169 results found
Dearden A, Demiris Y, 2007, From exploration to imitation: using learnt internal models to imitate others, AISB'07: Artificial and Ambient Intelligence, Publisher: AISB, Pages: 218-226
We present an architecture that enables asocial and social learning mechanisms to be combined in a unified framework on a robot. The robot learns two kinds of internal models by interacting with the environment with no a priori knowledge of its own motor system: internal object models are learnt about how its motor system and other objects appear in its sensor data; internal control models are learnt by babbling and represent how the robot controls objects. These asocially-learnt models of the robot’s motor system are used to understand the actions of a human demonstrator on objects that they can both interact with. Knowledge acquired through self-exploration is therefore used as a bootstrapping mechanism to understand others and benefit from their knowledge.
Dearden A, Demiris Y, Grau O, 2007, Learning models of camera control for imitation in football matches, AISB'07: Artificial and Ambient Intelligence, Publisher: AISB, Pages: 227-231
In this paper, we present ongoing work towards a system capable of learning from and imitating the movement of a trained cameraman and his director covering a football match. Useful features such as the pitch and the movement of players in the scene are detected using various computer vision techniques. In simulation, a robotic camera trains its own internal model for how it can affect these features. The movement of a real cameraman in an actual football game can be imitated by using this internal model.
Demiris Y, 2007, Prediction of intent in robotics and multi-agent systems., Cognitive Processing, Vol: 8, Pages: 151-158, ISSN: 1612-4782
Moving beyond the stimulus contained in observable agent behaviour, i.e. understanding the underlying intent of the observed agent is of immense interest in a variety of domains that involve collaborative and competitive scenarios, for example assistive robotics, computer games, robot-human interaction, decision support and intelligent tutoring. This review paper examines approaches for performing action recognition and prediction of intent from a multi-disciplinary perspective, in both single robot and multi-agent scenarios, and analyses the underlying challenges, focusing mainly on generative approaches.
Demiris Y, Billard A, 2007, Special Issue on Robot Learning by Observation, Demonstration, and Imitation, IEEE Transactions on Systems Man and Cybernetics Part B-Cybernetics, Vol: 37, Pages: 254-255, ISSN: 1083-4419
This special issue contains selected extended contributions from both the Adaptation in Artificial and Biological Systems symposium held in Hertforshire in 2006 and the wider academic community following a public call for papers in 2006. The papers presented serve as a good illustration of the challenges faced by robotics researchers today in the field of programming by observation, demonstration, and imitation.
Demiris Y K, 2007, Using Robots to study the mechanisms imitation, Neuroconstructivism: Perspectives and Prospects, Editors: Mareschal, Sirois, Westermann, Publisher: Oxford University Press, Pages: 159-178
Demiris Y K, Johnson M, 2007, Simulation Theory for Understanding Others: A Robotics Perspective, Imitation and Social Learning in Robots, Humans and Animals: Behavioural Social and Communicative Dimensions, Pages: 89-102
Johnson M, Demiris Y, 2007, Visuo-Cognitive Perspective Taking for Action Recognition, AISB'07: Artificial and Ambient Intelligence, Publisher: AISB, Pages: 262-269
Many excellent architectures exist that allow imitation of actions involving observable goals. In this paper, we develop a Simulation Theory-based architecture that uses continuous visual perspective taking to maintain a persistent model of the demonstrator's knowledge of object locations in dynamic environments; this allows an observer robot to attribute potential actions in the presence of goal occlusions, and predict the unfolding of actions through prediction of visual feedback to the demonstrator. The architecture is tested in robotic experiments, and results show that the approach also allows an observer robot to solve Theory-of-Mind tasks from the 'False Belief' paradigm.
Takács B, Butler S, Demiris Y, 2007, Multi-agent Behaviour Segmentation via Spectral Clustering, AAAI-2007 Workshop on Plan, Activity and Intention Recognition (PAIR), Publisher: AAAI, Pages: 74-81
We examine the application of spectral clustering for breaking up the behaviour of a multi-agent system in space and time into smaller, independent elements. We extend the clustering into the temporal domain and propose a novel similarity measure, which is shown to possess desirable temporal properties when clustering multi-agent behaviour. We also propose a technique to add knowledge about events of multi-agent interaction with different importance. We apply spectral clustering with this measure for analysing behaviour in a strategic game.
Tidemann A, Demiris Y, 2007, Imitating the Groove: Making Drum Machines more Human, AISB'07: Artificial and Ambient Intelligence, Publisher: AISB, Pages: 232-240
Current music production software allows rapid programming of drum patterns, but programmed patterns often lack the groove that a human drummer will provide, both in terms of being rhythmically too rigid and having no variation for longer periods of time. We have implemented an artificial software drummer that learns drum patterns by extracting user specific variations played by a human drummer. The artificial drummer then builds up a library of patterns it can use in different musical contexts. The artificial drummer models the groove and the variations of the human drummer, enhancing the realism of the produced patterns.
Chinellato E, Demiris Y, Pobil APD, 2006, Studying the human visual cortex for improving prehension capabilities in robotics., Publisher: IASTED/ACTA Press, Pages: 184-189
Chinellato E, Demiris Y, del Pobil AP, 2006, Studying the human visual cortex for achieving action-perception coordination with robots, Artificial Intelligence and Soft Computing, Editors: del Pobil, Publisher: Acta Press, Anaheim, CF, USA, Pages: 184-189
Dearden A, Demiris Y, Grau O, 2006, Tracking football player movement from a single moving camera using particle filters, European Conference on Visual Media Production (CVMP), Publisher: IET, Pages: 29-37
This paper deals with the problem of tracking football players in a football match using data from a single moving camera. Tracking footballers from a single video source is difficult: not only do the football players occlude each other, but they frequently enter and leave the camera's field of view, making initialisation and destruction of a player's tracking a difficult task. The system presented here uses particle filters to track players. The multiple state estimates used by a particle filter provide an elegant method for maintaining tracking of players following an occlusion. Automated tracking can be achieved by creating and stopping particle filters depending on the input player data.
Dearden A, Demiris Y, Grau O, 2006, Tracking football player movement from a single moving camera using particle filters, Pages: 29-37
Dearden A, Demiris Y, 2006, Active learning of probabilistic forward models in visuo-motor development, AISB'06: Adaptation in Artificial and Biological Systems, Publisher: AISB, Pages: 176-183
Forward models enable both robots and humans to predict the sensory consequences of their motor actions. To learn its own forward models a robot needs to experiment with its own motor system, in the same way that human infants need to babble as a part of their motor development. In this paper we investigate how this babbling with the motor system can be influenced by the forward models’ own knowledge of their predictive ability. By spending more time babbling in regions of motor space that require more accuracy in the forward model, the learning time can be reduced. The key to guiding this exploration is the use of probabilistic forward models, which are capable of learning and predicting not just the sensory consequence of a motor command, but also an estimate of how accurate this prediction is. An experiment was carried out to test this theory on a robotic pan tilt camera.
Demiris Y, Khadhouri B, 2006, Hierarchical attentive multiple models for execution and recognition of actions, Robotics and Autonomous Systems, Vol: 54, Pages: 361-369, ISSN: 0921-8890
According to the motor theories of perception, the motor systems of an observer are actively involved in the perception of actions when these are performed by a demonstrator. In this paper we review our computational architecture, HAMMER (Hierarchical Attentive Multiple Models for Execution and Recognition), where the motor control systems of a robot are organised in a hierarchical, distributed manner, and can be used in the dual role of (a) competitively selecting and executing an action, and (b) perceiving it when performed by a demonstrator. We subsequently demonstrate that such an arrangement can provide a principled method for the top-down control of attention during action perception, resulting in significant performance gains. We assess these performance gains under a variety of resource allocation strategies.
Demiris Y, Khadhouri B, 2006, Content-Based Control of Goal-Directed Attention During Human Action Perception, Pages: 226-231
Demiris Y, Simmons G, 2006, Perceiving the unusual: temporal properties of hierarchical motor representations for action perception, Neural Networks, Vol: 19, Pages: 272-284, ISSN: 0893-6080
Recent computational approaches to action imitation have advocated the use of hierarchical representations in the perception and imitation of demonstrated actions. Hierarchical representations present several advantages, with the main one being their ability to process information at multiple levels of detail. However, the nature of the hierarchies in these approaches has remained relatively unsophisticated, and their relation with biological evidence has not been investigated in detail, in particular with respect to the timing of movements. Following recent neuroscience work on the modulation of the premotor mirror neuron activity during the observation of unpredictable grasping movements, we present here an implementation of our HAMMER architecture using the minimum variance model for implementing reaching and grasping movements that have biologically plausible trajectories. Subsequently, we evaluate the performance of our model in matching the temporal dynamics of the modulation of cortical excitability during the passive observation of normal and unpredictable movements of human demonstrators.
Simmons G, Demiris Y, 2006, Object grasping using the minimum variance model, BIOLOGICAL CYBERNETICS, Vol: 94, Pages: 393-407, ISSN: 0340-1200
Veskos P, Demiris Y, 2006, Experimental comparison of the van der Pol and Rayleigh nonlinear oscillators for a robotic swinging task, AISB'06: Adaptation in Artificial and Biological Systems, Publisher: AISB, Pages: 197-202
In this paper, the effects of different lower-level building blocks of a robotic swinging system are explored, from the perspective of motor skill acquisition. The van der Pol and Rayleigh oscillators are used to entrain to the system’s natural dynamics, with two different network topologies being used: a symmetric and a hierarchical one. Rayleigh outperformed van der Pol regarding maximum oscillation amplitudes for every morphological configuration examined. However, van der Pol started large amplitude relaxation oscillations faster, attaining better performance during the first half of the transient period. Hence, even though there are great similarities between the oscillators, differences in their resultant behaviours are more pronounced than originally expected.
Veskos P, Demiris Y, 2006, Neuro-mechanical entrainment in a bipedal robotic walking platform, AISB'06: Adaptation in Artificial and Biological Systems, Publisher: AISB, Pages: 78-84
In this study, we investigated the use of van der Pol oscillators in a 4-dof embodied bipedal robotic platform for the purposes of planar walking. The oscillator controlled the hip and knee joints of the robot and was capable of generating waveforms with the correct frequency and phase so as to entrain with the mechanical system. Lowering its oscillation frequency resulted in an increase to the walking pace, indicating exploitation of the global natural dynamics. This is verified by its operation in absence of entrainment, where faster limb motion results in a slower overall walking pace.
Dearden A, Demiris YK, 2005, Learning forward models for robots, International Joint Conference on Artificial Intelligence (IJCAI), Publisher: International Joint Conferences on Artificial Intelligence, Pages: 1440-1445
Forward models enable a robot to predict the effects of its actions on its own motor system and its environment. This is a vital aspect of intelligent behaviour, as the robot can use predictions to decide the best set of actions to achieve a goal. The ability to learn forward models enables robots to be more adaptable and autonomous; this paper describes a system whereby they can be learnt and represented as a Bayesian network. The robot’s motor system is controlled and explored using 'motor babbling'. Feedback about its motor system comes from computer vision techniques requiring no prior information to perform tracking. The learnt forward model can be used by the robot to imitate human movement.
Demiris Y, Dearden A, 2005, From motor babbling to hierarchical learning by imitation: a robot developmental pathway, International Workshop on Epigenetic Robotics, Pages: 31-37
How does an individual use the knowledge acquired through self exploration as a manipulable model through which to understand others and benefit from their knowledge? How can developmental and social learning be combined for their mutual benefit? In this paper we review a hierarchical architecture (HAMMER) which allows a principled way for combining knowledge through exploration and knowledge from others, through the creation and use of multiple inverse and forward models. We describe how Bayesian Belief Networks can be used to learn the association between a robot’s motor commands and sensory consequences (forward models), and how the inverse association can be used for imitation. Inverse models created through self exploration, as well as those from observing others can coexist and compete in a principled unified framework, that utilises the simulation theory of mind approach to mentally rehearse and understand the actions of others.
Demiris Y, Dautenhahn K, Nehaniv C, 2005, AISB'05: Social Intelligence and Interaction in animals, robots and agents: proceedings of the 3rd international symposium on imitation in animals and artifacts, University of Hertfordshire, Hatfield, UK, 12 - 15 April 2005, Publisher: SSAISB
Johnson M, Demiris Y, 2005, Hierarchies of Coupled Inverse and Forward Models for Abstraction in Robot Action Planning, Recognition and Imitation, International Symposium on Imitation in Animals and Artifacts, Publisher: AISB, Pages: 69-76
Coupling internal inverse and forward models gives rise to on-line simulation processes that may be used as a common computational substrate for action execution, planning, recognition, imitation and learning. In this paper, multiple coupled internal inverse and forward models are arranged in a hierarchical fashion, with each level of the hierarchy interacting with other levels through top-down and bottom-up processes. Through experiments involving imitation of a human demonstrator performing object manipulation tasks, this architecture is shown to equip a robot with a multi-level motor abstraction capability. This is then used to solve the correspondence problem in action recognition. The architecture is inspired by biological evidence.
Johnson M, Demiris Y, 2005, Perceptual Perspective Taking and Action Recognition, International Journal of Advanced Robotic Systems, Vol: 2, Pages: 301-308, ISSN: 1729-8806
Robots that operate in social environments need to be able to recognise and understand the actions of other robots, and humans, in order to facilitate learning through imitation and collaboration. The success of the simulation theory approach to action recognition and imitation relies on the ability to take the perspective of other people, so as to generate simulated actions from their point of view. In this paper, simulation of visual perception is used to recreate the visual egocentric sensory space and egocentric behaviour space of an observed agent, and through this increase the accuracy of action recognition. To demonstrate the approach, experiments are performed with a robot attributing perceptions to and recognising the actions of a second robot.
Johnson MR, Demiris YK, 2005, Perspective Taking Through Simulation, Towards Autonomous Robotic Systems (TAROS), Pages: 119-126
Robots that operate among humans need to be able to attribute mental states in order to facilitate learning through imitation and collaboration. The success of the simulation theory approach for attributing mental states to another person relies on the ability to take the perspective of that person, typically by generating pretend states from that person’s point of view. In this paper, internal inverse and forward models are coupled to create simulation processes that may be used for mental state attribution: simulation of the visual process is used to attribute perceptions, and simulation of the motor control process is used to attribute potential actions. To demonstrate the approach, experiments are performed with a robot attributing perceptions and potential actions to a second robot.
Khadhouri B, Demiris Y, 2005, Attention shifts during action sequence recognition for social robots, International Conference on Advanced Robotics, Publisher: IEEE, Pages: 468-475
Human action understanding is an important component of our research towards social robots that can operate among humans. A crucial element of this component is visual attention - where should a robot direct its limited visual and computational resources during the perception of a human action? In this paper, we propose a computational model of an attention mechanism that combines the saliency of top-down elements, based on multiple hypotheses about the demonstrated action, with the saliency of bottom up components. We implement our attention mechanism on a robot, and examine its performance during the observation of object-directed human actions. Furthermore, we propose a method for resetting this model that allows it to work on multiple behaviours observed in a sequence. We also implement and investigate this method's performance on the robot.
Khadhouri B, Demiris Y, 2005, Compound effects of top-down and bottom-up influences on visual attention during action recognition, International Joint Conference on Artificial Intelligence (IJCAI), Publisher: International Joint Conferences on Artificial Intelligence, Pages: 1458-1463
The limited visual and computational resources available during the perception of a human action makes a visual attention mechanism essential. In this paper we propose an attention mechanism that combines the saliency of top-down (or goal-directed) elements, based on multiple hypotheses about the demonstrated action, with the saliency of bottom-up (or stimulus-driven) components. Furthermore, we use the bottom-up part to initialise the top-down, hence resulting in a selection of the behaviours that rightly require the limited computational resources. This attention mechanism is then combined with an action understanding model and implemented on a robot, where we examine its performance during the observation of object-directed human actions.
Khadhouri B, Demiris Y, 2005, Attention shifts during action sequence recognition for social robots, New York, 12th international conference on advanced robotics, 17 - 20 July 2005, Seattle, WA, Publisher: Ieee, Pages: 468-475
Simmons G, Demiris Y, 2005, Optimal robot arm control using the minimum variance model, Journal of Robotic Systems, Vol: 22, Pages: 677-690, ISSN: 0741-2223
Models of human movement from computational neuroscience provide a starting point for building a system that can produce flexible adaptive movement on a robot. There have been many computational models of human upper limb movement put forward, each attempting to explain one or more of the stereotypical features that characterize such movements. While these models successfully capture some of the features of human movement, they often lack a compelling biological basis for the criteria they choose to optimize. One that does provide such a basis is the minimum variance model (and its extension—task optimization in the presence of signal-dependent noise). Here, the variance of the hand position at the end of a movement is minimized, given that the control signals on the arm's actuators are subject to random noise with zero mean and variance proportional to the amplitude of the signal. Since large control signals, required to move fast, would have higher amplitude noise, the speed-accuracy trade-off emerges as a direct result of the optimization process. We chose to implement a version of this model that would be suitable for the control of a robot arm, using an optimal control scheme based on the discrete-time linear quadratic regulator. This implementation allowed us to examine the applicability of the minimum variance model to producing humanlike movement. In this paper, we describe our implementation of the minimum variance model, both for point-to-point reaching movements and for more complex trajectories involving via points. We also evaluate its performance in producing humanlike movement and show its advantages over other optimization based models (the well-known minimum jerk and minimum torque-change models) for the control of a robot arm.
This data is extracted from the Web of Science and reproduced under a licence from Thomson Reuters. You may not copy or re-distribute this data in whole or in part without the written consent of the Science business of Thomson Reuters.