A Highly Reliable and Cost-Efficient Multi-Sensor System for Land Vehicle Positioning.

Area of Science:

• Engineering
• Robotics
• Sensor Fusion

Background:

• Accurate land vehicle positioning is crucial for autonomous systems.
• MEMS-based inertial sensors suffer from nonlinear drift, impacting reliability.
• GPS outages present a significant challenge for continuous positioning.

Purpose of the Study:

• To develop a novel, highly reliable, and cost-efficient positioning solution for land vehicles.
• To fuse data from a MEMS-based reduced inertial sensor system (RISS), GPS, and supplementary sources.
• To compensate for positioning errors during GPS outages using a hybrid approach.

Main Methods:

• Accurate estimation of pitch and roll angles using a vehicle kinematic model.
• Elimination of MEMS sensor drift using an H∞ filter.
• Application of a distributed-dual-H∞ filtering (DDHF) mechanism for robust drift compensation.
• Integration of a generalized regression neural network (GRNN) with an auxiliary H∞ filter (AHF) for GPS outage compensation.

Main Results:

• The proposed system demonstrates accurate pitch and roll angle estimation.
• The H∞ filter effectively mitigates the negative effects of uncertain nonlinear drift in MEMS inertial sensors.
• The DDHF mechanism successfully addresses MEMS-RISS drift and utilizes supplementary sensor data.
• The hybrid GRNN-AHF methodology compensates for RISS position errors during GPS outages.

Conclusions:

• The developed positioning system offers a reliable and cost-efficient solution for land vehicles.
• The fusion of RISS, GPS, and supplementary data, coupled with advanced filtering techniques, enhances positioning accuracy.
• The system proves effective in maintaining accurate and reliable positioning even during GPS signal loss, validated through road tests.