Related Experiment Video
Updated: Feb 28, 2026
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
7.6K
A Cost-Effective Vehicle Localization Solution Using an Interacting Multiple Model-Unscented Kalman Filters (IMM-UKF)
Qimin Xu1, Xu Li2, Ching-Yao Chan3
1School of Instrument Science and Engineering, Southeast University, Nanjing 210096, China. jimmy.xqm@gmail.com.
Sensors (Basel, Switzerland)
|June 21, 2017
Summary
This study introduces an improved vehicle localization system using an Interacting Multiple Model-Unscented Kalman Filter (IMM-UKF) and Grey Neural Network (GNN). The solution effectively handles noisy sensors and Global Positioning System (GPS) signal loss for accurate positioning.
Area of Science:
- Robotics and Autonomous Systems
- Sensor Fusion
- Navigation and Positioning
Background:
- Accurate vehicle localization is critical for advanced driver-assistance systems and autonomous driving.
- Inertial sensors suffer from noise and drift, while Global Positioning System (GPS) signals are prone to outages in urban canyons or tunnels.
- Existing methods often struggle with adapting to dynamic noise characteristics and bridging GPS outages effectively.
Purpose of the Study:
- To propose a cost-effective and robust localization solution for land vehicles.
- To enhance adaptability to uncertain sensor noise and mitigate Global Positioning System (GPS) outages.
- To improve the accuracy and reliability of vehicle positioning systems.
Main Methods:
- Development of an Interacting Multiple Model-based Unscented Kalman Filter (IMM-UKF) integrating three UKFs with varying noise covariances.
- Parallel execution of two IMM-UKF systems: one fusing GPS, in-vehicle sensors, and MEMS-RISS; the other fusing only in-vehicle sensors and MEMS-RISS.
- Utilization of a Grey Neural Network (GNN) module trained on state vector differences from the two IMM-UKFs to predict and compensate for position errors during GPS outages.
Main Results:
- The proposed IMM-UKF approach demonstrated adaptability to different noise characteristics through soft switching.
- The integrated GNN module effectively predicted and compensated for position errors when GPS signals were unavailable.
- Road-test experiments confirmed the proposed solution's superior performance compared to existing methods across various driving scenarios.
Conclusions:
- The proposed IMM-UKF and GNN-based localization solution offers a cost-effective and robust approach for land vehicles.
- The system effectively addresses challenges posed by noisy inertial sensors and intermittent GPS availability.
- This method significantly enhances vehicle positioning accuracy and reliability, outperforming conventional techniques.
