Cluster Target Tracking Based on Multi-Sensor Adaptive GLMB Filter.

Area of Science:

• Engineering
• Computer Science
• Signal Processing

Background:

• Complex detection environments pose challenges for tracking cluster targets due to unknown detection probabilities and clutter rates.
• Accurate tracking is crucial for various applications, including surveillance and autonomous systems.

Purpose of the Study:

• To propose a novel multi-sensor adaptive generalized labeled multi-Bernoulli (MS-AGLMB) filter for enhanced cluster target tracking.
• To address the limitations of existing filters in handling unknown environmental parameters like detection probability and clutter rate.

Main Methods:

• A virtual leader-follower model is used to describe cluster kinematics, considering interactions among cluster members.
• An adaptive cardinalized probability hypothesis density (CPHD) filter estimates detection probability and clutter rate in real time.
• Gibbs sampling is employed for efficient truncation of GLMB association hypotheses and multi-sensor measurement partitioning.

Main Results:

• The proposed MS-AGLMB filter enables joint estimation of target trajectories, detection probability, clutter rate, and cluster structure.
• Simulations show the algorithm's superior robustness in scenarios with time-varying detection probability and clutter rate.
• The MS-AGLMB filter outperforms existing methods in cluster target tracking accuracy and reliability.

Conclusions:

• The developed MS-AGLMB filter provides a robust solution for cluster target tracking in challenging environments.
• Real-time estimation of environmental parameters significantly improves tracking performance.
• This work advances the state-of-the-art in multi-target tracking with adaptive filtering techniques.