3D Spatial Mapping of Sparse Featured Environments

Overview

Built an autonomous robot mapping system using RTAB-Map SLAM with ZED Mini stereo camera in ROS2. The system creates drift-free maps in challenging GPS-denied environments.

The Challenge

Testing in Northeastern University's tunnels revealed that mounting the camera on a rolling chair improved stability and accuracy compared to handheld use.

The Solution

Integrated stereo visual odometry with IMU data using Kalman filtering. Implemented Bayesian loop closure with GTSAM optimization to eliminate accumulated drift.

Key Results

• Precise alignment with ground truth maps
• Real-time 3D indoor mapping capabilities
• Successfully mapped Northeastern University underground tunnels
• Cost-effective solution for indoor mapping applications