Autonav_pkg — Autonomous Navigation Demo (ROS)

Overview

This project demonstrates the autonomous navigation of a differential-drive robot in a simulated static environment using ROS (Robot Operating System).
The robot is equipped with a LiDAR sensor for real-time obstacle detection and mapping.

The system integrates several core components:

• URDF robot model (differential drive chassis + LiDAR).
• Simulated environment publisher (complex_obstacles node).
• LiDAR emulator that converts published obstacles to /scan using TF.
• Odometry node that integrates /cmd_vel into /odom and publishes TF.
• Occupancy grid map builder using /scan and /odom.
• APF (Artificial Potential Field) controller for autonomous navigation.
• Service /set_goal to set navigation targets dynamically.

Navigation and obstacle avoidance are achieved using an Artificial Potential Field (APF) controller, which generates:

• Attractive forces toward the goal.
• Repulsive forces away from obstacles.

This package provides a complete end-to-end pipeline for robotic navigation, making it ideal for education, demonstration, and as a base for future robotics research.

---

Features / Highlights

• Real-time obstacle avoidance using APF (local planner).
• Occupancy grid mapping and visualization in RViz.
• TF frame publishing for world → body.
• Simple stuck-recovery routine.
• Modular ROS nodes so components can be swapped/extended.

---

Results and Discussion

The system successfully demonstrated obstacle avoidance and goal-oriented navigation within a static simulated environment.

• The occupancy grid map provided clear visualization of free and occupied regions.
• The APF controller ensured smooth trajectory generation and real-time reactions to obstacles.
• The robot’s path and LiDAR scan data were visualized in RViz, allowing real-time monitoring.

However, as typical with APF-based approaches, challenges such as local minima can occur in more complex environments.

---

Screenshots

1. Robot Model in RViz

This image shows the differential drive robot URDF model with a mounted LiDAR sensor.

---

2. Simulated Environment with Obstacles

The static environment generated by the complex_obstacles node, containing various obstacles for the robot to navigate around.

---

3. Occupancy Grid Map

An occupancy grid map generated using LiDAR data and odometry, representing free, occupied, and unknown regions of the environment.

---

4. Robot Path Towards Goal

The robot’s trajectory as it moves toward the target location while dynamically avoiding obstacles using the APF controller.

---

5. ROS Nodes Graph (rqt_graph)

This diagram shows the ROS computation graph, displaying how different nodes and topics in the system are connected.
It provides a clear view of the interactions between components such as the LiDAR, odometry, mapping, and controller.

---

Repository Structure

autonav_pkg/
│
├─ package.xml
├─ CMakeLists.txt
├─ README.md
├─ LICENSE
├─ .gitignore
│
├─ launch/
│ └─ autonav.launch
│
├─ rviz/
│ └─ autonav.rviz
│
├─ xacro/
│ └─ differential_robot.xacro
│
├─ srv/
│ └─ SetGoal.srv
│
├─ scripts/
│ ├─ complex_obstacles.py
│ ├─ lidar.py
│ ├─ odom_node.py
│ ├─ map.py
│ ├─ path.py
│ ├─ arduino.py
│ └─ apf_controller.py
│
├─ config/
│ └─ params.yaml
│
└─ docs/
└─ screenshots/

---

Nodes

• complex_obstacles_node — publishes static obstacles as visualization_msgs/MarkerArray on /obstacles.
• lidar_from_markers_tf — listens to /obstacles + world->body TF and publishes /scan (LaserScan).
• odom_node — subscribes to /cmd_vel, integrates motion, publishes /odom and TF.
• map_builder — subscribes to /scan & /odom and publishes /map (OccupancyGrid).
• apf_controller — subscribes to /scan & /odom, exposes /set_goal service, and publishes /cmd_vel.

Messages / Services

• Service srv/goal.srv:

float64 x
float64 y
---
int32 success
string message

---

How to Run

1. Clone the repository and Make sure you have a catkin workspace and the package is placed in src/:

   git clone https://github.com/<your-username>/autonav_pkg.git

2. Build the package:

 cd ~/catkin_ws
 catkin_make
 source devel/setup.bash

3. Launch the simulation:

roslaunch autonav_pkg autonav.launch

4. Set a navigation goal:

rosservice call /set_goal "x: 2.0 y: 3.0"

---

Troubleshooting & Notes

• If RViz shows empty map: confirm /scan and /odom are publishing and TF world->body exists.
• If robot stuck in local minima: APF is a local planner — consider hybrid global planner or randomized escape behavior.
• The odometry here is simulated by integrating /cmd_vel (no encoder noise/correction).

Suggested further work

• Replace local APF with a hybrid planner (global path + local obstacle avoidance).
• Add SLAM (gmapping / Cartographer) for robust localization.
• Plug in real robot hardware by replacing the lidar_from_markers_tf and odom_node with real topics.

---

License

This project is licensed under the MIT License — see LICENSE for details.

---

Authors

• Beilassan Hdewa
• Lana Al wazzeh
• Zain Alabidin Shbani