Use of Self-Selected Postures to Regulate Multi-Joint Stiffness During Unconstrained Tasks

Background

The human motor system is highly redundant, having more kinematic degrees of freedom than necessary to complete a given task. Understanding how kinematic redundancies are utilized in different tasks remains a fundamental question in motor control. One possibility is that they can be used to tune the mechanical properties of a limb to the specific requirements of a task. For example, many tasks such as tool usage compromise arm stability along specific directions. These tasks only can be completed if the nervous system adapts the mechanical properties of the arm such that the arm, coupled to the tool, remains stable. The purpose of this study was to determine if posture selection is a critical component of endpoint stiffness regulation during unconstrained tasks.

Methodology/Principal Findings

Three-dimensional (3D) estimates of endpoint stiffness were used to quantify limb mechanics. Most previous studies examining endpoint stiffness adaptation were completed in 2D using constrained postures to maintain a non-redundant mapping between joint angles and hand location. Our hypothesis was that during unconstrained conditions, subjects would select arm postures that matched endpoint stiffness to the functional requirements of the task. The hypothesis was tested during endpoint tracking tasks in which subjects interacted with unstable haptic environments, simulated using a 3D robotic manipulator. We found that arm posture had a significant effect on endpoint tracking accuracy and that subjects selected postures that improved tracking performance. For environments in which arm posture had a large effect on tracking accuracy, the self-selected postures oriented the direction of maximal endpoint stiffness towards the direction of the unstable haptic environment.

Conclusions/Significance

These results demonstrate how changes in arm posture can have a dramatic effect on task performance and suggest that postural selection is a fundamental mechanism by which kinematic redundancies can be exploited to regulate arm stiffness in unconstrained tasks.     

Motor control and learning in altered dynamic environments

Dynamic perturbations of reaching movements are an important technique for studying motor learning and adaptation. Adaptation to non-contacting, velocity-dependent inertial Coriolis forces generated by arm movements during passive body rotation is very rapid, and when complete the Coriolis forces are no longer sensed. Adaptation to velocity-dependent forces delivered by a robotic manipulandum takes longer and the perturbations continue to be perceived even when adaptation is complete. These differences reflect adaptive self-calibration of motor control versus learning the behavior of an external object or 'tool'. Velocity-dependent inertial Coriolis forces also arise in everyday behavior during voluntary turn and reach movements but because of anticipatory feedforward motor compensations do not affect movement accuracy despite being larger than the velocity-dependent forces typically used in experimental studies. Progress has been made in understanding: the common features that determine adaptive responses to velocity-dependent perturbations of jaw and limb movements; the transfer of adaptation to mechanical perturbations across different contact sites on a limb; and the parcellation and separate representation of the static and dynamic components of multiforce perturbations.     

Limb Dominance Results from Asymmetries in Predictive and Impedance Control Mechanisms

Handedness is a pronounced feature of human motor behavior, yet the underlying neural mechanisms remain unclear. We hypothesize that motor lateralization results from asymmetries in predictive control of task dynamics and in control of limb impedance. To test this hypothesis, we present an experiment with two different force field environments, a field with a predictable magnitude that varies with the square of velocity, and a field with a less predictable magnitude that varies linearly with velocity. These fields were designed to be compatible with controllers that are specialized in predicting limb and task dynamics, and modulating position and velocity dependent impedance, respectively. Because the velocity square field does not change the form of the equations of motion for the reaching arm, we reasoned that a forward dynamic-type controller should perform well in this field, while control of linear damping and stiffness terms should be less effective. In contrast, the unpredictable linear field should be most compatible with impedance control, but incompatible with predictive dynamics control. We measured steady state final position accuracy and 3 trajectory features during exposure to these fields: Mean squared jerk, Straightness, and Movement time. Our results confirmed that each arm made straighter, smoother, and quicker movements in its compatible field. Both arms showed similar final position accuracies, which were achieved using more extensive corrective sub-movements when either arm performed in its incompatible field. Finally, each arm showed limited adaptation to its incompatible field. Analysis of the dependence of trajectory errors on field magnitude suggested that dominant arm adaptation occurred by prediction of the mean field, thus exploiting predictive mechanisms for adaptation to the unpredictable field. Overall, our results support the hypothesis that motor lateralization reflects asymmetries in specific motor control mechanisms associated with predictive control of limb and task dynamics, and modulation of limb impedance.     

A neural network model for the intersensory coordination involved in goal-directed movements

A neural network model for a sensorimotor system, which was developed to simulate oriented movements in man, is presented. It is composed of a formal neural network comprising two layers: a sensory layer receiving and processing sensory inputs, and a motor layer driving a simulated arm. The sensory layer is an extension of the topological network previously proposed by Kohonen (1984). Two kinds of sensory modality, proprioceptive and exteroceptive, are used to define the arm position. Each sensory cell receives proprioceptive inputs provided by each arm-joint together with the exteroceptive inputs. This sensory layer is therefore a kind of associative layer which integrates two separate sensory signals relating to movement coding. It is connected to the motor layer by means of adaptive synapses which provide a physical link between a motor activity and its sensory consequences. After a learning period, the spatial map which emerges in the sensory layer clearly depends on the sensory inputs and an associative map of both the arm and the extra-personal space is built up if proprioceptive and exteroceptive signals are processed together. The sensorimotor transformations occuring in the junctions linking the sensory and motor layers are organized in such a manner that the simulated arm becomes able to reach towards and track a target in extra-personal space. Proprioception serves to determine the final arm posture adopted and to correct the ongoing movement in cases where changes in the target location occur. With a view to developing a sensorimotor control system with more realistic salient features, a robotic model was coupled with the formal neural network. This robotic implementation of our model shows the capacity of formal neural networks to control the displacement of mechanical devices.     

A bioinspired autonomous swimming robot as a tool for studying goal-directed locomotion

The bioinspired approach has been key in combining the disciplines of robotics with neuroscience in an effective and promising fashion. Indeed, certain aspects in the field of neuroscience, such as goal-directed locomotion and behaviour selection, can be validated through robotic artefacts. In particular, swimming is a functionally important behaviour where neuromuscular structures, neural control architecture and operation can be replicated artificially following models from biology and neuroscience. In this article, we present a biomimetic system inspired by the lamprey, an early vertebrate that locomotes using anguilliform swimming. The artefact possesses extra- and proprioceptive sensory receptors, muscle-like actuation, distributed embedded control and a vision system. Experiments on optimised swimming and on goal-directed locomotion are reported, as well as the assessment of the performance of the system, which shows high energy efficiency and adaptive behaviour. While the focus is on providing a robotic platform for testing biological models, the reported system can also be of major relevance for the development of engineering system applications.     

Development and Analysis of an Electrically Actuated Lower Extremity Assistive Exoskeleton

An electrically actuated lower extremity exoskeleton is developed,in which only the knee joint is actuated actively while other joints linked by elastic elements are actuated passively.This paper describes the critical design criteria and presents the process of design and calculation of the actuation system.A flexible physical Human-Robot-Interaction (pHRI) measurement device is designed and applied to detect the human movement,which comprises two force sensors and two gasbags attached to the inner surface of the connection cuff.An online adaptive pHRI minimization control strategy is proposed and implemented to drive the robotic exoskeleton system to follow the motion trajectory of human limb.The measured pHRI information is fused by the Variance Weighted Average (VWA) method.The Mean Square Values (MSV) of pHRI and control torque are utilized to evaluate the performance of the exoskeleton.To improve the comfort level and reduce energy consumption,the gravity compensation is taken into consideration when the control law is designed.Finally,practical experiments are performed on healthy users.Experimental results show that the proposed system can assist people to walk and the outlined control strategy is valid and effective.     

Development of an Anthropomorphic Robotic Arm and Hand for Interactive Humanoids

Humanoid robots are designed and built to mimic human form and movement.Ultimately,they are meant to resemble the size and physical abilities of a human in order to function in human-oriented environments and to work autonomously but to pose no physical threat to humans.Here,a humanoid robot that resembles a human in appearance and movement is built using powerful actuators paired with gear trains,joint mechanisms,and motor drivers that are all encased in a package no larger than that of the human physique.In this paper,we propose the construction of a humanoid-applicable anthropomorphic 7-DoF arm complete with an 8-DoF hand.The novel mechanical design of this humanoid arm makes it sufficiently compact to be compatible with currently available narrating-model humanoids,and to be sufficiently powerful and flexible to be functional; the number of degrees of freedom endowed in this robotic arm is sufficient for executing a wide range of tasks,including dexterous hand movements.The developed humanoid arm and hand are capable of sensing and interpreting incoming external force using the motor in each joint current without conventional torque sensors.The humanoid arm adopts an algorithm to avoid obstacles and the dexterous hand is capable of grasping objects.The developed robotic arm is suitable for use in an interactive humanoid robot.     

Impedance characteristics of a neuromusculoskeletal model of the human arm II. Movement control

The modulation of neuromusculoskeletal impedance during movements is analysed using a motor control model of the human arm. The motor control system combines feedback and feedforward control and both control modes are determined in one optimization process. In the model, the stiffness varies at the double movement frequency for 2-Hz oscillatory elbow movements and has high values at the movement reversals. During goal-directed two-degrees-of-freedom arm movements, the stiffness is decreased during the movement and may be increased in the initial and final phases, depending on the movement velocity. The stiffness has a considerable curl during the movement, as was also observed in experimental data. The dynamic stiffness patterns of the model can be explained basically by the α−γ coactivation scheme where feedback gains covary with motor control signals. In addition to the modulation of the gain factors, it is argued that the variation of the intrinsic stiffness has a considerable effect on movement control, especially during fast movements. Received: 14 October 1997 / Accepted in revised form: 18 May 1999     

A human motor control perspective to multiple manipulator modelling

This paper describes the aspects involved in modelling a multi-robot system from a human motor control perspective. The human motor control system has a hierarchical and decentralised structure, and building a control system for a multi-robot system that attains human features would require a decomposable model. Decomposition of a complex robotic system is difficult due to the interactions between the subsystems, so these have to be first separated before the system is modelled. The proposed method of separating the interconnections is applied with the aid of fuzzy modelling to derive a fully decomposable model of two manipulator robots handling a common object.     

Design of a Robotized Magnetic Platform for Targeted Drug Delivery in the Cochlea

Inner ear disorders' treatment remains challenging due to anatomical barriers. Robotic assistance seems to be a promising approach to enhance inner ear treatments and, more particularly, lead to effective targeted drug delivery into the human cochlea. In this paper we present a combination of a micro-macro system that was designed and realized in order to efficiently control the navigation of magnetic nanoparticles in an open-loop scheme throughout the cochlea, considering that the magnetic particles cannot be located in real time.In order to respect the anatomical constraints, we established the characteristics that the new platform must present then proceeded to the design of the latter. The developed system is composed of a magnetic actuator that aims to guide nanoparticles into the cochlea. Mounted on a robotic manipulator, it ensures its positioning around the patient's head. The magnetic device integrates four parallelepiped-rectangle permanent magnets. Their arrangement in space, position and orientation, allows the creation of an area of convergence of magnetic forces where nanoparticles can be pushed/pulled to. To ensure the reachability of the desired orientations and positions, a 3 DOF robot based on a Remote Centre of Motion (RCM) mechanism was developed. It features three concurrent rotational joints that generate a spherical workspace around the head. The control of the latter is based on kinematic models.A prototype of this platform was realized to validate the actuation process. Both magnetic actuator and robotic manipulator were realized using an additive manufacturing approach. We also designed a virtual human head with a life-size cochlea inside. A laser was mounted on the end effector to track the positioning of the actuator. This permitted to experimentally prove the capacity of the robotic system to reach the desired positions and orientations in accordance with the medical needs.This promising robotic approach, makes it possible to overcome anatomical barriers and steer magnetic nanoparticles to a targeted location in the inner ear and, more precisely, inside the cochlea.     

Simulation studies on the control of posture and movement in a multi-jointed limb

Simulation studies were performed to evaluate the effectiveness of different control schemes in stabilizing a multi-jointed limb (human arm) in response to force perturbations. The mechanical properties of the arm were modeled as a linear visco-elastic system and the effectiveness of negative feedback of angular position and torque was evaluated. The effectiveness of a given amount of position feedback depended strongly on the initial position of the arm and on the perturbation, while torque feedback was much more consistently effective in damping the motion of the limb.     

Iterative learning control applied to a hybrid rehabilitation exoskeleton system powered by PAM and FES

Currently upper limb exoskeleton rehabilitation robots powered by electric motors used in the hospitals are usually cumbersome, bulky and unmovable. Our developed RUPERT is a low-cost lightweight portable exoskeleton rehabilitation robot that can encourage stroke patients with high stiffness in arm flexor muscles to receive frequent intensive rehabilitation trainings in the community or home, but its joints are unidirectionally actuated by pneumatic artificial muscles (PAMs). RUPERT with one PAM of each joint is not suitable for stroke patients with weak muscles in the flaccid paralysis period. Functional electrical stimulation (FES) uses current with low frequency to activate paralyzed muscles, which can produce muscle torque and compensate the unidirectional drawbacks of RUPERT, so as to realize the two-way motion of its joints for passive reaching trainings. As both the exoskeleton robot driven by PAMs and neuromuscular skeletal system under FES possess the highly nonlinear and time-varying characteristics, which adds control difficulty to the hybrid dynamic system, iterative learning control (ILC) is chosen to control this newly designed hybrid rehabilitation system to realize repetitive task trainings.     

Independent Mobility Achieved through a Wireless Brain-Machine Interface

Individuals with tetraplegia lack independent mobility, making them highly dependent on others to move from one place to another. Here, we describe how two macaques were able to use a wireless integrated system to control a robotic platform, over which they were sitting, to achieve independent mobility using the neuronal activity in their motor cortices. The activity of populations of single neurons was recorded using multiple electrode arrays implanted in the arm region of primary motor cortex, and decoded to achieve brain control of the platform. We found that free-running brain control of the platform (which was not equipped with any machine intelligence) was fast and accurate, resembling the performance achieved using joystick control. The decoding algorithms can be trained in the absence of joystick movements, as would be required for use by tetraplegic individuals, demonstrating that the non-human primate model is a good pre-clinical model for developing such a cortically-controlled movement prosthetic. Interestingly, we found that the response properties of some neurons differed greatly depending on the mode of control (joystick or brain control), suggesting different roles for these neurons in encoding movement intention and movement execution. These results demonstrate that independent mobility can be achieved without first training on prescribed motor movements, opening the door for the implementation of this technology in persons with tetraplegia.     

A method for measuring endpoint stiffness during multi-joint arm movements

Current methods for measuring stiffness during human arm movements are either limited to one-joint motions, or lead to systematic errors. The technique presented here enables a simple, accurate and unbiased measurement of endpoint stiffness during multi-joint movements. Using a computer-controlled mechanical interface, the hand is displaced relative to a prediction of the undisturbed trajectory. Stiffness is then computed as the ratio of restoring force to displacement amplitude. Because of the accuracy of the prediction (< 1 cm error after 200 ms) and the quality of the implementation, the movement is not disrupted by the perturbation. This technique requires only 13 as many trials to identify stiffness as the method of Gomi and Kawato (1997, Biological Cybernetics 76, 163-171) and may, therefore, be used to investigate the evolution of stiffness during motor adaptation.     

Impedance characteristics of a neuromusculoskeletal model of the human arm I. Posture control

The mechanical impedance of neuromusculoskeletal models of the human arm is studied in this paper. The model analysis provides a better understanding of the contributions of possible intrinsic and reflexive components of arm impedance, makes clear the limitations of second-order mass-viscosity-stiffness models and reveals possible task effects on the impedance. The musculoskeletal model describes planar movements of the upper arm and forearm, which are moved by six lumped muscles with nonlinear dynamics. The motor control system is represented by a neural network which combines feedforward and feedback control. It is optimized for the control of movements or for posture control in the presence of external forces. The achieved impedance characteristics depend on the conditions during the learning process. In particular, the impedance is adapted in a suitable way to the frequency content and direction of external forces acting on the hand during an isometric task. The impedance characteristics of a model, which is optimized for movement control, are similar to experimental data in the literature. The achieved stiffness is, to a large extent, reflexively determined whereas the approximated viscosity is primarily due to intrinsic attributes. It is argued that usually applied Hill-type muscle models do not properly represent intrinsic muscle stiffness. Received: 14 October 1997 / Accepted in revised form: 18 May 1999     

Measuring multi-joint stiffness during single movements: numerical validation of a novel time-frequency approach

This study presents and validates a Time-Frequency technique for measuring 2-dimensional multijoint arm stiffness throughout a single planar movement as well as during static posture. It is proposed as an alternative to current regressive methods which require numerous repetitions to obtain average stiffness on a small segment of the hand trajectory. The method is based on the analysis of the reassigned spectrogram of the arm's response to impulsive perturbations and can estimate arm stiffness on a trial-by-trial basis. Analytic and empirical methods are first derived and tested through modal analysis on synthetic data. The technique's accuracy and robustness are assessed by modeling the estimation of stiffness time profiles changing at different rates and affected by different noise levels. Our method obtains results comparable with two well-known regressive techniques. We also test how the technique can identify the viscoelastic component of non-linear and higher than second order systems with a non-parametrical approach. The technique proposed here is very impervious to noise and can be used easily for both postural and movement tasks. Estimations of stiffness profiles are possible with only one perturbation, making our method a useful tool for estimating limb stiffness during motor learning and adaptation tasks, and for understanding the modulation of stiffness in individuals with neurodegenerative diseases.     

Offset of rotation centers creates a bias in isokinetics: a virtual model including stiffness or friction

The present paper deals with a virtual model devoted to isokinetics and isometrics assessment of a human muscular group in the common joints, knee, ankle, hip, shoulder, cervical spine, etc. This virtual model with an analytical analysis followed by a numerical simulation is able to predict measurement errors of the joint torque due to offset of rotation centers between the body segment and the ergometer arm. As soon as offset is present, errors increase due to the influence of inertial effects, gravity effects, stiffness due to the limb strapping on the ergometer arm or Coulomb friction between limb and ergometer. The analytical model is written in terms of Lagrange formalism and the numerical model uses ADAMS software adapted to multi-body dynamics simulations. Results of models show a maximal relative error of 11%, for a 10% relative offset between the rotation centers. Inertial contributions are found to be negligible but gravity effects must be discussed in regard to the measured torque. Stiffness or friction effects may also increase the torque error; in particular when offset occurs, it is shown that errors due to friction have to be considered for all torque level while only stiffness effects have to be considered for torque less than 25Nm. This study also emphasizes the influence of the angular range of motion at a given angular position.     

An interference effect of observed biological movement on action

It has been proposed that actions are intrinsically linked to perception and that imagining, observing, preparing, or in any way representing an action excites the motor programs used to execute that same action. There is neurophysiological evidence that certain brain regions involved in executing actions are activated by the mere observation of action (the so-called "mirror system;" ). However, it is unknown whether this mirror system causes interference between observed and simultaneously executed movements. In this study we test the hypothesis that, because of the overlap between action observation and execution, observed actions should interfere with incongruous executed actions. Subjects made arm movements while observing either a robot or another human making the same or qualitatively different arm movements. Variance in the executed movement was measured as an index of interference to the movement. The results demonstrate that observing another human making incongruent movements has a significant interference effect on executed movements. However, we found no evidence that this interference effect occurred when subjects observed a robotic arm making incongruent movements. These results suggest that the simultaneous activation of the overlapping neural networks that process movement observation and execution infers a measurable cost to motor control.     

Bispectrum-based features classification for myoelectric control

Surface electromyographic signals provide useful information about motion intentionality. Therefore, they are a suitable reference signal for control purposes. A continuous classification scheme of five upper limb movements applied to a myoelectric control of a robotic arm is presented. This classification is based on features extracted from the bispectrum of four EMG signal channels. Among several bispectrum estimators, this paper is focused on arithmetic mean, median, and trimmed mean estimators, and their ensemble average versions. All bispectrum estimators have been evaluated in terms of accuracy, robustness against outliers, and computational time. The median bispectrum estimator shows low variance and high robustness properties. Two feature reduction methods for the complex bispectrum matrix are proposed. The first one estimates the three classic means (arithmetic, harmonic, and geometric means) from the module of the bispectrum matrix, and the second one estimates the same three means from the square of the real part of the bispectrum matrix. A two-layer feedforward network for movement's classification and a dedicated system to achieve the myoelectric control of a robotic arm were used. It was found that the classification performance in real-time is similar to those obtained off-line by other authors, and that all volunteers in the practical application successfully completed the control task.     

A Flexible Fin with Bio-Inspired Stiffness Profile and Geometry

Biological evidence suggests that fish use mostly anterior muscles for steady swimming while the caudal part of the body is passive and,acting as a carrier of energy,transfers the momentum to the surrounding water.Inspired by those findings we hypothesize that certain swimming patterns can be achieved without copying the distributed actuation mechanism of fish but rather using a single actuator at the anterior part to create the travelling wave.To test the hypothesis a pitching flexible fin made of silicone rubber and silicone foam was designed by copying the stiffness distribution profile and geometry of a rainbow trout.The kinematics of the fin was compared to that of a steadily swimming trout.Fin's propulsive wave length and tail-beat amplitude were determined while it was actuated by a single servo motor.Results showed that the propulsive wave length and tail-beat amplitude of a steadily swimming 50 cm rainbow trout was achieved with our biomimetic fin while stimulated using certain actuation parameters (frequency 2.31 Hz and amplitude 6.6 degrees).The study concluded that fish-like swimming can be achieved by mimicking the stiffness and geometry of a rainbow trout and disregarding the details of the actuation mechanism.