🤖 Anibot — ROS2 Autonomous Navigation Robot

⚠️ Active development — BETA. Architecture and params may change.

Anibot is a compact ROS2 differential drive robot built around a Raspberry Pi 5, capable of autonomous navigation, real-time SLAM mapping, and sensor-fused odometry using wheel encoders and LiDAR. It runs the full Nav2 stack with a calibrated EKF for reliable localization.

---

📋 Table of Contents

• Overview
• System Architecture
• Hardware
• Software Stack
• Odometry & Sensor Fusion
• Calibrated Parameters
• Workspace Structure
• Setup & Usage
• Features & Status
• Roadmap
• Known Issues & Fixes
• Contributing
• Contact

---

🚀 Overview

Anibot implements a complete mobile robotics stack on affordable hardware:

• Autonomous navigation with Nav2 (DWB controller)
• SLAM mapping with slam_toolbox (lifelong mode)
• Sensor-fused odometry — wheel encoders + LiDAR via EKF (robot_localization)
• AMCL localization on saved maps
• Docker-based RViz for remote visualization from a laptop
• Arduino Mega motor controller with quadrature encoder tracking at 20 Hz

---

🧠 System Architecture

┌─────────────────────────────────────────────────────────┐
│                    Raspberry Pi 5                        │
│  ┌────────────┐   ┌──────────────┐   ┌───────────────┐  │
│  │ RPLIDAR A1 │──▶│  /scan topic │──▶│  slam_toolbox │  │
│  └────────────┘   └──────────────┘   └───────────────┘  │
│                                             │            │
│  ┌────────────┐   ┌──────────────┐   ┌─────▼─────────┐  │
│  │Arduino Mega│──▶│ /odom (wheel)│──▶│ robot_local.  │  │
│  │Cytron MDD3A│   └──────────────┘   │ (EKF fusion)  │  │
│  └────────────┘                      └───────┬───────┘  │
│       ▲                                      │           │
│  ┌────┴───────┐   ┌──────────────┐   ┌───────▼───────┐  │
│  │ /cmd_vel   │◀──│     Nav2     │◀──│  /odom_fused  │  │
│  └────────────┘   └──────────────┘   └───────────────┘  │
└─────────────────────────────────────────────────────────┘
                          │ SSH / ROS2 Domain 42
                ┌─────────▼─────────┐
                │  Laptop (Docker)  │
                │  RViz2 + Nav2 UI  │
                └───────────────────┘

---

🔧 Hardware

Component	Model	Notes
Compute	Raspberry Pi 5	Ubuntu 24.04, ROS2 Jazzy
Microcontroller	Arduino Mega 2560	Motor control + encoder reading
Motor Driver	Cytron MDD3A	Dual channel, PWM control
LiDAR	RPLIDAR A1M8	360° scan, ~6m range
Drive	Differential Drive	2WD with caster
Encoders	Quadrature (×4)	28 PPR per motor, gear ratio 19.2:1

Arduino Pin Mapping:

Signal	Pin
M1A (Right Motor)	9
M1B (Right Motor)	10
M2A (Left Motor)	5
M2B (Left Motor)	6
ENC_L_A	2 (INT)
ENC_L_B	3 (INT)
ENC_R_A	18 (INT)
ENC_R_B	19 (INT)

---

💾 Software Stack

Layer	Technology
OS	Ubuntu 24.04 LTS (RPi5)
Middleware	ROS2 Jazzy
Navigation	Nav2 (DWB Local Planner)
Mapping	slam_toolbox (lifelong)
Localization	AMCL + EKF (robot_localization)
Sensor Fusion	robot_localization (wheel odom + laser odom)
Visualization	RViz2 (via Docker on laptop)
Serial Comm	Python pyserial @ 115200 baud

---

🔀 Odometry & Sensor Fusion

Anibot fuses wheel odometry and laser odometry using an Extended Kalman Filter (EKF) via the robot_localization package.

/odom          (wheel encoders)  ──┐
                                    ├──▶ EKF ──▶ /odometry/filtered
/laser_odom    (LiDAR scan match) ──┘

• Wheel odom provides high-frequency velocity estimates (20 Hz) but drifts over time
• Laser odom corrects long-term drift using scan matching
• EKF output /odometry/filtered is fed to Nav2 as the primary odometry source

🧪 Fusion testing in progress — actively tuning EKF covariance matrices and validating against ground truth.

---

⚙️ Calibrated Parameters

These values are calibrated through physical testing (360° spin test, straight-line test):

# Wheel & Drive
wheel_radius:       0.041 m
wheel_base:         0.355 m     # Calibrated — 360° spin, 1.1° residual error
encoder_ppr:        28          # × 4 quadrature = 112 counts/rev
gear_ratio:         19.2

# Scale correction (motor asymmetry compensation)
left_scale:         0.82
right_scale:        1.00
swap_motors:        true        # Physical L/R wiring swap — corrected in software

# PWM limits
max_pwm:            150
pwm_min:            70          # Deadband — below this motor stalls
max_linear_speed:   0.20 m/s

# Nav2 controller
desired_linear_vel: 0.18 m/s
rotate_angular_vel: 0.42 rad/s
xy_goal_tolerance:  0.20 m
yaw_goal_tolerance: 0.25 rad

# AMCL
particles_min:      1000
particles_max:      5000
z_hit:              0.95
max_beams:          180

---

📦 Workspace Structure

anibot_ws/
├── src/
│   └── interface/          # Main ROS2 package
│       ├── interface/       # Python nodes
│       │   ├── odometery.py     # Wheel odometry node
│       │   └── ...
│       ├── launch/
│       │   ├── anibot_full.launch.py    # Nav2 full launch
│       │   └── slam_full.launch.py     # SLAM launch
│       ├── config/          # Nav2, EKF, AMCL params
│       ├── maps/            # Saved maps (.yaml + .pgm)
│       └── CMakeLists.txt
├── assets/                  # Demo GIFs, images
├── anibot_setup.sh          # One-shot dependency installer
├── docker_jazzy.sh          # RViz2 Docker launcher (laptop)
├── .gitignore
└── README.md

---

⚙️ Setup & Usage

1. Install Dependencies

chmod +x anibot_setup.sh
./anibot_setup.sh

2. Build Workspace

cd ~/anibot_ws
colcon build --symlink-install
source install/setup.bash

3. Launch SLAM (Mapping)

ros2 launch interface slam_full.launch.py

4. Save Map

ros2 run nav2_map_server map_saver_cli -f ~/maps/my_map \
  --ros-args -p save_map_timeout:=10.0

5. Launch Autonomous Navigation

ros2 launch interface anibot_full.launch.py map:=$HOME/maps/my_map.yaml

6. Teleoperation

ros2 run teleop_twist_keyboard teleop_twist_keyboard \
  --ros-args -p speed:=0.21 -p turn:=0.43

7. RViz2 on Laptop (Docker)

# On laptop — sets up ROS2 Domain 42, software OpenGL
./docker_jazzy.sh

Make sure ROS_DOMAIN_ID=42 is set on both RPi and laptop.

---

✨ Features & Status

Feature	Status
Differential drive control	✅ Working
Quadrature encoder odometry	✅ Working
RPLIDAR A1M8 integration	✅ Working
SLAM mapping (slam_toolbox)	✅ Working
Autonomous navigation (Nav2)	✅ Working
Map saving & reloading	✅ Working
Teleoperation	✅ Working
Docker RViz (laptop)	✅ Working
Wheel + Laser odom EKF fusion	🧪 Testing
IMU integration	🔜 Planned
Autonomous docking	🔜 Planned
Web monitoring dashboard	🔜 Planned

---

🗺️ Roadmap

• Basic diff drive + encoder odometry
• LiDAR scan integration
• SLAM mapping
• Nav2 autonomous navigation
• Docker RViz remote visualization
• Sensor fusion: Wheel odom + Laser odom (EKF) ← in progress
• IMU integration (MPU6050 or similar)
• Full 3-source EKF (wheel + laser + IMU)
• Dynamic obstacle avoidance tuning
• Web-based robot monitoring dashboard
• Autonomous charging dock return
• Multi-floor mapping

---

🐛 Known Issues & Fixes

Issue	Fix Applied
Left motor runs faster than right	left_scale=0.82 software compensation
Motors physically wired L/R swapped	swap_motors=True in odometry node
SLAM node crashes in ROS2 Jazzy	lifecycle_manager + bond_timeout: 0.0
Motors stall at low PWM	PWM_MIN=70 deadband enforced
Arduino resets mid-session causing odom jump	Triple reset on startup + clean encoder state
joint_state_publisher missing causing TF errors	Added to launch file
AMCL not converging	Set 2D Pose Estimate manually in RViz first
Wheel base drift in turns	Calibrated to 0.355m via 360° spin test

---

🤝 Contributing

Currently a solo project — contributions not open yet. Once the sensor fusion and IMU integration are stable, I'll open issues for community input. Feel free to fork and experiment!

---

📬 Contact

Omkarthik Dhudu

• GitHub: @Smokey8979
• LinkedIn: Omkarthik Dhudu
• Email: sdhudu@gmail.com

---

⭐ Acknowledgements

• ROS2 Community
• Nav2 Developers
• slam_toolbox
• robot_localization
• Open-source robotics ecosystem 🙏

---

Built with too much coffee and not enough sleep ☕