An IMM-BP based Algorithm for Tracking Maneuvering Underwater Targets by Multistatic Marine Robot Networks

• Published in: 2022 WRC Symposium on Advanced Robotics and Automation (WRC SARA) pp. 189 - 196
• Main Authors: Zheng, Feng, Tian, Yu, Zhan, Weicong, Yu, Jiancheng, Liu, Kaizhou
• Format: Conference Proceeding
• Language: English
• Published: IEEE 20.08.2022
• Subjects: Autonomous Marine Robot; Belief Propagation; Feature extraction; Interacting Multiple Model; Maintenance engineering; Multisensor-Multitarget Tracking; Prediction algorithms; Predictive models; Robotic Multistatic Sonar Network; Sonar measurements; Surveillance; Target tracking
• DOI: 10.1109/WRCSARA57040.2022.9903972

A mobile and distributed multistatic active sonar network formed by a collection of autonomous marine robots is a significant type of tool for underwater surveillance applications such as antisubmarine warfare. Developing a tracker that fuses the sonar data collected by the robots to accurately estimate the locations and velocities of many underwater maneuvering targets is a crucial issue for realizing such a robotic surveillance network. This paper investigates this issue and an IMM-BP tracking strategy. The strategy adopts the interacting multiple model (IMM) method to cope with the problem of accurate tracking of highly maneuvering targets and employs a belief propagation (BP)-based tracking method to implement the Bayesian multisensor-multitarget tracking. And a new algorithm implementing the integration of the IMM method with the BP-based tracker is developed and presented in this paper. The main features of the algorithm include the extraction of measurement-target association information and its utilization in the model probability update step, and using a track score-based strategy and predicted measurements in the track management step to improve the performance of track confirmation and maintenance. The performance of the developed algorithm is validated with numerical simulations. The results show that the developed algorithm could accurately estimate the number and states of multiple maneuvering targets in environments with clutter and conditions of missed measurements.