Behaviour and learning in completely autonomous mobile robots

A number of Intelligent Mobile Robots have been developed at the University of Reading. They are completely autonomous in that no umbilical cord attaches to them to extra power supplies or computer station: further, they are not radio controlled. In this paper, the robots are discussed, in their various forms, and the individual behaviours and characteristics which appear are considered.

Approximation of non-autonomous dynamic systems by continuous time recurrent neural networks

This work provides a framework for the approximation of a dynamic system of the form x˙=f(x)+g(x)u by dynamic recurrent neural network. This extends previous work in which approximate realisation of autonomous dynamic systems was proven. Given certain conditions, the first p output neural units of a dynamic n-dimensional neural model approximate at a desired proximity a p-dimensional dynamic system with n>p. The neural architecture studied is then successfully implemented in a nonlinear multivariable system identification case study.

Towards a modular robotics standard for AGV design and programming

Across the world there are many bodies currently involved in researching into the design of autonomous guided vehicles (AGVs). One of the greatest problems at present however, is that much of the research work is being conducted in isolated groups, with the resulting AGVs sensor/control/command systems being almost completely nontransferable to other AGV designs. This paper describes a new modular method for robot design which when applied to AGVs overcomes the above problems. The method is explained here with respect to all forms of robotics but the examples have been specifically chosen to reflect typical AGV systems.

Trajectory planning of a robot using learning algorithms

The authors consider the problem of a robot manipulator operating in a noisy workspace. The manipulator is required to move from an initial position P(i) to a final position P(f). P(i) is assumed to be completely defined. However, P(f) is obtained by a sensing operation and is assumed to be fixed but unknown. The authors approach to this problem involves the use of three learning algorithms, the discretized linear reward-penalty (DLR-P) automaton, the linear reward-penalty (LR-P) automaton and a nonlinear reinforcement scheme. An automaton is placed at each joint of the robot and by acting as a decision maker, plans the trajectory based on noisy measurements of P(f).

Intact automatic imitation of human and robot actions in autism spectrum disorders

The existence of a specialized imitation module in humans is hotly debated. Studies suggesting a specific imitation impairment in individuals with autism spectrum disorders (ASD) support a modular view. However, the voluntary imitation tasks used in these studies (which require socio-cognitive abilities in addition to imitation for successful performance) cannot support claims of a specific impairment. Accordingly, an automatic imitation paradigm (a ‘cleaner’ measure of imitative ability) was used to assess the imitative ability of 16 adults with ASD and 16 non-autistic matched control participants. Participants performed a prespecified hand action in response to observed hand actions performed either by a human or a robotic hand. On compatible trials the stimulus and response actions matched, while on incompatible trials the two actions did not match. Replicating previous findings, the Control group showed an automatic imitation effect: responses on compatible trials were faster than those on incompatible trials. This effect was greater when responses were made to human than to robotic actions (‘animacy bias’). The ASD group also showed an automatic imitation effect and a larger animacy bias than the Control group. We discuss these findings with reference to the literature on imitation in ASD and theories of imitation.

Assessing the effectiveness of robot facilitated neurorehabilitation for relearning motor skills following a stroke

A growing awareness of the potential for machine-mediated neurorehabilitation has led to several novel concepts for delivering these therapies. To get from laboratory demonstrators and prototypes to the point where the concepts can be used by clinicians in practice still requires significant additional effort, not least in the requirement to assess and measure the impact of any proposed solution. To be widely accepted a study is required to use validated clinical measures but these tend to be subjective, costly to administer and may be insensitive to the effect of the treatment. Although this situation will not change, there is good reason to consider both clinical and mechanical assessments of recovery. This article outlines the problems in measuring the impact of an intervention and explores the concept of providing more mechanical assessment techniques and ultimately the possibility of combining the assessment process with aspects of the intervention.

Connectivity measures for internet topologies on the level of autonomous systems

Classical measures of network connectivity are the number of disjoint paths between a pair of nodes and the size of a minimum cut. For standard graphs, these measures can be computed efficiently using network flow techniques. However, in the Internet on the level of autonomous systems (ASs), referred to as AS-level Internet, routing policies impose restrictions on the paths that traffic can take in the network. These restrictions can be captured by the valley-free path model, which assumes a special directed graph model in which edge types represent relationships between ASs. We consider the adaptation of the classical connectivity measures to the valley-free path model, where it is -hard to compute them. Our first main contribution consists of presenting algorithms for the computation of disjoint paths, and minimum cuts, in the valley-free path model. These algorithms are useful for ASs that want to evaluate different options for selecting upstream providers to improve the robustness of their connection to the Internet. Our second main contribution is an experimental evaluation of our algorithms on four types of directed graph models of the AS-level Internet produced by different inference algorithms. Most importantly, the evaluation shows that our algorithms are able to compute optimal solutions to instances of realistic size of the connectivity problems in the valley-free path model in reasonable time. Furthermore, our experimental results provide information about the characteristics of the directed graph models of the AS-level Internet produced by different inference algorithms. It turns out that (i) we can quantify the difference between the undirected AS-level topology and the directed graph models with respect to fundamental connectivity measures, and (ii) the different inference algorithms yield topologies that are similar with respect to connectivity and are different with respect to the types of paths that exist between pairs of ASs.

Design of a biomimetic skin for an octopus-inspired robot– Part I: characterising octopus skin

Octopus skin samples were tested under quasi-static and scissor cutting conditions to measure the in-plane material properties and fracture toughness. Samples from all eight arms of one octopus were tested statically to investigate how properties vary from arm to arm. Another nine octopus skins were measured to study the influence of body mass on skin properties. Influence of specimen location on skin mechanical properties was also studied. Material properties of skin, i.e. the Young's modulus, ultimate stress, failure strain and fracture toughness have been plotted against the position of skin along the length of arm or body. Statistical studies were carried out to help analyzing experimental data obtained. Results of this work will be used as guidelines for the design and development of artificial skins for an octopus-inspired robot.

Design of a biomimetic skin for an octopus-inspired robot - Part II: Development of the skin artefact

In order to develop skin artefact for an octopus-inspired robot arm, which is designed to be able to elongate 60% of its original length, silicone rubber and knitted nylon sheet were selected to manufacture an artificial skin, due to their higher elastic strain and high flexibility. Tensile and scissors cutting tests were conducted to characterise the matrix and reinforcing materials and the skin artefact. Material properties of the individual and the composite materials were compared with the measured properties of real octopus skin presented in Part I. The Young’s modulus of the skin should be below 20 MPa and the elastic strain range should be over 60%. The fracture toughness should be at least 0.9 kJ·m−2. Tubes made of the skin artefact filled with liquid were tested to study volume change under deformation. Finite element analysis model was developed to simulate the material and arm structure under tensile loading. Results show that the skin artefact developed has similar mechanical properties as the real octopus skin and satisfies all the design specifications of the OCTOPUS robot.

Predictive computed-torque control of a PUMA 560 manipulator robot

This paper describes the integration of constrained predictive control and computed-torque control, and its application on a six degree-of-freedom PUMA 560 manipulator arm. The real-time implementation was based on SIMULINK, with the predictive controller and the computed-torque control law implemented in the C programming language. The constrained predictive controller solved a quadratic programming problem at every sampling interval, which was as short as 10 ms, using a prediction horizon of 150 steps and an 18th order state space model.

Assisted remote viewing of a teleoperation work cell

Intelligent viewing systems are required if efficient and productive teleoperation is to be applied to dynamic manufacturing environments. These systems must automatically provide remote views to an operator which assist in the completion of the task. This assistance increases the productivity of the teleoperation task if the robot controller is responsive to the unpredictable dynamic evolution of the workcell. Behavioral controllers can be utilized to give reactive 'intelligence.' The inherent complex structure of current systems, however, places considerable time overheads on any redesign of the emergent behavior. In industry, where the remote environment and task frequently change, this continual redesign process becomes inefficient. We introduce a novel behavioral controller, based on an 'ego-behavior' architecture, to command an active camera (a camera mounted on a robot) within a remote workcell. Using this ego-behavioral architecture the responses from individual behaviors are rapidly combined to produce an 'intelligent' responsive viewing system. The architecture is single-layered, each behavior being autonomous with no explicit knowledge of the number, description or activity of other behaviors present (if any). This lack of imposed structure decreases the development time as it allows each behavior to be designed and tested independently before insertion into the architecture. The fusion mechanism for the behaviors provides the ability for each behavior to compete and/or co-operate with other behaviors for full or partial control of the viewing active camera. Each behavior continually reassesses this degree of competition or co-operation by measuring its own success in controlling the active camera against pre-defined constraints. The ego-behavioral architecture is demonstrated through simulation and experimentation.

Transputer control of a Puma 560 robot via the virtual bus

The robot control problem is discussed with regard to controller implementation on a multitransputer array. Some high-performance aspects required of such controllers are described, with particular reference to robot force control. The implications for the architecture required for controllers based on computed torque are discussed and an example is described. The idea of treating a transputer array as a virtual bus is put forward for the implementation of fast real-time controllers. An example is given of controlling a Puma 560 industrial robot. Some of the practical considerations for using transputers for such control are described.