superxslam/SuperOdom

What it solves

SuperOdometry provides a lightweight solution for LiDAR-inertial odometry and mapping, enabling robots to estimate their position and orientation (pose) in real-time. It specifically addresses the challenge of maintaining accuracy in "degenerate" environments where sensor data might be ambiguous, while remaining computationally efficient enough for various hardware platforms.

How it works

The system uses a fusion of LiDAR and Inertial Measurement Unit (IMU) data. The IMU odometry module provides fast pose predictions that serve as initial guesses for the LiDAR odometry's Iterative Closest Point (ICP) optimization. In return, the LiDAR odometry provides pose constraints that help the system estimate and correct the IMU's bias. The project also includes a specialized module for online degeneracy detection to identify when the environment lacks sufficient geometric features for reliable alignment.

Who it’s for

It is designed for robotics developers and researchers working with platforms equipped with Livox, Velodyne, or Ouster LiDAR sensors, including humanoid robots and other autonomous vehicles using ROS 2 Humble.

Highlights

• Flexible Sensor Support: Works in either LiDAR-only or LiDAR-inertial fusion modes.
• Degeneracy Awareness: Detects environmental degeneracy with 6 DOF uncertainty visualization.
• Dual-Mode Operation: Supports both mapping (creating a map) and localization (positioning within a map).
• Risk Prediction: Predicts alignment risks for ICP algorithms to improve robustness.
• ROS 2 Integration: Built natively for ROS 2 Humble.