Open Collections

Multisensor fusion within an Encapsulated Logical Devices Architecture

University of British Columbia

This work is concerned with increasing the efficiency of the implementation, maintainability and operational reliability of automated workcells. These systems consist of a locally contained set of sensors and actuators that are integrated to perform a set of automated tasks. The specific contribution of this work includes the specification of the Heuristic-based Geometric Redundant Fusion (HGRF) method and discussions regarding the fusion of specific classes of complementary sensor measurements. Further, the Encapsulated Logical Device (ELD) Architecture is presented as a framework that facilitates, the systematic and efficient implementation of automated workcells. This architecture incorporates the fusion mechanisms proposed in this work. The HGRF method is an extension of the Geometric Redundant Fusion (GRF) method and is capable of fusing m redundant measurements in n-dimensional space. Unlike the GRF method, the uncertainty of the HGRF method's result increases as the level of disparity of the measurements increases. This ensures a realistic estimate of the uncertainty of the data even when unexpected sensor errors occur and when a small amount of data is available, as is common with automation workcells. The uncertainty ellipsoid representation, based on the Gaussian distribution, is utilized by this method. This limits the application. of this method to linear measurement spaces, unless a linear approximation to a non-linear measurement space is acceptable. This work also investigates the fusion of three classes of complementary sensor data. The dimensional extension function is useful for combining measurements taken of the same feature in non-overlapping linear measurement spaces. The uncertainty modification mechanism is applicable when a sensor's measurement can be used to modify the uncertainty of another sensor's measurement. The range enhancement mechanism is useful for combining different measurements taken in overlapping measurement spaces over different ranges of that space. Additionally, the projection method is a tool useful for manipulating the dimensionality of uncertain sensor data by projecting an ellipsoid into a lower dimension. The ELD Architecture allows systematic and efficient implementation of automation workcells across multiple hardware and software platforms. The ELD Architecture is based upon the Logical Sensor (LS). paradigm and includes sensor and actuator functionality. Within the ELD the uncertainty of all sensor data is quantified and manipulated using the sensor fusion mechanisms detailed in this work. Consideration of the uncertainty of sensor data will enhance the operational reliability of industrial automation workcells.

Thesis/Dissertation

application/pdf

2009-08-12

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

http://hdl.handle.net/2429/12078

Mechanical Engineering

University of British Columbia

UBCV

DSpace