The work reported in this paper is motivated by the need for developing swarm pattern transformation methodologies. Two methods, namely a macroscopic method and a mathematical method are investigated for pattern transformation. The first method is based on macroscopic parameters while the second method is based on both microscopic and macroscopic parameters. A formal definition to pattern transformation considering four special cases of transformation is presented. Simulations on a physics simulation engine are used to confirm the feasibility of the proposed transformation methods. A brief comparison between the two methods is also presented.

Current and planned robotic rovers for space exploration are focused on science and correspondingly carry a science payload. Future missions will need robotic rovers that can demonstrate a wider range of functionality. This paper proposes an approach to offering this greater functionality by employing science and/or tool packs aboard a highly mobile robotic chassis. The packs are interchangeable and each contains different instruments or tools. The appropriate selection of science and/or tool packs enables the robot to perform a great variety of tasks either alone or in cooperation with other robots. The multi-tasking rover (MTR), thus conceived, provides a novel method for high return on investment. This paper describes the mobility system of the MTR and reports on initial experimental evaluation of the robotic chassis

In this paper we present the networked robotics approach to dynamic architecture creation. Building on our prior work we highlight the ease at which system and architecture creation can be moved from the single robot domain to the cooperative/multiple robotic domain; indeed under the networked robotic framework there are no differences between the two, as a multiple, cooperative, robotic architecture simply emerges from a richer network environment (the module pool). Essentially task-driven architectures are instantiated on an as needed basis, allowing conceptualised designs to be run wherever a suitable framework (i.e. a module pool) exists. In this paper we demonstrate this framework applied in the creation of a multi robot system, highlighting the flexibility and robustness of the approach, and the experimental freedom inherent within it. The approach has been tested experimentally.

In this paper we report on the design, implementation and testing of a manipulator for a multi-robot task involving the transport of an extended payload by two robots. The overall task involves three main manipulation components, namely payload pickup, terrain traversal and putdown at the deployment site. The manipulator was specifically designed to meet the requirements of these manipulation operations. It comprises three main components, a two-degree of freedom arm, a passive but compliant one-degree of freedom wrist, and an active gripper mechanism. Low-level control is accomplished using a small network of PIC microcontrollers. A key design goal of the control architecture was software and hardware modularity: separate modules drive the motors, measure encoder counts and support a range of low-level behaviors suitable for the task. Several sensors were specifically designed to enhance the compliance of the system. High-level control is based on an on-board laptop computer. The paper describes the motivation, the design and implementation, and the experimental results to date.

The visual acts theory aims to provide intelligent assistance for camera viewpoint selection during teleoperation. It combines a top-down model to partition a task into a sequence of operations and a bottom up component that monitors the operator in order to select task-relevant camera viewpoints in a timely manner for the operator. In a previous paper the authors (2001) presented results showing that visual acts does indeed provide a camera view of sufficient quality to allow an operator to complete a task. In this paper we present further results that bear on the performance of visual acts. The results provide further insights into the provision of operator controls for teleoperation and the role of additional cues, such as shadows, in task completion.

This paper reports on the development of an online environment supporting a network-based robotics educational project. The project was undertaken by a large body of students and involved adding a level of intelligence to control a toy device that was accessible remotely over the network. The students were to design and implement a stop-look-act control strategy for a simple pick-up task. The paper describes the educational context of the project, the task the students were required to undertake, the hardware and software environment used to support the project, and the performance of the environment over the duration of the assignment. The results contribute to the understanding of the infrastructure support required for Internet-based robotics projects and online robot demonstrations.

In this paper we present research aimed at exploring the configuration of robot architecture from robot components that are distributed about a network environment. The research is motivated by previous work focusing on representation and reasoning methods for modeling modular robotics systems. We look specifically at self-configuring networked robotic agents. The proposed model combines the notions of a task factory that generate task modules and a pool of modules representing robotics components such as sensors and effectors. A number of successively more complex scenarios are used to illustrate the concepts. These have been implemented and tested experimentally. The paper describes the experimental implementation.

This paper describes the Visual Acts theory and architecture for automated camera placement during teleoperation. The theory is motivated by the goal to provide operators with task relevant information in a timely manner. The Visual Acts architecture combines top-down deliberative task modelling with bottom-up reactive observation of an operator to select camera views which provide the operator with task-relevant information. The paper describes the reactive component of the architecture, its implementation in an agent-based architecture and experimental results that confirm the validity of the approach.

Modular robotics systems (MRS) model robotics resources as distinct functional modules which are combined to perform tasks. MRS are flexible in both functionality and control, allowing a range of modules and control architectures to be used without modification of modules. This flexibility must be supported by a modelling strategy which allows specification of a module's functionality, and a well-defined mechanism for combining modules. This paper presents a sensor-oriented functional model of modular robotics systems. We describe our approach to modelling robotics resources, which uses semantic annotations to enhance the specification of functionality. A mechanism which successfully detects and resolves consequences of module combination is described, and results from our implementation of this mechanism presented. We outline the application of this work to the automatic configuration of varying control architectures without modification of modules or additional burden on the operator.

This paper reports on research currently underway at the University of Reading aimed at developing a resource management model for providing laboratory resources as shared facilities on the Internet. The model is based on the concept of a resource allocator which maintains details of local, "registered" resources. This allocator, in turn, provides the mechanism for client applications to request and gain access to these resources: the resource allocator is publicly accessible via the Internet. We describe the model and its implementation. The model we propose can form the basis for resource sharing, providing a model for many laboratories to place laboratory facilities on the Internet. The work has implications for modular and cooperative robotics systems, where robots are configured into cooperative groups, and for human-robot systems in educations, research and industry.

In this paper we present research which is exploring the use of perceptual models in the construction of automated viewing systems for operator viewing during teleoperation. We show how a perceptual discrimination model can be used for selecting a set of viewing directions for one phase of the peg-in-hole task and hence a pattern of looking, or "visual act". We introduce "visibility modelling" as a way of mapping from the task environment to the discrimination model.

By automating viewing during teleoperation the operator can be released from the need to manually set camera viewing parameters, which distracts from the central component of the task, robotic manipulation. The challenge of automation is to manage the duality of automatic-control versus operator-in-control. Specifically, the operator needs to 'feel' in control, since otherwise confidence in the system, and hence performance, will be poor. We propose an approach that combines deliberative task models, models of human perception and reactive architectures. We use the peg-in-hole task as a case study and discuss the mapping of viewing requirements onto the camera resources defining the remote viewing facilities.

We propose a sensor-based approach to modelling robotic systems. In this approach a robot is modelled as a colony of sensors between which constraints have been imposed. When all constraint are off, each member of the colony behaves as an autonomous agent. A graph structure in which nodes represent sensors and edges represent constraints between sensors, is used to represent a robot modelled in this way. A colony of robots is represented as a set of these graphs, and corresponds to a particular partition of the sensor colony. A different partition will correspond to a different robot colony. We distinguish between physical and virtual robotic systems and propose the latter as an approach to implementing adaptable robotic systems.