Vestibular humanoid postural control

Human body-sway steady-state responses to small amplitude tilts and translations of the support surface - Effects of superposition of the two stimuli

BACKGROUND

For humans, control of upright standing posture is a prerequisite for many physical activities. Experimentally, this control is often challenged by the motion of the support surface presented as tilt or translation, or some combination thereof. In particular, we have investigated subjects balancing in situations where tilt and translation stimuli were presented in isolation and compared it to a situation where such stimuli occurred simultaneously.

RESEARCH QUESTIONS

Is the human posture control system in the case of two or more superimposed external disturbances responding to these as if it were dealing with one disturbance? Or does it identify the disturbances individually and as such and respond to them specifically, as suggested in a current concept of disturbance-specific estimations and compensations?

METHODS

We had healthy human subjects controlling their balancing of upright stance on a motion platform while we presented them with different combinations of pseudorandom support surface tilt and translation stimuli alone or in superposition (with peak-to-peak amplitude of 0.5° and 1° for tilt, and 0.8 cm and 1.5 cm for translation). In one set of trials they kept their eyes closed and in a second set open. Furthermore, a simulation was performed to qualitatively evaluate the impact of sensory non-linearities and joint stiffness modulation.

RESULTS

We found that the experimental conditions 'eyes open' vs. 'eyes closed' always created significant differences (p < 0.05) between the frequency response functions. In contrast to this, with different combinations of the tilt and translation stimuli, significant differences between the responses were observed only in 5 cases over the 24 that have been tested. Significance The superposition of translation and tilt can be used to characterize the responses to both stimuli with one trial. When the amplitude of the stimuli is unbalanced (e.g. very small tilt superimposed with a larger translation) the effect of stiffness modulation can be studied.

Gymnastics Experience Enhances the Development of Bipedal-Stance Multi-Segmental Coordination and Control During Proprioceptive Reweighting

Performance and control of upright bipedal posture requires a constant and dynamic integration of relative contributions of different sensory inputs (i. e., sensory reweighting) to enable effective adaptations as individuals face environmental changes and perturbations. Children with gymnastic experience showed balance performance closer to that of adults during and after proprioceptive alteration than children without gymnastic experience when their center of pressure (COP) was analyzed. However, a particular COP sway can be achieved through performing and coordinating different postural movements. The aim of this study was to assess how children and adults of different gymnastic experience perform and control postural movements while they have to adjust balance during and after bilateral tendon vibration. All participants were equipped with spherical markers attached to their skin and two vibrators strapped over the Achilles tendons. Bipedal stance was performed in three 45-s trials in two visual conditions (eyes open, EO, and eyes closed, EC) ordered randomly in which vibration lasted 10 s. Posture movements were analyzed by a principal component analysis (PCA) calculated on normalized and weighted markers coordinates. The relative standard deviation of each principal movement component (principal position, PP-rSTD) quantified its contribution to the whole postural movements, i.e., quantified the coordinative structure. The first (principal velocities, PV-rSTD) and second (principal accelerations, PA-rSTD) time-derivatives characterized the rate-dependent sensory information associated with and the neuromuscular control of the postural movements, respectively. Children without gymnastic experience showed a different postural coordinative structure and different sensory-motor control characteristics. They used less ankle movements in the anterior-posterior direction but increased ankle movements in medio-lateral direction, presented larger hip and trunk velocities, and exhibited more hip actions. Gymnastic experience during childhood seemed to benefit the development of proprioceptive reweighting processes in children, leading to a more mature form of coordinating and controlling posture similarly to adults.

Visual Modulation of Human Responses to Support Surface Translation

Vision is known to improve human postural responses to external perturbations. This study investigates the role of vision for the responses to continuous pseudorandom support surface translations in the body sagittal plane in three visual conditions: with the eyes closed (EC), in stroboscopic illumination (EO/SI; only visual position information) and with eyes open in continuous illumination (EO/CI; position and velocity information) with the room as static visual scene (or the interior of a moving cabin, in some of the trials). In the frequency spectrum of the translation stimulus we distinguished on the basis of the response patterns between a low-frequency, mid-frequency, and high-frequency range (LFR: 0.0165-0.14 Hz; MFR: 0.15–0.57 Hz; HFR: 0.58–2.46 Hz). With EC, subjects’ mean sway response gain was very low in the LFR. On average it increased with EO/SI (although not to a significant degree p = 0.078) and more so with EO/CI (p < 10⁻⁶). In contrast, the average gain in the MFR decreased from EC to EO/SI (although not to a significant degree, p = 0.548) and further to EO/CI (p = 0.0002). In the HFR, all three visual conditions produced, similarly, high gain levels. A single inverted pendulum (SIP) model controlling center of mass (COM) balancing about the ankle joints formally described the EC response as being strongly shaped by a resonance phenomenon arising primarily from the control’s proprioceptive feedback loop. The effect of adding visual information in these simulations lies in a reduction of the resonance, similar as in the experiments. Extending the model to a double inverted pendulum (DIP) suggested in addition a biomechanical damping effective from trunk sway in the hip joints on the resonance.

Physiological Vibration Acceleration (Phybrata) Sensor Assessment of Multi-System Physiological Impairments and Sensory Reweighting Following Concussion

Objective

To assess the utility of a head-mounted wearable inertial motion unit (IMU)-based physiological vibration acceleration (“phybrata”) sensor to support the clinical diagnosis of concussion, classify and quantify specific concussion-induced physiological system impairments and sensory reweighting, and track individual patient recovery trajectories.

Methods

Data were analyzed from 175 patients over a 12-month period at three clinical sites. Comprehensive clinical concussion assessments were first completed for all patients, followed by testing with the phybrata sensor. Phybrata time series data and spatial scatter plots, eyes open (Eo) and eyes closed (Ec) phybrata powers, average power (Eo+Ec)/2, Ec/Eo phybrata power ratio, time-resolved phybrata spectral density (TRPSD) distributions, and receiver operating characteristic (ROC) curves are compared for individuals with no objective impairments and those clinically diagnosed with concussions and accompanying vestibular impairment, other neurological impairment, or both vestibular and neurological impairments. Finally, pre- and post-injury phybrata case report results are presented for a participant who was diagnosed with a concussion and subsequently monitored during treatment, rehabilitation, and return-to-activity clearance.

Results

Phybrata data demonstrate distinct features and patterns for individuals with no discernable clinical impairments, diagnosed vestibular pathology, and diagnosed neurological pathology. ROC curves indicate that the average power (Eo+Ec)/2 may be utilized to support clinical diagnosis of concussion, while Eo and Ec/Eo may be utilized as independent measures to confirm accompanying neurological and vestibular impairments, respectively. All 3 measures demonstrate area under the curve (AUC), sensitivity, and specificity above 90% for their respective diagnoses. Phybrata spectral analyses demonstrate utility for quantifying the severity of concussion-induced physiological impairments, sensory reweighting, and subsequent monitoring of improvements throughout treatment and rehabilitation.

Conclusion

Phybrata testing assists with objective concussion diagnosis and provides an important adjunct to standard concussion assessment tools by objectively ascertaining neurological and vestibular impairments, guiding targeted rehabilitation strategies, monitoring recovery, and assisting with return-to-sport/work/learn decision-making.

Reductions in body sway responses to a rhythmic support surface tilt perturbation can be caused by other mechanisms than prediction

Studies investigating balance control often use external perturbations to probe the system. These perturbations can be administered as randomized, pseudo-randomized, or predictable sequences. As predictability of a given perturbation can affect balance performance, the way those perturbations are constructed may affect the results of the experiments. In the present study, we hypothesized that subjects are able to adapt to short, rhythmic support surface tilt stimuli, but not to long pseudo-random stimuli. 19 subjects were standing with eyes closed on a servo-controlled platform tilting about the ankle joint axis. Pre and post to the learning intervention, pseudo-random tilt sequences were applied. For the learning phase, a rhythmic and easy-to-memorize 8-s long sequence was applied 75 times, where subjects were instructed to stand as still as possible. Body kinematics were measured and whole body center of mass sway was analyzed. Results showed reduced sway and less forward lean of the body across the learning phase. The sway reductions were similar for stimulus and non-stimulus frequencies. Surprisingly, for the pseudo-random sequences, comparable changes were found from pre- to post-tests. In summary, results confirmed that considerable adaptations exist when exposing subjects to an 8-s long rhythmic perturbation. No indications of predictions of the learning tilt sequence were found, since similar changes were also observed in response to pseudo-random sequences. We conclude that changes in body sway responses following 75 repetitions of an 8-s long rhythmic tilt sequence are due to adaptations in the dynamics of the control mechanism (presumably stiffness).

Basin of Attraction and Limit Cycle Amplitude of an Ankle-Hip Model of Balance on a Balance Board

The study of upright posture (UP) stability is of relevance to estimating risk of falls, especially among people with neuromuscular deficits. Several studies have addressed this problem from a system dynamic approach based on parameter bifurcation analyses, which provide the region of stability (RoS) and the delimiting bifurcation curves (usually Hopf and pitchfork) in some parameter-spaces. In contrast, our goal is to determine the effect of parameter changes on the size of the basin of attraction (BoA) of the UP equilibrium and the amplitude of the limit cycle oscillations (LCOs) emerging from the Hopf bifurcations (HBs). The BoA is an indicator of the ability of the UP to maintain balance without falling while LCOs may explain the sway motion commonly observed during balancing. In this study, a three degree of freedom model for a human balancing on a balance board was developed. Analysis of the model revealed the BoAs and the amplitude of the LCOs. Results show that physical parameters (time-delays and feedback control gains) have a large impact on the size of the BoA and the amplitude of the LCOs. Particularly, the size of the BoA increases when balancing on a rigid surface and decreases when either proprioceptive or combined visual and vestibular feedback gain is too high. With respect to the LCOs, it is shown that they emerge from both the subcritical and supercritical HBs and increase their amplitudes as some parameters vary.

The parieto-insular vestibular cortex in humans: more than a single area?

Here, we review the structure and function of a core region in the vestibular cortex of humans that is located in the midposterior Sylvian fissure and referred to as the parieto-insular vestibular cortex (PIVC). Previous studies have investigated PIVC by using vestibular or visual motion stimuli and have observed activations that were distributed across multiple anatomical structures, including the temporo-parietal junction, retroinsula, parietal operculum, and posterior insula. However, it has remained unclear whether all of these anatomical areas correspond to PIVC and whether PIVC responds to both vestibular and visual stimuli. Recent results suggest that the region that has been referred to as PIVC in previous studies consists of multiple areas with different anatomical correlates and different functional specializations. Specifically, a vestibular but not visual area is located in the parietal operculum, close to the posterior insula, and likely corresponds to the nonhuman primate PIVC, while a visual-vestibular area is located in the retroinsular cortex and is referred to, for historical reasons, as the posterior insular cortex area (PIC). In this article, we review the anatomy, connectivity, and function of PIVC and PIC and propose that the core of the human vestibular cortex consists of at least two separate areas, which we refer to together as PIVC+. We also review the organization in the nonhuman primate brain and show that there are parallels to the proposed organization in humans.

Posture Control-Human-Inspired Approaches for Humanoid Robot Benchmarking: Conceptualizing Tests, Protocols and Analyses

Posture control is indispensable for both humans and humanoid robots, which becomes especially evident when performing sensorimotor tasks such as moving on compliant terrain or interacting with the environment. Posture control is therefore targeted in recent proposals of robot benchmarking in order to advance their development. This Methods article suggests corresponding robot tests of standing balance, drawing inspirations from the human sensorimotor system and presenting examples from robot experiments. To account for a considerable technical and algorithmic diversity among robots, we focus in our tests on basic posture control mechanisms, which provide humans with an impressive postural versatility and robustness. Specifically, we focus on the mechanically challenging balancing of the whole body above the feet in the sagittal plane around the ankle joints in concert with the upper body balancing around the hip joints. The suggested tests target three key issues of human balancing, which appear equally relevant for humanoid bipeds: (1) four basic physical disturbances (support surface (SS) tilt and translation, field and contact forces) may affect the balancing in any given degree of freedom (DoF). Targeting these disturbances allows us to abstract from the manifold of possible behavioral tasks. (2) Posture control interacts in a conflict-free way with the control of voluntary movements for undisturbed movement execution, both with "reactive" balancing of external disturbances and "proactive" balancing of self-produced disturbances from the voluntary movements. Our proposals therefore target both types of disturbances and their superposition. (3) Relevant for both versatility and robustness of the control, linkages between the posture control mechanisms across DoFs provide their functional cooperation and coordination at will and on functional demands. The suggested tests therefore include ankle-hip coordination. Suggested benchmarking criteria build on the evoked sway magnitude, normalized to robot weight and Center of mass (COM) height, in relation to reference ranges that remain to be established. The references may include human likeness features. The proposed benchmarking concept may in principle also be applied to wearable robots, where a human user may command movements, but may not be aware of the additionally required postural control, which then needs to be implemented into the robot.

Evidence in Support of the Independent Channel Model Describing the Sensorimotor Control of Human Stance Using a Humanoid Robot

The Independent Channel (IC) model is a commonly used linear balance control model in the frequency domain to analyze human balance control using system identification and parameter estimation. The IC model is a rudimentary and noise-free description of balance behavior in the frequency domain, where a stable model representation is not guaranteed. In this study, we conducted firstly time-domain simulations with added noise, and secondly robot experiments by implementing the IC model in a real-world robot (PostuRob II) to test the validity and stability of the model in the time domain and for real world situations. Balance behavior of seven healthy participants was measured during upright stance by applying pseudorandom continuous support surface rotations. System identification and parameter estimation were used to describe the balance behavior with the IC model in the frequency domain. The IC model with the estimated parameters from human experiments was implemented in Simulink for computer simulations including noise in the time domain and robot experiments using the humanoid robot PostuRob II. Again, system identification and parameter estimation were used to describe the simulated balance behavior. Time series, Frequency Response Functions, and estimated parameters from human experiments, computer simulations, and robot experiments were compared with each other. The computer simulations showed similar balance behavior and estimated control parameters compared to the human experiments, in the time and frequency domain. Also, the IC model was able to control the humanoid robot by keeping it upright, but showed small differences compared to the human experiments in the time and frequency domain, especially at high frequencies. We conclude that the IC model, a descriptive model in the frequency domain, can imitate human balance behavior also in the time domain, both in computer simulations with added noise and real world situations with a humanoid robot. This provides further evidence that the IC model is a valid description of human balance control.

Haptic Cues for Balance: Use of a Cane Provides Immediate Body Stabilization

Haptic cues are important for balance. Knowledge of the temporal features of their effect may be crucial for the design of neural prostheses. Touching a stable surface with a fingertip reduces body sway in standing subjects eyes closed (EC), and removal of haptic cue reinstates a large sway pattern. Changes in sway occur rapidly on changing haptic conditions. Here, we describe the effects and time-course of stabilization produced by a haptic cue derived from a walking cane. We intended to confirm that cane use reduces body sway, to evaluate the effect of vision on stabilization by a cane, and to estimate the delay of the changes in body sway after addition and withdrawal of haptic input. Seventeen healthy young subjects stood in tandem position on a force platform, with eyes closed or open (EO). They gently lowered the cane onto and lifted it from a second force platform. Sixty trials per direction of haptic shift (Touch → NoTouch, T-NT; NoTouch → Touch, NT-T) and visual condition (EC-EO) were acquired. Traces of Center of foot Pressure (CoP) and the force exerted by cane were filtered, rectified, and averaged. The position in space of a reflective marker positioned on the cane tip was also acquired by an optoelectronic device. Cross-correlation (CC) analysis was performed between traces of cane tip and CoP displacement. Latencies of changes in CoP oscillation in the frontal plane EC following the T-NT and NT-T haptic shift were statistically estimated. The CoP oscillations were larger in EC than EO under both T and NT (p < 0.001) and larger during NT than T conditions (p < 0.001). Haptic-induced effect under EC (Romberg quotient NT/T ~ 1.2) was less effective than that of vision under NT condition (EC/EO ~ 1.5) (p < 0.001). With EO cane had little effect. Cane displacement lagged CoP displacement under both EC and EO. Latencies to changes in CoP oscillations were longer after addition (NT-T, about 1.6 s) than withdrawal (T-NT, about 0.9 s) of haptic input (p < 0.001). These latencies were similar to those occurring on fingertip touch, as previously shown. Overall, data speak in favor of substantial equivalence of the haptic information derived from both “direct” fingertip contact and “indirect” contact with the floor mediated by the cane. Cane, finger and visual inputs would be similarly integrated in the same neural centers for balance control. Haptic input from a walking aid and its processing time should be considered when designing prostheses for locomotion.

A Sensitivity Analysis of an Inverted Pendulum Balance Control Model

Balance control models are used to describe balance behavior in health and disease. We identified the unique contribution and relative importance of each parameter of a commonly used balance control model, the Independent Channel (IC) model, to identify which parameters are crucial to describe balance behavior. The balance behavior was expressed by transfer functions (TFs), representing the relationship between sensory perturbations and body sway as a function of frequency, in terms of amplitude (i.e., magnitude) and timing (i.e., phase). The model included an inverted pendulum controlled by a neuromuscular system, described by several parameters. Local sensitivity of each parameter was determined for both the magnitude and phase using partial derivatives. Both the intrinsic stiffness and proportional gain shape the magnitude at low frequencies (0.1–1 Hz). The derivative gain shapes the peak and slope of the magnitude between 0.5 and 0.9 Hz. The sensory weight influences the overall magnitude, and does not have any effect on the phase. The effect of the time delay becomes apparent in the phase above 0.6 Hz. The force feedback parameters and intrinsic stiffness have a small effect compared with the other parameters. All parameters shape the TF magnitude and phase and therefore play a role in the balance behavior. The sensory weight, time delay, derivative gain, and the proportional gain have a unique effect on the TFs, while the force feedback parameters and intrinsic stiffness contribute less. More insight in the unique contribution and relative importance of all parameters shows which parameters are crucial and critical to identify underlying differences in balance behavior between different patient groups.

Devices and tasks involved in the objective assessment of standing dynamic balancing - A systematic literature review

BACKGROUND

Static balancing assessment is often complemented with dynamic balancing tasks. Numerous dynamic balancing assessment methods have been developed in recent decades with their corresponding balancing devices and tasks.

OBJECTIVE

The aim of this systematic literature review is to identify and categorize existing objective methods of standing dynamic balancing ability assessment with an emphasis on the balancing devices and tasks being used.

DATA SOURCES

Three major scientific literature databases (Science Direct, Web of Science, PLoS ONE) and additional sources were used.

STUDY SELECTION

Studies had to use a dynamic balancing device and a task described in detail. Evaluation had to be based on objectively measureable parameters. Functional tests without instrumentation evaluated exclusively by a clinician were excluded. A total of 63 articles were included.

DATA EXTRACTION

The data extracted during full-text assessment were: author and date; the balancing device with the balancing task and the measured parameters; the health conditions, size, age and sex of participant groups; and follow-up measurements.

DATA SYNTHESIS

A variety of dynamic balancing assessment devices were identified and categorized as 1) Solid ground, 2) Balance board, 3) Rotating platform, 4) Horizontal translational platform, 5) Treadmill, 6) Computerized Dynamic Posturography, and 7) Other devices. The group discrimination ability of the methods was explored and the conclusions of the studies were briefly summarized.

LIMITATIONS

Due to the wide scope of this search, it provides an overview of balancing devices and do not represent the state-of-the-art of any single method.

CONCLUSIONS

The identified dynamic balancing assessment methods are offered as a catalogue of candidate methods to complement static assessments used in studies involving postural control.

Human-Derived Disturbance Estimation and Compensation (DEC) Method Lends Itself to a Modular Sensorimotor Control in a Humanoid Robot

The high complexity of the human posture and movement control system represents challenges for diagnosis, therapy, and rehabilitation of neurological patients. We envisage that engineering-inspired, model-based approaches will help to deal with the high complexity of the human posture control system. Since the methods of system identification and parameter estimation are limited to systems with only a few DoF, our laboratory proposes a heuristic approach that step-by-step increases complexity when creating a hypothetical human-derived control systems in humanoid robots. This system is then compared with the human control in the same test bed, a posture control laboratory. The human-derived control builds upon the identified disturbance estimation and compensation (DEC) mechanism, whose main principle is to support execution of commanded poses or movements by compensating for external or self-produced disturbances such as gravity effects. In previous robotic implementation, up to 3 interconnected DEC control modules were used in modular control architectures separately for the sagittal plane or the frontal body plane and successfully passed balancing and movement tests. In this study we hypothesized that conflict-free movement coordination between the robot's sagittal and frontal body planes emerges simply from the physical embodiment, not necessarily requiring a full body control. Experiments were performed in the 14 DoF robot Lucy Posturob (i) demonstrating that the mechanical coupling from the robot's body suffices to coordinate the controls in the two planes when the robot produces movements and balancing responses in the intermediate plane, (ii) providing quantitative characterization of the interaction dynamics between body planes including frequency response functions (FRFs), as they are used in human postural control analysis, and (iii) witnessing postural and control stability when all DoFs are challenged together with the emergence of inter-segmental coordination in squatting movements. These findings represent an important step toward controlling in the robot in future more complex sensorimotor functions such as walking.

Human-Inspired Eigenmovement Concept Provides Coupling-Free Sensorimotor Control in Humanoid Robot

Control of a multi-body system in both robots and humans may face the problem of destabilizing dynamic coupling effects arising between linked body segments. The state of the art solutions in robotics are full state feedback controllers. For human hip-ankle coordination, a more parsimonious and theoretically stable alternative to the robotics solution has been suggested in terms of the Eigenmovement (EM) control. Eigenmovements are kinematic synergies designed to describe the multi DoF system, and its control, with a set of independent, and hence coupling-free, scalar equations. This paper investigates whether the EM alternative shows “real-world robustness” against noisy and inaccurate sensors, mechanical non-linearities such as dead zones, and human-like feedback time delays when controlling hip-ankle movements of a balancing humanoid robot. The EM concept and the EM controller are introduced, the robot's dynamics are identified using a biomechanical approach, and robot tests are performed in a human posture control laboratory. The tests show that the EM controller provides stable control of the robot with proactive (“voluntary”) movements and reactive balancing of stance during support surface tilts and translations. Although a preliminary robot-human comparison reveals similarities and differences, we conclude (i) the Eigenmovement concept is a valid candidate when different concepts of human sensorimotor control are considered, and (ii) that human-inspired robot experiments may help to decide in future the choice among the candidates and to improve the design of humanoid robots and robotic rehabilitation devices.

Sensory reweighting dynamics following removal and addition of visual and proprioceptive cues

Removing or adding sensory cues from one sensory system during standing balance causes a change in the contribution of the remaining sensory systems, a process referred to as sensory reweighting. While reweighting changes have been described in many studies under steady-state conditions, less is known about the temporal dynamics of reweighting following sudden transitions to different sensory conditions. The present study changed sensory conditions by periodically adding or removing visual (lights On/Off) or proprioceptive cues (surface sway referencing On/Off) in 12 young, healthy subjects. Evidence for changes in sensory contributions to balance was obtained by measuring the time course of medial-lateral sway responses to a constant-amplitude 0.56-Hz sinusoidal stimulus, applied as support surface tilt (proprioceptive contribution), as visual scene tilt (visual contribution), or as binaural galvanic vestibular stimulation (vestibular contribution), and by analyzing the time course of sway variability. Sine responses and variability of body sway velocity showed significant changes following transitions and were highly correlated under steady-state conditions. A dependence of steady-state responses on upcoming transitions was observed, suggesting that knowledge of impending changes can influence sensory weighting. Dynamic changes in sway in the period immediately following sensory transitions were very inhomogeneous across sway measures and in different experimental tests. In contrast to steady-state results, sway response and variability measures were not correlated with one another in the dynamic transition period. Several factors influence sway responses following addition or removal of sensory cues, partly instigated by but also obscuring the effects of reweighting dynamics.

Human upright posture control models based on multisensory inputs; in fast and slow dynamics

Posture control to maintain an upright stance is one of the most important and basic requirements in the daily life of humans. The sensory inputs involved in posture control include visual and vestibular inputs, as well as proprioceptive and tactile somatosensory inputs. These multisensory inputs are integrated to represent the body state (body schema); this is then utilized in the brain to generate the motion. Changes in the multisensory inputs result in postural alterations (fast dynamics), as well as long-term alterations in multisensory integration and posture control itself (slow dynamics). In this review, we discuss the fast and slow dynamics, with a focus on multisensory integration including an introduction of our study to investigate "internal force control" with multisensory integration-evoked posture alteration. We found that the study of the slow dynamics is lagging compared to that of fast dynamics, such that our understanding of long-term alterations is insufficient to reveal the underlying mechanisms and to propose suitable models. Additional studies investigating slow dynamics are required to expand our knowledge of this area, which would support the physical training and rehabilitation of elderly and impaired persons.

Controlling legs for locomotion-insights from robotics and neurobiology

Walking is the most common terrestrial form of locomotion in animals. Its great versatility and flexibility has led to many attempts at building walking machines with similar capabilities. The control of walking is an active research area both in neurobiology and robotics, with a large and growing body of work. This paper gives an overview of the current knowledge on the control of legged locomotion in animals and machines and attempts to give walking control researchers from biology and robotics an overview of the current knowledge in both fields. We try to summarize the knowledge on the neurobiological basis of walking control in animals, emphasizing common principles seen in different species. In a section on walking robots, we review common approaches to walking controller design with a slight emphasis on biped walking control. We show where parallels between robotic and neurobiological walking controllers exist and how robotics and biology may benefit from each other. Finally, we discuss where research in the two fields diverges and suggest ways to bridge these gaps.

Bioinspired dynamic inclination measurement using inertial sensors

Biologically, the vestibular feedback is critical to the ability of human body to balance in different conditions. This balancing ability inspires analysis of the reference equilibrium position in dynamic environments. The research proposes and experimentally validates the concept of equilibrium for the human body modeled as an inverted pendulum, which is instrumental in explaining why we align the body along the surface normal when standing on a surface but not on an incline, and tend to lean backward or forward on non-static surfaces e.g. accelerating or decelerating bus. This equilibrium position--the dynamic equilibrium axis--is dependent only on the acceleration of surface of contact (e.g. gravity) and acts as the reference to the orientation measurements. The research also draws design inspiration from the two human ears--symmetry and plurality of inertial sensors. The vestibular dynamic inclinometer and planar vestibular dynamic inclinometer consist of multiple (two or four) symmetrically placed accelerometers and a gyroscope. The sensors measure the angular acceleration and absolute orientation, not the change in orientation, from the reference equilibrium position and are successful in separating gravity from motion for objects moving on ground. The measurement algorithm is an analytical solution that is not time-recursive, independent of body dynamics and devoid of integration errors. The experimental results for the two sensor combinations validate the theoretically (kinematics) derived analytical solution of the measurement algorithm.

Visual contribution to human standing balance during support surface tilts

•

Contributions of visual position and velocity cues to standing balance are analyzed.

•

Both visual cues reduce sway responses to support surface tilt and sway variability.

•

Model simulations are used for data interpretation and data reproduction.

•

Visual cues improve disturbance estimates by reduction of estimation thresholds.

•

Reduction of noise by visual cues appears to be an instrumental factor.

Visual position and velocity cues improve human standing balance, reducing sway responses to external disturbances and sway variability. Previous work suggested that human balancing is based on sensory estimates of external disturbances and their compensation using feedback mechanisms (Disturbance Estimation and Compensation, DEC model). This study investigates the visual effects on sway responses to pseudo-random support surface tilts, assuming that improvements result from lowering the velocity threshold in a tilt estimate and the position threshold in an estimate of the gravity disturbance. Center of mass (COM) sway was measured with four different tilt amplitudes, separating the effect of visual cues across the conditions ‘Eyes closed’ (no visual cues), ‘4 Hz stroboscopic illumination’ (visual position cues), and ‘continuous illumination’ (visual position and velocity cues). In a model based approach, parameters of disturbance estimators were identified. The model reproduced experimental results and showed a specific reduction of the position and velocity threshold when adding visual position and velocity cues, respectively. Sway variability was analyzed to explore a hypothesized relation between estimator thresholds and internal noise. Results suggest that adding the visual cues reduces the contribution of vestibular noise, thereby reducing sway variability and allowing for lower thresholds, which improves the disturbance compensation.

Time-interval for integration of stabilizing haptic and visual information in subjects balancing under static and dynamic conditions

Maintaining equilibrium is basically a sensorimotor integration task. The central nervous system (CNS) continually and selectively weights and rapidly integrates sensory inputs from multiple sources, and coordinates multiple outputs. The weighting process is based on the availability and accuracy of afferent signals at a given instant, on the time-period required to process each input, and possibly on the plasticity of the relevant pathways. The likelihood that sensory inflow changes while balancing under static or dynamic conditions is high, because subjects can pass from a dark to a well-lit environment or from a tactile-guided stabilization to loss of haptic inflow. This review article presents recent data on the temporal events accompanying sensory transition, on which basic information is fragmentary. The processing time from sensory shift to reaching a new steady state includes the time to (a) subtract or integrate sensory inputs; (b) move from allocentric to egocentric reference or vice versa; and (c) adjust the calibration of motor activity in time and amplitude to the new sensory set. We present examples of processes of integration of posture-stabilizing information, and of the respective sensorimotor time-intervals while allowing or occluding vision or adding or subtracting tactile information. These intervals are short, in the order of 1–2 s for different postural conditions, modalities and deliberate or passive shift. They are just longer for haptic than visual shift, just shorter on withdrawal than on addition of stabilizing input, and on deliberate than unexpected mode. The delays are the shortest (for haptic shift) in blind subjects. Since automatic balance stabilization may be vulnerable to sensory-integration delays and to interference from concurrent cognitive tasks in patients with sensorimotor problems, insight into the processing time for balance control represents a critical step in the design of new balance- and locomotion training devices.

Stance width changes how sensory feedback is used for multisegmental balance control

A multilink sensorimotor integration model of frontal plane balance control was developed to determine how stance width influences the use of sensory feedback in healthy adults. Data used to estimate model parameters came from seven human participants who stood on a continuously rotating surface with three different stimulus amplitudes, with eyes open and closed, and at four different stance widths. Dependent variables included lower body (LB) and upper body (UB) sway quantified by frequency-response functions. Results showed that stance width had a major influence on how parameters varied across stimulus amplitude and between visual conditions. Active mechanisms dominated LB control. At narrower stances, with increasing stimulus amplitude, subjects used sensory reweighting to shift reliance from proprioceptive cues to vestibular and/or visual cues that oriented the LB more toward upright. When vision was available, subjects reduced reliance on proprioception and increased reliance on vision. At wider stances, LB control did not exhibit sensory reweighting. In the UB system, both active and passive mechanisms contributed and were dependent on stance width. UB control changed across stimulus amplitude most in wide stance (opposite of the pattern found in LB control). The strong influence of stance width on sensory integration and neural feedback control implies that rehabilitative therapies for balance disorders can target different aspects of balance control by using different stance widths. Rehabilitative strategies designed to assess or modify sensory reweighting will be most effective with the use of narrower stances, whereas wider stances present greater challenges to UB control.

Sensory reweighting dynamics in human postural control

Healthy humans control balance during stance by using an active feedback mechanism that generates corrective torque based on a combination of movement and orientation cues from visual, vestibular, and proprioceptive systems. Previous studies found that the contribution of each of these sensory systems changes depending on perturbations applied during stance and on environmental conditions. The process of adjusting the sensory contributions to balance control is referred to as sensory reweighting. To investigate the dynamics of reweighting for the sensory modalities of vision and proprioception, 14 healthy young subjects were exposed to six different combinations of continuous visual scene and platform tilt stimuli while sway responses were recorded. Stimuli consisted of two components: 1) a pseudorandom component whose amplitude periodically switched between low and high amplitudes and 2) a low-amplitude sinusoidal component whose amplitude remained constant throughout a trial. These two stimuli were mathematically independent of one another and, thus, permitted separate analyses of sway responses to the two components. For all six stimulus combinations, the sway responses to the constant-amplitude sine were influenced by the changing amplitude of the pseudorandom component in a manner consistent with sensory reweighting. Results show clear evidence of intra- and intermodality reweighting. Reweighting dynamics were asymmetric, with slower reweighting dynamics following a high-to-low transition in the pseudorandom stimulus amplitude compared with low-to-high amplitude shifts, and were also slower for inter- compared with intramodality reweighting.

Contribution of visual velocity and displacement cues to human balancing of support surface tilt

Vision helps humans in controlling bipedal stance, interacting mainly with vestibular and proprioceptive cues. This study investigates how postural compensation of support surface tilt is compromised by selectively reducing visual velocity cues by stroboscopic illumination of a stationary visual scene. Healthy adult subjects were presented with pseudorandom tilt sequences in the sagittal plane (tilt frequency range 0.017-2.2 Hz; velocity amplitude spectrum constant up to a frequency of 0.6 Hz, angular displacement amplitude spectrum increasing with decreasing frequencies). Center of mass (COM) sway responses were recorded for stroboscopic illuminations at 48, 32, 16, 8, and 4 Hz, as well as under continuous illumination and with eyes closed. With strobe duration (5 ms) and mean luminance (1 lx) kept constant, visual acuity and perceived brightness remained constant and the visual scene was perceived as stationary. Yet, tilt-evoked COM excursions increased with decreasing strobe frequency in a graded way, with largest effects occurring at tilt frequencies where large tilt velocities coincided with small displacements. In addition, COM excursions were reduced at the lowest strobe frequency compared to eyes closed, with the largest effect occurring at tilt frequencies where tilt displacements were large. We conclude that two mechanisms exist, a velocity mechanism that deals with tilt compensation and is foremost affected by the stroboscopic illumination and a displacement mechanism. This compares favorably to previous findings that, transferred to a stance control model, suggest a velocity mechanism for tilt compensation and a position mechanism for gravity compensation.

Humanoids Learning to Walk: A Natural CPG-Actor-Critic Architecture

The identification of learning mechanisms for locomotion has been the subject of much research for some time but many challenges remain. Dynamic systems theory (DST) offers a novel approach to humanoid learning through environmental interaction. Reinforcement learning (RL) has offered a promising method to adaptively link the dynamic system to the environment it interacts with via a reward-based value system. In this paper, we propose a model that integrates the above perspectives and applies it to the case of a humanoid (NAO) robot learning to walk the ability of which emerges from its value-based interaction with the environment. In the model, a simplified central pattern generator (CPG) architecture inspired by neuroscientific research and DST is integrated with an actor-critic approach to RL (cpg-actor-critic). In the cpg-actor-critic architecture, least-square-temporal-difference based learning converges to the optimal solution quickly by using natural gradient learning and balancing exploration and exploitation. Futhermore, rather than using a traditional (designer-specified) reward it uses a dynamic value function as a stability indicator that adapts to the environment. The results obtained are analyzed using a novel DST-based embodied cognition approach. Learning to walk, from this perspective, is a process of integrating levels of sensorimotor activity and value.

A model of the basal ganglia in voluntary movement and postural reactions

A basal ganglia central pattern generator (CPG) is developed and its role in voluntary movements on the ground and postural reactions on a disturbed platform are studied and analysed by simulation. Biped dynamics and platform kinematics are utilised. The effects of agonist-antagonist muscular co-activation and joint stiffness are formulated. The implementation of the necessary counter-manoeuvres for maintaining balance and postural stability is studied. A control strategy, applicable to large systems, is formulated. The biped manoeuvres and transitions terminate in pre-specified intervals of time. Gravity is included and compensated for. Certain voluntary and postural adjustment strategies are the same but are initiated differently. Further experimental/computational research may identify the central nervous system and sensory paths that lead to the CPG. All actuator forces linearly evolve in time from their original values to their terminal values. There are no central continuous feedback loops present. Monitoring and sensing, however, are ongoing. The counter-manoeuvres are based on learned human-like voluntary movements that are triggered by the disturbance. The required central inputs to the musculoskeletal system are designed in the CPG. A functional structure for the CPG is proposed. The effect of certain disorders and malfunctions of the CPG are studied by simulation.

Navigating sensory conflict in dynamic environments using adaptive state estimation

Most conventional robots rely on controlling the location of the center of pressure to maintain balance, relying mainly on foot pressure sensors for information. By contrast,humans rely on sensory data from multiple sources, including proprioceptive, visual, and vestibular sources. Several models have been developed to explain how humans reconcile information from disparate sources to form a stable sense of balance. These models may be useful for developing robots that are able to maintain dynamic balance more readily using multiple sensory sources. Since these information sources may conflict, reliance by the nervous system on any one channel can lead to ambiguity in the system state. In humans, experiments that create conflicts between different sensory channels by moving the visual field or the support surface indicate that sensory information is adaptively reweighted. Unreliable information is rapidly down-weighted,then gradually up-weighted when it becomes valid again.Human balance can also be studied by building robots that model features of human bodies and testing them under similar experimental conditions. We implement a sensory reweighting model based on an adaptive Kalman filter in abipedal robot, and subject it to sensory tests similar to those used on human subjects. Unlike other implementations of sensory reweighting in robots, our implementation includes vision, by using optic flow to calculate forward rotation using a camera (visual modality), as well as a three-axis gyro to represent the vestibular system (non-visual modality), and foot pressure sensors (proprioceptive modality). Our model estimates measurement noise in real time, which is then used to recompute the Kalman gain on each iteration, improving the ability of the robot to dynamically balance. We observe that we can duplicate many important features of postural sw ay in humans, including automatic sensory reweighting,effects, constant phase with respect to amplitude, and a temporal asymmetry in the reweighting gains.