SLAM with SC-PHD filters: an underwater vehicle application

Abstract

The random finite-set formulation for multiobject estimation provides a means of estimating the number of objects in cluttered environments with missed detections within a unified probabilistic framework. This methodology is now becoming the dominant mathematical framework within the sensor fusion community for developing multiple-target tracking algorithms. These techniques are also gaining traction in the field of feature-based simultaneous localization and mapping (SLAM) for mobile robotics. Here, we present one such instance of this approach with an underwater vehicle using a hierarchical multiobject estimation method for estimating both landmarks and vehicle position