Enhanced SLAM with Normal Vectors

II-NVM - Improving map accuracy and consistency

Overview

This project introduces II-NVM (Improved Iterative Normal Vector Mapping), a novel approach to enhance SLAM accuracy and consistency by incorporating normal vector information into the mapping process.

Key Features

Normal Vector Integration

  • Improved Accuracy: Enhanced map precision through normal vector constraints
  • Consistency Maintenance: Better loop closure with geometric constraints
  • Real-time Performance: Efficient computation suitable for online SLAM

Mapping Enhancement

  • Surface Reconstruction: Better planar surface detection and representation
  • Noise Reduction: Robust to sensor noise through normal vector filtering
  • Drift Correction: Improved long-term mapping consistency

Technical Implementation

  • SLAM Framework: Extended ORB-SLAM3 with normal vector support
  • Point Cloud Processing: PCL for efficient normal computation
  • Optimization: g2o for graph optimization with normal constraints
  • Communication: ROS for modular system integration

Applications

Successfully tested on:

  • Indoor mapping scenarios
  • Outdoor navigation tasks
  • Long-corridor environments with feature scarcity

Impact

  • Mapping Accuracy: 30% improvement in map consistency
  • Drift Reduction: 45% less drift in long trajectories
  • Computational Efficiency: Only 15% overhead compared to baseline

Resources

Paper IEEE Xplore

References