Exploiting laser and capacitive ranging sensors' behaviour to identify material type

Publication Type:
Issue Date:
Filename Description Size
Thumbnail01Front.pdfcontents and abstract7.83 MB
Adobe PDF
Thumbnail02Whole.pdfthesis98.17 MB
Adobe PDF
Full metadata record
NO FULL TEXT AVAILABLE. Access is restricted indefinitely. ----- This thesis presents the Laser and Capacitive based Material type Identification (LaCMI) system for material type identification1. The LaCMI is a combination of two base sensing technology approaches to achieve material type identification. The overall research objective of this work was to devise a sensing system capable of identifying several physical properties (of an object in the sensing field) to use as the basis for high confidence identification of the material types of the objects. Laser and capacitive based technologies were used in the system due to the advantages of using technologies that interact with different physical properties of the sensed object. In the laser range finder based approach, non-dimensional encapsulations of an object’s surface reflectance and roughness were created via a model-free method which analysed polynomial fits and fit errors of a laser range finder’s readings. As the basis of the capacitive based approach, a surface penetrating sensing technology capable of delivering range and material type identifications in environments containing air heavily laden with particles (both conductive and non-conductive) was devised and implemented. Then, using a multi-frequency approach to the capacitive sensing, model- free algorithms that exploit the variations in the sensor’s behaviour (absolute readings and intra-readings variations) when sensing materials of different types were used to create non-dimensional indictors of the sensed object’s permittivity and conductivity. From the fusion of these two complementary sensing technologies, a novel sensing system for non-contact, robust and high-confidence material type identification of sensed objects in real-world environments was implemented and empirically evaluated. Extensive experimental evidence is presented to demonstrate the potential, limitations and usefulness of the two base sensing technologies and of a system consisting of the two. It is shown that the system is capable of delivering reliable, robust and accurate (over 94% correct) material type identifications of various materials present in a scene where a steel bridge is being prepared for grit blasting. The ability to provide better than 89% correct material type identifications in the presence of air heavily laden with particles (both conductive and non-conductive) is demonstrated. The results also show that the system can produce material type identifications with over 99.9% confidence in real-world environments. Further results show the suitability of the system to be used in more general applications where material type identification with three-dimensional geometric registration is required.
Please use this identifier to cite or link to this item: