Visual Sonar ROS v2.0

Advanced Mobile Robot Navigation Using Omnidirectional Vision and Virtual Sonar

Author: Mohammad Hossein Bamorovat Abadi
Email: m.bamorovvat@gmail.com
Website: https://bamorovat.com
Project Page: https://bamorovat.com/projects/visual-sonar.html
Version: 2.0.0

---

Overview

Visual Sonar ROS is a sophisticated navigation system for mobile robots that uses omnidirectional vision to simulate sonar-like obstacle detection. Instead of relying on physical sonar sensors, this system processes camera images with advanced computer vision algorithms to detect obstacles and compute safe navigation paths.

Scientific Background

This implementation is based on peer-reviewed research published in multiple IEEE conferences:

1. Bamorovat Abadi, M.H., Asghari Oskoei, M. "Effects of Mirrors in Mobile Robot Navigation Based on Omnidirectional Vision." 8th International Conference, ICIRA 2015, Portsmouth, UK, Springer International Publishing, 2015. [PDF] [BibTex]

2. Bamorovat Abadi, M.H., Asghari Oskoei, M., Fakharian, A. "Mobile robot navigation using sonar vision algorithm applied to omnidirectional vision." AI & Robotics (IRANOPEN), 2015, IEEE, pp. 1-6, 2015. [PDF] [BibTex]

3. Bamorovat Abadi, M.H., Asghari Oskoei, M., Fakharian, A. "Side Sonar Vision Applied to Omni-directional Images to Navigate Mobile Robots." 16th Conference on Fuzzy Systems and 14th Conference on Intelligent Systems, IEEE, 2017. [PDF] [BibTex]

Demo Video

Watch our robot in action: Robot Navigation Demo

---

Key Features (New in v2.0)

Multiple Edge Detection Algorithms

• Sobel - Robust general-purpose edge detection
• Canny - Precise edges with excellent noise handling
• Laplacian - Fast detection for high-contrast scenarios
• Scharr - Improved rotation invariance over Sobel
• Roberts - Simple and fast for basic edge detection
• Prewitt - Alternative gradient operator

Advanced Processing Pipeline

• Light Reflection Elimination - HSV-based analysis to reduce false positives
• Virtual Sonar Beams - Configurable number and positioning (default: 24 beams)
• Graduated Speed Control - Dynamic velocity adjustment based on obstacle proximity
• Safety Mechanisms - Emergency stop and velocity limiting

Modern Architecture

• Object-Oriented Design - Clean, maintainable C++14 code
• Namespace Organization - Proper code organization and conflict prevention
• Parameter Server Integration - Runtime configuration via ROS parameters
• Comprehensive Error Handling - Robust operation with graceful degradation
• Performance Monitoring - Real-time statistics and diagnostics

Professional Features

• Real-time Visualization - Debug images with navigation vectors and statistics
• Multiple Configuration Scenarios - Optimized presets for indoor/outdoor/low-light conditions
• Thread-Safe Operations - Proper concurrent processing
• Memory Efficient - Optimized resource usage with RAII principles

---

Requirements

System Requirements

• Operating System: Ubuntu 16.04+ (tested on 16.04, 18.04, 20.04)
• ROS Versions: Kinetic, Melodic, or Noetic
• Compiler: GCC 5.4+ with C++14 support
• Memory: Minimum 2GB RAM (4GB+ recommended)
• Storage: ~100MB for package installation

Dependencies

• ROS Core Packages:

  • roscpp - ROS C++ client library
  • sensor_msgs - Standard sensor message definitions
  • geometry_msgs - Geometry-related message definitions
  • cv_bridge - OpenCV-ROS image conversion
  • image_transport - Efficient image publishing/subscribing

• Computer Vision:

  • OpenCV 3.0+ - Image processing and computer vision
  • OpenCV Contrib (optional) - Additional algorithms

• Build System:

  • CMake 3.0.2+ - Modern build system
  • catkin - ROS build tools

---

Installation

Important

Ensure you have Ubuntu 16.04+ and at least 4GB RAM for optimal performance.

1. Prerequisites

Warning

ROS installation requires sudo privileges and internet connection. This process may take 30-60 minutes.

Install ROS (if not already installed):

# For Ubuntu 18.04 (ROS Melodic)
sudo sh -c 'echo "deb http://packages.ros.org/ros/ubuntu $(lsb_release -sc) main" > /etc/apt/sources.list.d/ros-latest.list'
sudo apt-key adv --keyserver 'hkp://keyserver.ubuntu.com:80' --recv-key C1CF6E31E6BADE8868B172B4F42ED6FBAB17C654
sudo apt update
sudo apt install ros-melodic-desktop-full

Initialize ROS Environment:

echo "source /opt/ros/melodic/setup.bash" >> ~/.bashrc
source ~/.bashrc
sudo rosdep init
rosdep update

2. Create Catkin Workspace

mkdir -p ~/catkin_ws/src
cd ~/catkin_ws/
catkin_make
echo "source ~/catkin_ws/devel/setup.bash" >> ~/.bashrc
source ~/.bashrc

3. Clone and Build Visual Sonar ROS

Tip

Use git clone --depth 1 for faster download if you don't need the full commit history.

cd ~/catkin_ws/src
git clone https://github.com/Bamorovat/VisualSonarRos.git visual_sonar_ros
cd ~/catkin_ws
catkin_make

Note

Build time typically takes 2-5 minutes depending on your system specifications.

4. Install Additional Dependencies

# Install OpenCV (if not included with ROS)
sudo apt install libopencv-dev libopencv-contrib-dev

# Install other dependencies
rosdep install --anonymous --from-paths src --ignore-src --rosdistro melodic -y

---

🎮 Usage

Important

Always ensure a camera is connected and publishing images before starting the Visual Sonar node.

Basic Operation

1. Start ROS Core:

roscore

2. Launch Camera Node (Example with USB camera):

Tip

Test your camera first with lsusb and ls /dev/video* to verify it's detected.

# Install usb_cam if needed
sudo apt install ros-melodic-usb-cam

# Launch camera node
roslaunch usb_cam usb_cam-test.launch

3. Run Visual Sonar Node:

rosrun visual_sonar_ros visual_sonar_ros_node

Note

The node will automatically start processing images and publishing navigation commands.

Advanced Usage with Parameters

Tip

Use rosparam list to see all available parameters and rosparam get /visual_sonar/param_name to check current values.

Launch with custom parameters:

rosrun visual_sonar_ros visual_sonar_ros_node \
  _num_sonars:=24 \
  _first_sonar:=10 \
  _last_sonar:=27 \
  _start_point:=60 \
  _edge_algorithm:=canny \
  _debug_mode:=true

Warning

Debug mode significantly increases CPU usage and should only be enabled for development.

---

🔧 Configuration

Core Parameters

Parameter	Type	Default	Description
num_sonars	int	24	Number of virtual sonar beams
first_sonar	int	10	Index of first active sonar
last_sonar	int	27	Index of last active sonar
start_point	int	60	Starting radius for sonar beams
max_range	double	300.0	Maximum sonar detection range
edge_algorithm	string	"canny"	Edge detection algorithm
debug_mode	bool	false	Enable debug visualization

Edge Detection Algorithms

Tip

Start with canny for general use, then switch to laplacian for better performance or sobel for noisy environments.

Algorithm	Best For	Parameters
canny	General use, precise edges	low_threshold: 50, high_threshold: 150
sobel	Robust, noise-tolerant	kernel_size: 3, scale: 1.0
laplacian	High contrast scenarios	kernel_size: 3
scharr	Rotation invariant	scale: 1.0, delta: 0.0
roberts	Fast, simple edges	threshold: 128
prewitt	Alternative gradient	threshold: 128

Note

Algorithm performance varies with lighting conditions. Test different algorithms in your specific environment.

Navigation Parameters

Caution

Always test navigation parameters in a safe environment before deploying on actual robots.

Parameter	Type	Default	Description
max_linear_vel	double	0.5	Maximum forward velocity (m/s)
max_angular_vel	double	1.0	Maximum angular velocity (rad/s)
safety_distance	double	0.3	Minimum obstacle distance (m)
emergency_stop_distance	double	0.15	Emergency stop threshold (m)
speed_reduction_factor	double	0.8	Speed reduction near obstacles

Important

The emergency_stop_distance should always be less than safety_distance for proper safety operation.

Light Reflection Parameters

Parameter	Type	Default	Description
hue_tolerance	int	15	HSV hue tolerance for reflection detection
saturation_threshold	int	30	Minimum saturation for real obstacles
value_threshold	int	50	Minimum brightness for obstacle detection
enable_hsv_analysis	bool	true	Enable HSV-based reflection filtering

---

ROS Interface

Subscribed Topics

Topic	Type	Description
/camera/image_raw	sensor_msgs/Image	Input camera images
/camera/camera_info	sensor_msgs/CameraInfo	Camera calibration data

Published Topics

Topic	Type	Description
/cmd_vel	geometry_msgs/Twist	Navigation commands
/visual_sonar/debug_image	sensor_msgs/Image	Debug visualization
/visual_sonar/sonar_data	sensor_msgs/LaserScan	Virtual sonar measurements
/visual_sonar/statistics	std_msgs/String	Performance statistics

Services

Service	Type	Description
/visual_sonar/set_algorithm	visual_sonar_ros/SetAlgorithm	Change edge detection algorithm
/visual_sonar/emergency_stop	std_srvs/Trigger	Emergency stop command
/visual_sonar/reset_navigation	std_srvs/Trigger	Reset navigation state

---

Testing and Validation

Unit Tests

# Run all tests
cd ~/catkin_ws
catkin_make run_tests visual_sonar_ros

# Run specific test categories
rostest visual_sonar_ros edge_detection_test.launch
rostest visual_sonar_ros navigation_test.launch

Troubleshooting

Tip

Enable debug mode first to visualize what the system is seeing: _debug_mode:=true

Common Issues

1. No Image Received

Warning

Check camera permissions before proceeding. USB cameras often need proper permissions.

# Check if camera topic is publishing
rostopic list | grep camera
rostopic hz /camera/image_raw

# Verify camera permissions
ls -la /dev/video*
sudo chmod 666 /dev/video0

2. High CPU Usage

Note

CPU usage above 80% may cause frame drops and degraded performance.

• Reduce image resolution in camera launch file
• Use faster edge detection algorithm (laplacian/roberts)
• Decrease number of sonar beams
• Enable debug mode only when needed

3. Navigation Oscillations

Caution

Oscillating behavior can damage robot motors. Stop the robot and adjust parameters immediately.

• Increase safety distance parameter
• Reduce maximum velocities
• Adjust speed reduction factor
• Check for light reflections causing false obstacles

4. Memory Leaks

Tip

Use htop or rosrun rqt_top rqt_top to monitor memory usage in real-time.

• Monitor with: rosrun rqt_top rqt_top
• Check debug images are properly released
• Verify OpenCV Mat objects are not accumulating

Debug Mode

Important

Debug mode creates additional ROS topics and significantly increases resource usage.

Enable comprehensive debugging:

rosrun visual_sonar_ros visual_sonar_ros_node _debug_mode:=true _log_level:=DEBUG

Debug output includes:

• Processing Statistics - Frame rate, timing, memory usage
• Algorithm Performance - Edge detection quality, reflection analysis
• Navigation Decisions - Velocity commands, obstacle distances
• Visual Overlays - Sonar beams, detected edges, navigation vectors

Tip

View debug images with: rosrun image_view image_view image:=/visual_sonar/debug_image

---

Performance Optimization

Real-time Performance Tips

1. Camera Configuration:

# Optimize camera parameters for performance
rosrun dynamic_reconfigure dynparam set /usb_cam_node image_width 320
rosrun dynamic_reconfigure dynparam set /usb_cam_node image_height 240
rosrun dynamic_reconfigure dynparam set /usb_cam_node framerate 15

2. Algorithm Selection:

• Indoor environments: Use canny or sobel
• Outdoor bright conditions: Use laplacian or roberts
• Low-light conditions: Use scharr or sobel
• High-speed navigation: Use roberts or laplacian

3. Parameter Tuning:

Note

These configurations are starting points. Fine-tune based on your specific robot and environment.

# Optimized parameters for different scenarios
indoor_config:
  edge_algorithm: "canny"
  num_sonars: 24
  max_range: 250.0
  debug_mode: false

outdoor_config:
  edge_algorithm: "laplacian" 
  num_sonars: 18
  max_range: 400.0
  enable_hsv_analysis: true

high_speed_config:
  edge_algorithm: "roberts"
  num_sonars: 16
  max_range: 300.0
  max_linear_vel: 0.8

Caution

High-speed configuration requires extensive testing and safety measures.

---

Algorithm Details

Visual Sonar Principle

The Visual Sonar algorithm converts omnidirectional images into distance measurements similar to sonar sensors:

1. Image Preprocessing: Apply edge detection to identify potential obstacles
2. Beam Simulation: Cast virtual rays from robot center at regular angular intervals
3. Distance Calculation: For each beam, find the first significant edge representing an obstacle
4. Light Reflection Filtering: Use HSV analysis to distinguish real obstacles from reflections
5. Navigation Vector Computation: Convert distance measurements to velocity commands

Mathematical Foundation

Sonar Beam Equations:

Beam_angle(i) = (2π × i) / num_sonars
Beam_x(r, i) = r × cos(Beam_angle(i))
Beam_y(r, i) = r × sin(Beam_angle(i))
Distance(i) = min(r) where Edge_detected(Beam_x(r,i), Beam_y(r,i)) = true

Navigation Vector Calculation:

Linear_velocity = max_vel × (1 - obstacle_proximity_factor)
Angular_velocity = weighted_sum(beam_distances × beam_angles) / total_weight
Emergency_stop = any(beam_distance < emergency_threshold)

Edge Detection Comparison

Sobel Operator:

• Gradient-based edge detection
• Good noise tolerance
• Computationally efficient
• Formula: G = √(Gx² + Gy²)

Canny Algorithm:

• Multi-stage edge detection
• Excellent precision with hysteresis thresholding
• More computationally intensive
• Steps: Gaussian blur → Gradient → Non-maximum suppression → Hysteresis

Laplacian Method:

• Second-derivative edge detection
• Very fast processing
• Sensitive to noise
• Formula: ∇²f = ∂²f/∂x² + ∂²f/∂y²

---

Use Cases and Applications

Autonomous Navigation

• Warehouse Robots: Navigate between shelves and avoid obstacles
• Service Robots: Indoor navigation in offices, hospitals, homes
• Agricultural Robots: Crop field navigation with row detection
• Security Patrol: Autonomous surveillance with obstacle avoidance

Research Applications

• Vision-Based SLAM: Integration with simultaneous localization and mapping
• Multi-Robot Systems: Cooperative navigation with visual sonar
• Human-Robot Interaction: Safe navigation around people
• Outdoor Exploration: Terrain mapping and obstacle detection

Educational Projects

• Computer Vision Learning: Practical implementation of edge detection algorithms
• ROS Development: Understanding message passing and node communication
• Robot Control: Real-time sensor processing and actuator control
• Algorithm Comparison: Empirical evaluation of different approaches

---

Citation

If you use Visual Sonar ROS in your research, please cite the original papers:

@inproceedings{bamorovat2015effects,
  title={Effects of Mirrors in Mobile Robot Navigation Based on Omnidirectional Vision},
  author={Bamorovat Abadi, Mohammad Hossein and Asghari Oskoei, Mohsen},
  booktitle={Intelligent Robotics and Applications: 8th International Conference, ICIRA 2015},
  pages={1--12},
  year={2015},
  publisher={Springer International Publishing},
  address={Portsmouth, UK}
}

@inproceedings{bamorovat2015mobile,
  title={Mobile robot navigation using sonar vision algorithm applied to omnidirectional vision},
  author={Bamorovat Abadi, Mohammad Hossein and Asghari Oskoei, Mohsen and Fakharian, Ahmad},
  booktitle={AI \& Robotics (IRANOPEN), 2015},
  pages={1--6},
  year={2015},
  organization={IEEE}
}

@inproceedings{bamorovat2017side,
  title={Side Sonar Vision Applied to Omni-directional Images to Navigate Mobile Robots},
  author={Bamorovat Abadi, Mohammad Hossein and Asghari Oskoei, Mohsen and Fakharian, Ahmad},
  booktitle={16th Conference on Fuzzy Systems and 14th Conference on Intelligent Systems},
  year={2017},
  organization={IEEE}
}

---

Contributing

We welcome contributions! Please see our Contributing Guidelines for details.

Development Setup

Tip

Always create a feature branch for your contributions to keep the main branch clean.

# Fork the repository and clone your fork
git clone https://github.com/YOUR_USERNAME/VisualSonarRos.git
cd VisualSonarRos

# Create development branch
git checkout -b feature/your-feature-name

# Make changes and run tests
catkin_make run_tests

# Submit pull request

Note

Please ensure all tests pass before submitting a pull request.

Code Style

• Follow Google C++ Style Guide
• Use meaningful variable names and comprehensive comments
• Include unit tests for new functionality
• Update documentation for API changes

---

License

This project is licensed under the MIT License

Support and Contact

• Author: Mohammad Hossein Bamorovat Abadi
• Email: m.bamorovvat@gmail.com
• Website: https://bamorovat.com
• Project Page: https://bamorovat.com/projects/visual-sonar.html
• Research Group: https://bamorovat.wordpress.com
• Issues: GitHub Issues
• Discussions: GitHub Discussions

---

Version History

v2.0.0 (2024)

• Major Rewrite: Complete modernization to C++14 standards
• Multi-Algorithm Support: 6 edge detection algorithms (Sobel, Canny, Laplacian, Scharr, Roberts, Prewitt)
• Enhanced Safety: Emergency stop mechanisms and graduated speed control
• Advanced Light Filtering: HSV-based reflection analysis
• Modern Architecture: Object-oriented design with namespace organization
• ROS Integration: Parameter server support and comprehensive diagnostics
• Performance Optimization: Memory efficient with real-time processing
• Critical Bug Fix: Fixed XOR power calculation error in navigation
• Complete Documentation: Comprehensive README with examples and troubleshooting

v1.0.0 (2015-2017)

• Initial Release: Basic visual sonar implementation
• Core Features: Sobel edge detection and basic navigation
• Research Foundation: Based on published IEEE papers
• ROS Support: Basic ROS node implementation

---

Star this repository if Visual Sonar ROS helps your project!

For the latest updates and research, visit bamorovat.com