Density trees for efficient nonlinear state estimation

Abstract

-In this paper, a new class of nonlinear Bayesian estimators based on a special space partitioning structure, generalized Octrees, is presented. This structure minimizes memory and calculation overhead. It is used as a container framework for a set of node functions that approximate a density piecewise. All necessary operations are derived in a very general way in order to allow for a great variety of Bayesian estimators. The presented estimators are especially well suited for multi-modal nonlinear estimation problems. The running time performance of the resulting estimators is first analyzed theoretically and then backed by means of simulations. All operations have a linear running time in the number of tree nodes.

A Deep Reinforcement Learning Approach for the Patrolling Problem of Water Resources Through Autonomous Surface Vehicles: The Ypacarai Lake Case

“…• Ypacaraí Lake has a total area of 60 km 2 with an average depth of 1.31 m. Different representations of the space can be used to obtain different levels of accuracy in the generated path like octrees discretization (for non-homogeneous resolution) [27] or a simple grid discretization. For the simplicity and computational efficiency, the map has been discretized in squared cells of 580 m × 580 m, which is a representative size of the sensor effective area and the variables resolution (temperature, PH, dissolved oxygen, ...).…”

Information fusion for automotive applications – An overview

“…To remedy this problem, either numerical approximation techniques are applied or further constraints that lead to an analytic solution are imposed on the models. Examples for the former are particle system-or grid-based approaches [25][26][27][28], whereas the Kalman filter [29] is the most prominent example of the latter. Automotive sensors are typically operated asynchronously and often possess different processing times.…”

Textured/textureless object recognition and pose estimation using RGB-D image

“…It is difficult to constrain the objects' localization for segmentation and clustering, as the target objects have the possibilities of being with various pose as shown in Figure 8b. Aiming to solve that, the initial background image is trained in off-line phase based on Octree data structure [40]. With the extracted foreground data, the object clusters C k will be segmented and clustered by Euclidean distance [42].…”