← Back ICRA 2026

RESPLE: Recursive Spline Estimation for LiDAR-Based Odometry

Ziyu Cao, William Talbot, Kailai Li

PDF

AI summary

Key figure (auto-extracted from paper)

RESPLE delivers a lightweight, real-time recursive B-spline estimator that achieves accurate and robust 6-DoF motion tracking across diverse multi-sensor LiDAR setups.

LiDAR odometry B-splines recursive estimation sensor fusion 6-DoF motion SLAM

Problem

Discrete-time and existing continuous-time motion estimation methods struggle to efficiently process dense, asynchronous multi-sensor data while maintaining high fidelity for highly dynamic motions.

Approach

The authors introduce RESPLE, which embeds 6-DoF cubic B-splines into a state-space model and uses a modified iterated extended Kalman filter to recursively update position and orientation control points without error-state formulations.

Key results

First B-spline recursive framework for full 6-DoF dynamic motion estimation
Unified direct LiDAR, LiDAR-inertial, and multi-sensor odometry suite
Comparable or superior accuracy and robustness with real-time efficiency
Proven strength in highly dynamic motions and complex environments

Why it matters

Offers a computationally efficient, universal framework for reliable real-time odometry, benefiting autonomous vehicles, mobile robots, and search-and-rescue platforms.

Abstract

We present a novel recursive Bayesian estimation framework using B-splines for continuous-time 6-DoF dynamic motion estimation. The state vector consists of a recurrent set of position control points and orientation control point increments, enabling efficient estimation via a modified iterated extended Kalman filter without involving error-state formula- tions. The resulting recursive spline estimator (RESPLE) is further leveraged to develop a versatile suite of direct LiDAR- based odometry solutions, supporting the integration of one or multiple LiDARs and an IMU. We conduct extensive real-world evaluations using public datasets and our own experiments, covering diverse sensor setups, platforms, and environments. Compared to existing systems, RESPLE achieves comparable or superior estimation accuracy and robustness, while attain- ing real-time efficiency. Our results and analysis demonstrate RESPLE’s strength in handling highly dynamic motions and complex scenes within a lightweight and flexible design, showing strong potential as a universal framework for multi-sensor motion estimation. We release the source code and experimental datasets at https://github.com/ASIG-X/RESPLE.

Index terms

Sensor Fusion Range Sensing