Grant B. Wigley

The Development of an Operating System for Reconfigurable Computing

An operating system (OS) for reconfigurable computing uses new versions of algorithms for the allocation of area to tasks, the partitioning of an application to fit selected allocated areas and the placement and routing inside partitions. The algorithms have small deterministic and bounded run times with near linear time complexity making them suitable to run in between time slices or at the initial loading stage of applications. Tests on the prototype with benchmark examples show that it is a feasible and that fragmentation of the area of the FPGA among many users is manageable.

The first real operating system for reconfigurable computers

Traditional reconfigurable computing platforms are designed to be single user and have been acknowledged to be difficult to design applications for. The design tools are still primitive and as reconfigurable computing becomes mainstream the development of new design tools and run time environments is essential. As the number of system gates is reaching 10 million on current FPGAs, there is an increase in demand to share a single FPGA amongst multiple applications. A third party must be introduced to handle the sharing of the FPGA and we therefore introduce the first real single FPGA concurrent multi-user operating system for reconfigurable computers. In this paper we describe the complete operating system for reconfigurable architecture and the implementation details for the first limited multi-user operating system. The first OS is a loader, it allocates FPGA area and it can dynamically partition, place and route applications at run-time. As OS for reconfigurable computing is a new area of research, we also had to develop techniques for regression testing and performance comparison. This involved the development of a test suite.

Introducing ReConfigME: An Operating System for Reconfigurable Computing

ReConfigME is a complete package to manage the dynamic reconfiguration of applications running on field programmable gate arrays. ReConfigME can also be viewed as an operating system for reconfigurable computing that handles the loading of IP cores on the FPGA platform and the dynamic arrangement and rearrangement of cores on the surface of the FPGA as the execution needs of multiple applications and multiple uses sharing the same platform evolve. ReConfigME can integrate with compilers for hardware software codesigned applications. We describe all the major components that make up the operating system and give preliminary results from the first prototype. These indicate that ReConfigME is a feasible basis for software like development of reconfigurable applications.

Image fusion for uninhabited airborne vehicles

In image fusion, information from a set of images is extracted and then combined intelligently to form a new composite image with extended information content. The original data may come from different viewing conditions (bracketed focus or exposure) or various sensors (visible and infrared or a cat scan and magnetic resonance imagery). Uninhabited Airborne Vehicles (UAVs) often have visible, infrared and synthetic aperture radar imaging sensors, so image fusion is an appropriate onboard processing task for UAVs. Some forms of image fusion are computationally intensive tasks, but like many other image processing applications are naturally suited to acceleration in hardware. This potential for hardware acceleration, and the ability to reconfigure the UAV to implement new algorithms as it moves towards objects of interest make reconfigurable computing a natural route for a hardware implementation. In this paper we present what we believe is the first implementation of image fusion on a reconfigurable platform alone, and the first investigation of adaptive image fusion which makes use of dynamic reconfiguration to change the fusion algorithm as the UAV approaches an object of interest.

ReConfigME: a detailed implementation of an operating system for reconfigurable computing

Reconfigurable computing applications have traditionally had the exclusive use of the field programmable gate array, primarily because the logic densities of the available devices have been relatively similar in size compared to the application. But with the modern FPGA expanding beyond 10 million system gates, and through the use of dynamic reconfiguration, it has become feasible for several applications to share a single high density device. However, developing applications that share a device is difficult as the current design flow assumes the exclusive use of the FPGA resources. As a consequence, the designer must ensure that resources have been allocated for all possible combinations of loaded applications at design time. If the sequence of application loading and unloading is not known in advance, all resource allocation cannot be performed at design time because the availability of resources changes dynamically. In this paper, we present an implementation of an operating system that has the ability to share its FPGA resources dynamically among multiple executing applications.

A Web-Based Multiuser Operating System for Reconfiguarble Computing

Traditional reconfigurable computing platforms are designed to be used by a single user at a time, and are acknowleged to be difficult to design applications for. These factors limit the usefulness of such machines in education, where one might want to share such a machine and initially hide some of the technical difficulties so as to explore issues of greater value. We have developed a multitasking operating system to share our SPACE.2 coprocessing board among up to 8 simultaneous users. A suite of pre-configured tasks and a web based client allows novices to run reconfigurable computing applications. As users develop a knowledge of the FPGA design process they are able to make use of a more advanced PC client to build and upload their own designs. The development aims to increase access to the machine and generate interest in the further study of reconfigurable computing. We report on the design, our experience to date, and directions for further development.

Mobile hand tracking using FPGAs for low powered augmented reality

Many augmented reality systems use general purpose computing hardware to perform tasks such as rendering computer graphics, video overlay, and vision tracking. This can result in systems being large and bulky due to the hardware complexity required and the power consumed. We have developed a hand tracking solution in a reconfigurable computer, which reduces power consumption and transfers some of the processing into specialised hardware. This paper presents a summary of the design and its implementation details.

A low cost, high performance reconfigurable computing based unmanned aerial vehicle

It has previously been stated that connectivity, computational processing power, and lack of resource integration are the three major limiting factors in developing the capabilities of small low-cost autonomous unmanned aerial vehicles (UAV). In an endeavor to address and overcome these limitations, we present details on a new UAV platform consisting of a commercially available airframe, off-the-shelf reconfigurable computing hardware, and a custom built operating system which does just that. This involved firstly selecting an appropriate UAV airframe which met all of the necessary specifications including cargo size and takeoff weight. Secondly, computing hardware which can provide enough computational processing power was then selected and installed into the UAV. Thirdly, the necessary software that will manage the UAV flight controls and applications was then developed and deployed. To verify the UAV platform, it was put through a series of flight trials while performing applications such as target recognition and edge detection. The significance of this research is that we have shown that complex computational applications can be performed on small, low cost UAVs

Bushfire Hotspot Detection Through Uninhabited Aerial Vehicles and Reconfigurable Computing

Hotspots or smouldering embers left in the wake of a bushfire can, if not extinguished, reignite causing further destruction and loss of life as was the case on Eyre Peninsular in Australia in January 2005. The current method of detecting hotspots is very labour intensive, time consuming and inexact. To overcome these limitations, we propose a system that employs small uninhabited aerial vehicles (UAV) and reconfigurable computing (RC) technologies to enable fire fighting personnel to quickly and effectively locate hotspots. This paper explores the technologies proposed for the hotspot detection system including the algorithms for detecting and tracking of hotspots. It investigates the characterisation of these hotspots for autonomous detection, including data collection and testing techniques. It also describes the systems requirements as well as its components and architecture.

A data set for evaluating the performance of multi-class multi-object video tracking

One of the challenges in evaluating multi-object video detection, tracking and classification systems is having publically available data sets with which to compare different systems. However, the measures of performance for tracking and classification are different. Data sets that are suitable for evaluating tracking systems may not be appropriate for classification. Tracking video data sets typically only have ground truth track IDs, while classification video data sets only have ground truth class-label IDs. The former identifies the same object over multiple frames, while the latter identifies the type of object in individual frames. This paper describes an advancement of the ground truth meta-data for the DARPA Neovision2 Tower data set to allow both the evaluation of tracking and classification. The ground truth data sets presented in this paper contain unique object IDs across 5 different classes of object (Car, Bus, Truck, Person, Cyclist) for 24 videos of 871 image frames each. In addition to the object IDs and class labels, the ground truth data also contains the original bounding box coordinates together with new bounding boxes in instances where un-annotated objects were present. The unique IDs are maintained during occlusions between multiple objects or when objects re-enter the field of view. This will provide: a solid foundation for evaluating the performance of multi-object tracking of different types of objects, a straightforward comparison of tracking system performance using the standard Multi Object Tracking (MOT) framework, and classification performance using the Neovision2 metrics. These data have been hosted publically.