Multiple grasp-specific representations of tool dynamics mediate skillful manipulation.

Skillful tool use requires knowledge of the dynamic properties of tools in order to specify the mapping between applied force and tool motion. Importantly, this mapping depends on the orientation of the tool in the hand. Here we investigate the representation of dynamics during skillful manipulation of a tool that can be grasped at different orientations. We ask whether the motor system uses a single general representation of dynamics for all grasp contexts or whether it uses multiple grasp-specific representations. Using a novel robotic interface, subjects rotated a virtual tool whose orientation relative to the hand could be varied. Subjects could immediately anticipate the force direction for each orientation of the tool based on its visual geometry, and, with experience, they learned to parameterize the force magnitude. Surprisingly, this parameterization of force magnitude showed limited generalization when the orientation of the tool changed. Had subjects parameterized a single general representation, full generalization would be expected. Thus, our results suggest that object dynamics are captured by multiple representations, each of which encodes the mapping associated with a specific grasp context. We suggest that the concept of grasp-specific representations may provide a unifying framework for interpreting previous results related to dynamics learning.

Optimal control predicts human performance on objects with internal degrees of freedom.

On a daily basis, humans interact with a vast range of objects and tools. A class of tasks, which can pose a serious challenge to our motor skills, are those that involve manipulating objects with internal degrees of freedom, such as when folding laundry or using a lasso. Here, we use the framework of optimal feedback control to make predictions of how humans should interact with such objects. We confirm the predictions experimentally in a two-dimensional object manipulation task, in which subjects learned to control six different objects with complex dynamics. We show that the non-intuitive behavior observed when controlling objects with internal degrees of freedom can be accounted for by a simple cost function representing a trade-off between effort and accuracy. In addition to using a simple linear, point-mass optimal control model, we also used an optimal control model, which considers the non-linear dynamics of the human arm. We find that the more realistic optimal control model captures aspects of the data that cannot be accounted for by the linear model or other previous theories of motor control. The results suggest that our everyday interactions with objects can be understood by optimality principles and advocate the use of more realistic optimal control models for the study of human motor neuroscience.

Context-dependent partitioning of motor learning in bimanual movements.

Human subjects easily adapt to single dynamic or visuomotor perturbations. In contrast, when two opposing dynamic or visuomotor perturbations are presented sequentially, interference is often observed. We examined the effect of bimanual movement context on interference between opposing perturbations using pairs of contexts, in which the relative direction of movement between the two arms was different across the pair. When each perturbation direction was associated with a different bimanual context, such as movement of the arms in the same direction versus movement in the opposite direction, interference was dramatically reduced. This occurred over a short period of training and was seen for both dynamic and visuomotor perturbations, suggesting a partitioning of motor learning for the different bimanual contexts. Further support for this was found in a series of transfer experiments. Having learned a single dynamic or visuomotor perturbation in one bimanual context, subjects showed incomplete transfer of this learning when the context changed, even though the perturbation remained the same. In addition, we examined a bimanual context in which one arm was moved passively and show that the reduction in interference requires active movement. The sensory consequences of movement are thus insufficient to allow opposing perturbations to be co-represented. Our results suggest different bimanual movement contexts engage at least partially separate representations of dynamics and kinematics in the motor system.

Composition and decomposition in bimanual dynamic learning.

Our ability to skillfully manipulate an object often involves the motor system learning to compensate for the dynamics of the object. When the two arms learn to manipulate a single object they can act cooperatively, whereas when they manipulate separate objects they control each object independently. We examined how learning transfers between these two bimanual contexts by applying force fields to the arms. In a coupled context, a single dynamic is shared between the arms, and in an uncoupled context separate dynamics are experienced independently by each arm. In a composition experiment, we found that when subjects had learned uncoupled force fields they were able to transfer to a coupled field that was the sum of the two fields. However, the contribution of each arm repartitioned over time so that, when they returned to the uncoupled fields, the error initially increased but rapidly reverted to the previous level. In a decomposition experiment, after subjects learned a coupled field, their error increased when exposed to uncoupled fields that were orthogonal components of the coupled field. However, when the coupled field was reintroduced, subjects rapidly readapted. These results suggest that the representations of dynamics for uncoupled and coupled contexts are partially independent. We found additional support for this hypothesis by showing significant learning of opposing curl fields when the context, coupled versus uncoupled, was alternated with the curl field direction. These results suggest that the motor system is able to use partially separate representations for dynamics of the two arms acting on a single object and two arms acting on separate objects.

Learning the Structure of Deep Sparse Graphical Models

Deep belief networks are a powerful way to model complex probability distributions. However, learning the structure of a belief network, particularly one with hidden units, is difficult. The Indian buffet process has been used as a nonparametric Bayesian prior on the directed structure of a belief network with a single infinitely wide hidden layer. In this paper, we introduce the cascading Indian buffet process (CIBP), which provides a nonparametric prior on the structure of a layered, directed belief network that is unbounded in both depth and width, yet allows tractable inference. We use the CIBP prior with the nonlinear Gaussian belief network so each unit can additionally vary its behavior between discrete and continuous representations. We provide Markov chain Monte Carlo algorithms for inference in these belief networks and explore the structures learned on several image data sets.

Viewing the future

As many industrial organizations have learned to apply roadmapping successfully, they have also learned that it is "roadmapping" rather than "the roadmap" that generates value. This two-part special report has focused primarily on product and technology roadmapping in industry. The first part (RTM, March-April 2003, pp. 26-59) examined the workings of the process at Lucent Technologies, Rockwell Automation, the pharmaceutical/biotechnology industry, and United Kingdom-based Domino Printing Sciences. This second part examines roadmapping in the UK, Motorola, General Motors, the services sector, and in cases that demand major investment decisions under conditions of volatility.

Effects of the user model on simulation-based learning of dialogue strategies

Over the past decade, a variety of user models have been proposed for user simulation-based reinforcement-learning of dialogue strategies. However, the strategies learned with these models are rarely evaluated in actual user trials and it remains unclear how the choice of user model affects the quality of the learned strategy. In particular, the degree to which strategies learned with a user model generalise to real user populations has not be investigated. This paper presents a series of experiments that qualitatively and quantitatively examine the effect of the user model on the learned strategy. Our results show that the performance and characteristics of the strategy are in fact highly dependent on the user model. Furthermore, a policy trained with a poor user model may appear to perform well when tested with the same model, but fail when tested with a more sophisticated user model. This raises significant doubts about the current practice of learning and evaluating strategies with the same user model. The paper further investigates a new technique for testing and comparing strategies directly on real human-machine dialogues, thereby avoiding any evaluation bias introduced by the user model. © 2005 IEEE.

Learning shape priors for single view reconstruction

In this paper, we aim to reconstruct free-from 3D models from a single view by learning the prior knowledge of a specific class of objects. Instead of heuristically proposing specific regularities and defining parametric models as previous research, our shape prior is learned directly from existing 3D models under a framework based on the Gaussian Process Latent Variable Model (GPLVM). The major contributions of the paper include: 1) a probabilistic framework for prior-based reconstruction we propose, which requires no heuristic of the object, and can be easily generalized to handle various categories of 3D objects, and 2) an attempt at automatic reconstruction of more complex 3D shapes, like human bodies, from 2D silhouettes only. Qualitative and quantitative experimental results on both synthetic and real data demonstrate the efficacy of our new approach. ©2009 IEEE.

Parameter learning for POMDP spoken dialogue models

The partially observable Markov decision process (POMDP) provides a popular framework for modelling spoken dialogue. This paper describes how the expectation propagation algorithm (EP) can be used to learn the parameters of the POMDP user model. Various special probability factors applicable to this task are presented, which allow the parameters be to learned when the structure of the dialogue is complex. No annotations, neither the true dialogue state nor the true semantics of user utterances, are required. Parameters optimised using the proposed techniques are shown to improve the performance of both offline transcription experiments as well as simulated dialogue management performance. ©2010 IEEE.

Tailored sustainability assessment for urban masterplanning

Tailored sustainability assessment represents one approach to addressing sustainability issues on large-scale urban projects with varying geographical, social and political constraints and diverse incentives among stakeholders. This paper examines the value and limitations of this approach. Three case studies of tailored systems developed by the authors for three unique masterplanning projects are discussed in terms of: contextual sustainability drivers; nature and evolution of systems developed; outcomes of implementation; and overall value delivered. Analysis Leads to conclusions on the key features of effective tailored assessment, the value of tailored sustainability assessment from various perspectives (including client, designer, end-users and the environment), and the limitations of tailored assessment as a tool for comparative analysis between projects. Although systems considered here are specific to individual projects and developed commercially, the challenges and lessons learned are relevant to a range of sustainability assessment approaches developed under different conditions.

Adaptation to stable and unstable dynamics achieved by combined impedance control and inverse dynamics model.

This study compared adaptation in novel force fields where trajectories were initially either stable or unstable to elucidate the processes of learning novel skills and adapting to new environments. Subjects learned to move in a null force field (NF), which was unexpectedly changed either to a velocity-dependent force field (VF), which resulted in perturbed but stable hand trajectories, or a position-dependent divergent force field (DF), which resulted in unstable trajectories. With practice, subjects learned to compensate for the perturbations produced by both force fields. Adaptation was characterized by an initial increase in the activation of all muscles followed by a gradual reduction. The time course of the increase in activation was correlated with a reduction in hand-path error for the DF but not for the VF. Adaptation to the VF could have been achieved solely by formation of an inverse dynamics model and adaptation to the DF solely by impedance control. However, indices of learning, such as hand-path error, joint torque, and electromyographic activation and deactivation suggest that the CNS combined these processes during adaptation to both force fields. Our results suggest that during the early phase of learning there is an increase in endpoint stiffness that serves to reduce hand-path error and provides additional stability, regardless of whether the dynamics are stable or unstable. We suggest that the motor control system utilizes an inverse dynamics model to learn the mean dynamics and an impedance controller to assist in the formation of the inverse dynamics model and to generate needed stability.

Impedance control balances stability with metabolically costly muscle activation.

Humans are able to stabilize their movements in environments with unstable dynamics by selectively modifying arm impedance independently of force and torque. We further investigated adaptation to unstable dynamics to determine whether the CNS maintains a constant overall level of stability as the instability of the environmental dynamics is varied. Subjects performed reaching movements in unstable force fields of varying strength, generated by a robotic manipulator. Although the force fields disrupted the initial movements, subjects were able to adapt to the novel dynamics and learned to produce straight trajectories. After adaptation, the endpoint stiffness of the arm was measured at the midpoint of the movement. The stiffness had been selectively modified in the direction of the instability. The stiffness in the stable direction was relatively unchanged from that measured during movements in a null force field prior to exposure to the unstable force field. This impedance modification was achieved without changes in force and torque. The overall stiffness of the arm and environment in the direction of instability was adapted to the force field strength such that it remained equivalent to that of the null force field. This suggests that the CNS attempts both to maintain a minimum level of stability and minimize energy expenditure.

Impedance control is tuned to multiple directions of movement

Humans are able to learn tool-handling tasks, such as carving, demonstrating their competency to make and vary the direction of movements in unstable environments. It has been shown that when a single reaching movement is repeated in unstable dynamics, the central nervous system (CNS) learns an impedance internal model to compensate for the environment instability. However, there is still no explanation for how humans can learn to move in various directions in such environments. In this study, we investigated whether and how humans compensate for instability while learning two different reaching movements simultaneously. Results show that when performing movements in two different directions, separated by a 35° angle, the CNS was able to compensate for the unstable dynamics. After adaptation, the force was found to be similar to the free movement condition, but stiffness increased in the direction of instability, specifically for each direction of movement. Our findings suggest that the CNS either learned an internal model generalizing over different movements, or alternatively that it was able to switch between specific models acquired simultaneously. © 2008 IEEE.

Reflex contributions to the directional tuning of arm stiffness

It has been shown that during arm movement, humans selectively change the endpoint stiffness of their arm to compensate for the instability in an unstable environment. When the direction of the instability is rotated with respect to the direction of movement, it was found that humans modify the antisymmetric component of their endpoint stiffness. The antisymmetric component of stiffness arises due to reflex responses suggesting that the subjects may have tuned their reflex responses as part of the feedforward adaptive control. The goal of this study was to examine whether the CNS modulates the gain of the reflex response for selective tuning of endpoint impedance. Subjects performed reaching movements in three unstable force fields produced by a robotic manipulandum, each field differing only in the rotational component. After subjects had learned to compensate for the field, allowing them to make unperturbed movements to the target, the endpoint stiffness of the arm was estimated in the middle of the movements. At the same time electromyographic activity (EMG) of six arm muscles was recorded. Analysis of the EMG revealed differences across force fields in the reflex gain of these muscles consistent with stiffness changes. This study suggests that the CNS modulates the reflex gain as part of the adaptive feedforward command in which the endpoint impedance is selectively tuned to overcome environmental instability. © 2008 Springer-Verlag Berlin Heidelberg.

Learning feedforward commands to muscles using time-shifted sensory feedback

In adapting to changing forces in the mechanical environment, humans change the force being applied by the limb by reciprocal changes in the activation of antagonistic muscles. However, they also cocontract these muscles when interaction with the environment is mechanically unstable to increase the mechanical impedance of the limb. We have postulated that appropriate patterns of muscle activation could be learned using a simple scheme in which the naturally occurring stretch reflex is used as a template to adjust feedforward commands to muscles. Feedforward commands are modified iteratively by shifting a scaled version of the reflex response forward in time and adding it to the previous feedforward command. We show that such an algorithm can account for the principal features of changes in muscle activation observed when human subjects adapt to instabilities in the mechanical environment. © 2006.