Multi Sensor 3D SLAM

Project Description

This project involves the use of a multi-sensor system for 3D SLAM (Simultaneous Localization and Mapping) using Agilex Scout Mini in a Gazebo simulation environment. The goal is to create a point cloud map of the environment using various sensors and compare the results based on quality metrics.

Sensors

1. 3D LiDAR:

   • 360° horizontal and 30° vertical Field of View (FOV)
   • 0.5° horizontal and vertical resolution
   • 50m range
   • 10Hz refresh rate

2. RGBD Camera:

   • 120° FOV
   • 5m depth range
   • 30Hz frame rate

Objectives

1. Use the Office Earthquake Gazebo world for simulation.
2. Create a Point Cloud Map of the environment using an algorithm such as Octomap (octomap ROS Wiki).
3. Perform SLAM using the following configurations:
   • a) Only RGBD camera-based SLAM
   • b) Only 3D LiDAR-based SLAM
   • c) Both 3D LiDAR and RGBD camera-based SLAM
4. Compare the quality of the generated 3D maps using metrics like:
   • Point cloud density
   • Peak Signal-to-Noise Ratio (PSNR)
   • Structural Similarity Index (SSIM)

---

How to Run the Simulation

Step 1: Launch the simulation

roslaunch scout_gazebo_sim scout_earthquake.launch

Step 2: Control the robot

rosrun teleop_twist_keyboard teleop_twist_keyboard.py

Step 3: Mapping using Octomap Server

For RGBD Camera:

rosrun octomap_server octomap_server_node cloud_in:=/rgbd/depth/points _frame_id:=odom _debug:=true

For 3D LiDAR:

rosrun octomap_server octomap_server_node cloud_in:=/lidar/points _frame_id:=odom _debug:=true

Step 4: Save the Point Cloud as PCD File

rosrun pcl_ros pointcloud_to_pcd input:=/octomap_point_cloud_centers _prefix:="output"

Step 5: Open the PCD File

pcl_viewer filename.pcd

---