A Positioning and Navigation Method Combining Multimotion Features Dead Reckoning with Acoustic Localization.

Area of Science:

• Navigation Systems
• Sensor Fusion
• Geomatics Engineering

Background:

• Accurate location information is crucial for commercial and social applications.
• Acoustic positioning offers high accuracy but suffers from anomalies.
• Inertia-assisted positioning (IAP) provides autonomous capabilities but accumulates errors.

Purpose of the Study:

• To develop a novel positioning navigation system integrating acoustic estimation and pedestrian dead reckoning (PDR).
• To address limitations of individual positioning methods, namely anomalies in acoustic systems and error accumulation in PDR.
• To enhance localization accuracy and robustness through sensor fusion and a novel step-length model.

Main Methods:

• Feature extraction from motion data (acceleration peak-to-valley amplitude difference, walk frequency, variance, mean, peak/valley medians).
• Step length prediction using previous steps and current acceleration data, optimized with LASSO regularization.
• Acoustic estimation via a hybrid CHAN-Taylor algorithm.
• Location determination using an extended Kalman filter (EKF) fusing improved PDR and acoustic estimations.

Main Results:

• Achieved localization accuracy ranging from 8 to 56.28 cm in comparative experiments across two scenarios and heterogeneous devices.
• Demonstrated significant mitigation of cumulative errors inherent in PDR.
• Showcased high-robustness localization performance under varying experimental conditions.

Conclusions:

• The proposed fusion positioning navigation system effectively integrates acoustic and PDR methods.
• The novel step-length model and fusion strategy enhance accuracy and overcome individual system limitations.
• The method offers a robust and accurate solution for location-based services, particularly in challenging environments.