A recursive B-spline-based state estimation framework for 6-DoF LiDAR odometry, supporting LiDAR-only, LiDAR-inertial, and multi-LiDAR configurations.
RESPLE is a recursive B-spline-based state estimation framework for LiDAR-based odometry. It estimates 6-degree-of-freedom dynamic motions by modeling continuous trajectories using splines, serving as a backbone for various LiDAR odometry systems including LiDAR-only, LiDAR-inertial, and multi-LiDAR configurations. It solves the problem of accurate and smooth motion estimation in dynamic environments for robotic platforms.
Robotics researchers and engineers working on LiDAR-based navigation, state estimation, and odometry for autonomous vehicles, drones, legged robots, and wearable systems.
Developers choose RESPLE for its novel spline-based recursive estimation, which provides continuous motion modeling and supports a unified suite of odometry variants. Its flexibility with multiple LiDAR types and validation across diverse real-world datasets makes it a robust choice for challenging dynamic scenarios.
The first 6-DoF spline-based recursive motion esimator for LiDAR-based odometry
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Uses B-splines to recursively estimate 6-DoF motion, providing smooth trajectories as validated on high-precision datasets like HelmDyn with motion capture ground truth.
Supports LO, LIO, MLO, and MLIO configurations, tested across aerial, wheeled, legged, and wearable platforms in diverse environments per the README examples.
Works with various LiDAR types including Livox, Ouster, and Hesai, demonstrated in usage instructions for datasets like GrandTour and Newer College.
Extensively evaluated on public datasets (e.g., MCD, Newer College) and custom datasets (HelmDyn, TudoRun), ensuring robustness in dynamic scenarios.
Requires ROS2 Humble on Ubuntu 22.04, as stated in dependencies, limiting portability and increasing setup complexity for non-ROS environments.
Compilation involves recursive cloning, specific CMake flags, and package selections, which can be error-prone and time-consuming for newcomers.
README focuses on running pre-configured datasets with launch files; it lacks detailed documentation for tuning parameters or integrating new sensors.