Acoustic sensor network for relative positioning of nodes

Area of Science:

• Robotics and Autonomous Systems
• Sensor Networks
• Localization and Navigation

Background:

• Traditional localization systems often rely on external infrastructure, limiting their application in dynamic or unmapped environments.
• Developing infrastructure-free relative localization is crucial for mobile robots and autonomous agents operating in diverse settings.

Purpose of the Study:

• To analyze the accuracy of an acoustic sensor network for infrastructure-free relative localization.
• To evaluate a system capable of computing spatial relations using only acoustic emissions.

Main Methods:

• Utilized multidimensional scaling (MDS) for initial position computation.
• Applied a least-squares algorithm, specifically the Levenberg-Marquardt algorithm (LMA), for refining position estimates.
• Modeled ranging errors, including Gaussian and non-Gaussian (line-of-sight obstruction) distributions.

Main Results:

• Demonstrated that fine-grained localization is achievable with a Gaussian error model in the acoustic ranging mechanism.
• Showcased suitable position estimation accuracy even with up to 25% bias due to lost line-of-sight measurements.

Conclusions:

• The proposed acoustic sensor network offers a viable solution for relative localization without external infrastructure.
• The system exhibits robustness to line-of-sight obstructions, making it suitable for real-world applications.