Multi-Target Data Association Algorithm in Underwater BOT System with Spatial Bias and Signal Delay

Area of Science:

• Robotics and Autonomous Systems
• Signal Processing
• Underwater Acoustics

Background:

• Unmanned underwater vehicles (UUVs) face challenges in multi-target tracking due to perceptual inconsistencies.
• Accurate tracking requires addressing sensor biases, including spatial bias and signal propagation delay.
• Existing bearing-only tracking (BOT) systems often lack robust methods for these complex biases.

Purpose of the Study:

• To develop a novel measurement model for underwater BOT systems that explicitly accounts for sensor spatial bias and signal propagation delay.
• To enhance initial target state estimation using maximum likelihood estimation (MLE) and particle swarm optimization (PSO).
• To introduce an iterative multi-target data association (MDA) algorithm integrated with the expectation-maximization (EM) method for joint bias mitigation.

Main Methods:

• A new measurement model incorporating sensor spatial bias and signal propagation delay.
• Maximum Likelihood Estimation (MLE) and Particle Swarm Optimization (PSO) for initial target state estimation.
• An iterative Expectation-Maximization (EM) based Multi-Target Data Association (MDA) algorithm.

Main Results:

• The proposed MDA framework effectively mitigates the effects of signal delay and spatial bias in multi-target BOT.
• The EM-based spatial bias estimation method demonstrated accurate bias estimation capabilities.
• Monte Carlo simulations validated the overall effectiveness and accuracy of the developed tracking framework.

Conclusions:

• The integrated MDA framework offers a significant advancement in underwater multi-target tracking accuracy for UUVs.
• The explicit modeling of sensor biases and the EM-based estimation are crucial for reliable BOT systems.
• This research provides a robust solution for improving perceptual consistency and state estimation in challenging underwater environments.