Date of Award
Master of Engineering Science
Mechanical and Materials Engineering
Dr. Michael D. Naish
In this thesis, a novel software architecture and knowledge representation scheme is described that facilitates the combination and reconfiguration of modular sensor and actuator components, termed transducer interface modules (TIMs), to produce flexible modular sensor systems. Each TIM provides a core sensing or actuation functionality. A composite sensor is able to automatically determine its overall geometry and assume an appropriate collective identity, and if reconfigured, may then assume a different identity to match its new geometry. In current practice, a fixed combination of sensors and actuators is typically utilized, and is tailored to a specific application. Such systems cannot be cheaply or quickly reconfigured to handle a change in process requirements. Domains that may benefit from easily reconfigurable modular sensing systems include flexible inspection, mobile robotics, surveillance, and even space exploration. The software architecture is distributed, and is comprised of six layers where the implementation of each layer is encapsulated from the layer above, to which it provides service. The use of a distributed and layered architecture promotes scalability, mitigates against a single point of failure, and enables each layer to be easily implemented, modified, and debugged independently of the others. The modularization of the software architecture is further facilitated through the utilization of a pre-emptive real-time operating system, which enables the concurrent execution of the various software components specific to the architecture that implement the services provided within most of its layers. Among the layers comprising the software architecture is a virtual machine layer, which implements a lightweight, architecture-specific version of Sun Microsystems’ Java Virtual Machine that runs on top of the real-time operating system. The integration of a virtual machine enables the platform-independent template algorithms utilized at the composition layer to be written once and executed on any TIM irrespective of its underlying hardware architecture. These template algorithms are unique to this software architecture and provide intelligence to a set of heterogeneous TIMs, enabling them to collaborate and behave as a single entity termed a logical module. The evaluation of the software architecture consists of performing multiple runs of two tests in which select sensors and actuators are associated with TIMs that are then allowed to interact in order to form a logical entity. The first test evaluates the behaviour of a logical module in which the constituent TIMs interact entirely through wireless communication. The second test evaluates the behaviour of a logical module in which the constituent TIMs are physically connected in various orientations, and interact through both wireless communication as well as through their physically connected faces. In both tests, correct behaviour was exhibited. However, the performance and scalability of the architecture was somewhat restricted by the limited processing and memory resources present in the current implementation of the TIMs. The design of the software architecture facilitates easy portability between embedded platforms and scales with increasing hardware capability. Therefore, utilization of future TIM hardware variations possessing increased processing and memory resources will reduce the latencies introduced throughout the architecture and lead to tangible improvements in its performance.
Lyle, Andrew C., "A Software Architecture for Adaptive Modular Sensing Systems" (2008). Digitized Theses. 4087.