A New Approach to Real-Time Odometry Calibration Using an Adaptive Particle Filter Design
Bibiana Fariña, Jonay Toledo, Leopoldo Acosta Sánchez
AI summary
Problem
Traditional offline odometry calibration fails to correct dynamic errors like wheel wear or terrain changes during operation. Existing online methods often struggle with high computational costs or cannot effectively handle the non-linear, non-Gaussian noise inherent in mobile robot sensors.
Approach
The authors develop an Adaptive Particle Filter that continuously estimates and corrects odometric parameters by dynamically adjusting particle counts and resampling frequency based on real-time sensor uncertainty and particle dispersion.
Key results
- Systematic resampling with Neff-based frequency yields optimal calibration accuracy
- Dynamic adjustment of particle count and prediction error reduces computational overhead
- Outperforms Dual Kalman Filter (DKF) in localization precision and robustness
- Validated on an autonomous wheelchair integrating wheel encoders, IMU, and LIDAR
Why it matters
Provides a computationally efficient, real-time calibration solution that enhances localization reliability for wheeled mobile robots and assistive devices in dynamic environments.
Abstract
This paper presents a novel calibration system for odometric sensors using an Adaptive Particle Filter (Adaptive- PF) to achieve high precision pose estimation and improve localization in wheeled mobile robots. The system reduces for intrinsic systematic errors in the odometric sensor by adjusting its parameters in realtime. Likewise, a comparative analysis of resampling methods —multinomial, stratified, systematic, and residual resampling— is conducted to evaluate their impact on calibration performance. The system validation is demon- strated by its implementation in an autonomous wheelchair, where the localization module integrates wheel encoders, an Inertial Measurement Unit (IMU), and a LIDAR sensor, provid- ing robust navigation in dynamic environments. Experimental results demonstrate that systematic approach and resampling based on the effective number of particles (Neff) yield the best performance. Additionally, the system dynamically adjusts prediction error based on the differences between LIDAR and odometry data. It also adapts the number of particles according to the dispersion and uncertainty, optimizing computational time without sacrificing accuracy. The proposed system outperforms another well-known method, namely the DKF (Dual Kalman Filter). Consequently, this research introduces a new Adaptive-PF for odometric parameter calibration under changing conditions.