auto-detected /dev/arduino port. Encoders moved from A4/A5 → A0/A1 to free I²C pins for MPU6050.

📄 ros2_control.xacro

<plugin>diffdrive_arduino/DiffDriveArduino</plugin>
<param name="device">/dev/arduino</param>
<param name="baud_rate">57600</param>
<param name="enc_counts_per_rev">1135</param>

Automatic USB Port Detection

LiDAR and Arduino devices auto-identified to avoid manual port edits. 📄 sllidar_a1_launch.py:

serial_port = LaunchConfiguration('serial_port', default='/dev/lidar')

Robot Parameters

Parameter	Value	Notes
Robot radius	0.15 m	Tuned for Nav2
Wheel diameter	0.0425 m	Measured
Encoder ticks/rev	1135	Accurate odometry

📁 Repository Structure

breakers_ws/
├── src/
│   ├── dojo/                     # Core control & navigation
│   ├── image_processor/          # Color & ML perception
│   ├── dojo_servo/               # Servo control nodes
│   ├── gazebo_ignition_fortress/ # Behavior tree tests
│   ├── diffdrive_arduino/        # Motor + encoder driver
│   └── sllidar_ros2/             # LiDAR driver
├── config/                       # EKF, Nav2, SLAM YAMLs
├── launch/                       # Launch files
├── rviz/                         # RViz setups
└── README.md

Setup Instructions

Development PC (Ubuntu 22.04 + ROS 2 Humble)

sudo apt update && sudo apt install -y python3-colcon-common-extensions git
mkdir -p ~/breakers_ws/src && cd ~/breakers_ws/src
git clone -b breakerpc https://github.com/Nathan-bot-design/breakers_ws
cd ~/breakers_ws
rosdep install --from-paths src --ignore-src -r -y
colcon build
source install/setup.bash

Raspberry Pi (Ubuntu 22.04 Server + ROS 2 Humble)

git clone https://github.com/Nathan-bot-design/breakers_ws
cd breakers_ws && colcon build && source install/setup.bash
ros2 launch dojo launch_robot.launch.py
ros2 run sllidar_ros2 sllidar_node
ros2 launch dojo ekf_launch.py
ros2 launch dojo robot_stack_launch.py
ros2 run dojo_servo servo_node

Network Configuration (ROS 2 DDS)

Add to ~/.bashrc on both devices:

export ROS_DOMAIN_ID=7
export RMW_IMPLEMENTATION=rmw_fastrtps_cpp

AI Models

Stored in: src/image_processor/image_processor/models/ Used by:

colour_detector.launch.py
disease_detector.launch.py

Mapping → Localization → Navigation

ros2 launch slam_toolbox online_async_launch.py \
  slam_params_file:=./src/dojo/config/mapper_params_online_async.yaml use_sim_time:=false
ros2 run nav2_map_server map_saver_cli -f ~/my_map123
ros2 launch dojo localization_launch.py map:=/home/nathan/my_map123.yaml
ros2 launch dojo navigation_launch.py use_sim_time:=false

Navigation Stack Tuning

The navigation stack in the BREAKERS Robot was heavily optimized beyond default Nav2 parameters to achieve smooth, reliable, and real-time motion on a Raspberry Pi + PC setup.

This section summarizes the key improvements and tuning decisions made during development.

1️⃣ Localization (AMCL)

Goal: Improve pose accuracy and reduce drift.

Changes:

Parameter	New	Old
α1–α5	0.05	0.2
update_rate	5.0	1.0

Effect:
AMCL now trusts wheel odometry more, producing a tighter and more stable particle cloud with minimal jitter — essential for precise mapping and navigation.

2️⃣ Controller Server (Path Following)

Goal: Achieve smoother, more natural motion and stronger obstacle avoidance.

Changes:

Parameter	New	Old
xy_goal_tolerance	0.20	0.25
yaw_goal_tolerance	0.20	0.25
PathAlign.scale	15.0	32.0
BaseObstacle.scale	0.5	0.02

Effect:
Reduced oscillations and zig-zagging. The robot now executes fluid turns and maintains stable paths across gravel, ramps, and uneven terrain.

3️⃣ Local Costmap

Goal: Optimize for real-time performance on limited hardware.

Changes:

Parameter	New	Old
resolution	0.05 m	0.02 m
robot_radius	0.18 m	0.15 m
inflation_radius	0.08 m	0.10 m
cost_scaling_factor	2.0	20.0

Effect:
CPU load reduced significantly while maintaining safe obstacle margins and smooth local pathing.

4️⃣ Global Costmap

Goal: Accelerate global planning and reduce computational overhead.

Changes:

Parameter	New	Old
resolution	0.05 m	0.02 m
cost_scaling_factor	2.5	20.0
robot_radius	0.18 m	0.15 m

Effect:
Global planner now computes paths faster and more efficiently, ideal for open or semi-structured fields.

5️⃣ Planner & Smoother Servers

Goal: Combine stable global paths with smooth local execution.

Changes:

Parameter	New	Old
planner	Dijkstra	A*
goal_tolerance	0.75	0.5
smoother.refinement_enabled	true	—
smoother.max_iterations	1000	—

Effect:
The Dijkstra planner performs better in partially mapped areas, while the smoother ensures continuous, curve-optimized motion without pausing between waypoints.

Summary of Improvements

Subsystem	Change	Result
Localization	Higher odometry trust	Low drift, stable positioning
Controller	Balanced DWB critics	Smooth obstacle-aware motion
Local Map	Coarser resolution	Lower CPU load, faster updates
Global Map	Soft inflation zones	Faster global planning
Planner	Dijkstra + smoother	Robust real-world navigation

Overall Result

These tuned parameters enabled the BREAKERS Robot to achieve:

Collision-free, power-efficient navigation
Human-like motion profiles
Reliable performance across grass, gravel, ramps, and sawdust

Autonomy — Behavior Tree–Driven Mission Control

“With guidance from Dr. Shohei Aoki, we learned that true autonomy isn’t just about navigation — it’s about decision-making.”

The BREAKERS Robot achieves full autonomy through a Behavior Tree (BT) framework app.py , integrating all subsystems — navigation, camera, LiDAR, servo control, and AI detection — into a single decision engine.

Mission Flow Overview

Stage	Task	ROS Integration	Status
1️⃣	Weeding	Navigate + Disease Detection	✅ Success
2️⃣	Terrain Traversal	Ramp → Grass → Gravel → Sawdust	⏳ In Progress
3️⃣	Loading Area	Color Detection	✅/⚠️
4️⃣	Delivery Route	Follows route based on color (“white” / “blue”)	✅ Final Stage

Core Autonomy Features:

Dynamic Task Sequencing
ROS 2 Action Integration (Nav2 navigate_to_pose)
Context Awareness via Blackboard
Servo Control Hooks for camera tilt
Failure Recovery for incomplete goals

Launching the Autonomy Node

cd ~/breakers_ws/src/gazebo_ignition_fortress/test_folder
python3 app.py

This initializes the full mission logic — subscribing to detection topics, publishing visualization markers, and autonomously executing the full workflow.

Key Dependencies

ros-humble-slam-toolbox
ros-humble-nav2-bringup
ros-humble-rviz2
ros-humble-robot-localization
ros-humble-ros2-control
sllidar_ros2
OpenCV, PyTorch

Credits

Role	Contributors
Developers	Limit Breakers Team
Mentors & References	Josh Newans, Collins Omariba
Inspired by	2025 KNIGHTS Robot
Special Thanks	Robotics Dojo, Dr. Shohei Aoki, Lenny Ng’ang’a

CAD & Robot Previews

CAD Preview

Physical Robot

Competition & Publications

Resource	Link
Robotics Dojo Competition 2025	roboticsdojo.github.io/competition2025.html
Technical Paper (PDF)	2_limit_breakers.pdf
Poster (PDF)	2_limit_breakers_poster.pdf

Name		Name	Last commit message	Last commit date
Latest commit History 26 Commits
breakerpc		breakerpc
pi/src		pi/src
ros_bridge_with_imu		ros_bridge_with_imu
.gitignore		.gitignore
README.md		README.md

Folders and files

Latest commit

History

Repository files navigation

BREAKERS Robot — Robotics Dojo Challenge 2025

Repository Branch Overview

System Overview

Hardware Architecture

Inspirations & References

Node Launch Summary

On Raspberry Pi — Robot Side

On PC — Base Station / AI & Autonomy

Sensor Fusion & EKF Configuration

ROS 2 Control & Arduino Interface

Automatic USB Port Detection

Robot Parameters

📁 Repository Structure

Setup Instructions

Development PC (Ubuntu 22.04 + ROS 2 Humble)

Raspberry Pi (Ubuntu 22.04 Server + ROS 2 Humble)

Network Configuration (ROS 2 DDS)

AI Models

Mapping → Localization → Navigation

Navigation Stack Tuning

1️⃣ Localization (AMCL)

2️⃣ Controller Server (Path Following)

3️⃣ Local Costmap

4️⃣ Global Costmap

5️⃣ Planner & Smoother Servers

Summary of Improvements

Overall Result

Autonomy — Behavior Tree–Driven Mission Control

Mission Flow Overview

Launching the Autonomy Node

Key Dependencies

Credits

CAD & Robot Previews

Competition & Publications

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages