Andrej Olenšek

A novel robot for imposing perturbations during overground walking: mechanism, control and normative stepping responses

BackgroundThe most common approach to studying dynamic balance during walking is by applying perturbations. Previous studies that investigated dynamic balance responses predominantly focused on applying perturbations in frontal plane while walking on treadmill. The goal of our work was to develop balance assessment robot (BAR) that can be used during overground walking and to assess normative balance responses to perturbations in transversal plane in a group of neurologically healthy individuals.MethodsBAR provides three passive degrees of freedom (DoF) and three actuated DoF in pelvis that are admittance-controlled in such a way that the natural movement of pelvis is not significantly affected. In this study BAR was used to assess normative balance responses in neurologically healthy individuals by applying linear perturbations in frontal and sagittal planes and angular perturbations in transversal plane of pelvis. One way repeated measure ANOVA was used to statistically evaluate the effect of selected perturbations on stepping responses.ResultsStandard deviations of assessed responses were similar in unperturbed and perturbed walking. Perturbations in frontal direction evoked substantial pelvis displacement and caused statistically significant effect on step length, step width and step time. Likewise, perturbations in sagittal plane also caused statistically significant effect on step length, step width and step time but with less explicit impact on pelvis movement in frontal plane. On the other hand, except from substantial pelvis rotation angular perturbations did not have substantial effect on pelvis movement in frontal and sagittal planes while statistically significant effect was noted only in step length and step width after perturbation in clockwise direction.ConclusionsResults indicate that the proposed device can repeatedly reproduce similar experimental conditions. Results also suggest that “stepping strategy” is the dominant strategy for coping with perturbations in frontal plane, perturbations in sagittal plane are to greater extent handled by “ankle strategy” while angular perturbations in transversal plane do not pose substantial challenge for balance. Results also show that specific perturbation in general elicits responses that extend also to other planes of movement that are not directly associated with plane of perturbation as well as to spatio temporal parameters of gait.

Two-level control strategy of an eight link biped walking model

Abstract This paper presents an adaptive two-level control strategy for a biped walking model and demonstrates its performance in a wide range of walking modes with considerably diverse model and control parameter settings. Proposed control strategy inherits a push off that resembles considerably to forceful extension of the trailing leg during push off in human locomotion and represents a very important source of forward propulsion. Extensive simulations have shown that adjustments in the push off related parameter on higher between-step control level after each step enable evolution of various walking modes of the biped walker at selected walking speeds and distinctive gait patterns. It also allows us to investigate the changes in gait kinematics and kinetics of the biped walking model due to changes in gait velocity, torso inclination and propulsion distribution profiles.

Measuring motor actions and psychophysiology for task difficulty estimation in human-robot interaction

In this paper, we describe a method for estimating task difficulty in human-robot interaction using a combination of motor actions and psychophysiology. A number of variables are calculated from kinematics, dynamics, heart rate, skin conductance, respiration and skin temperature. Discriminant analysis of the variables is used to determine whether the user finds the task too easy or too hard. The discriminant function is recursively updated with Kalman filtering in order to better adapt to the current user. The method was tested offline in a task with 20 subjects. In cross-validation, nonadaptive discriminant analysis yielded a classification accuracy of 80.2% while adaptive discriminant analysis yielded a classification accuracy of 84.3%.

Psychophysiological responses to robot training in different recovery phases after stroke

Psychophysiological responses have become a valuable tool in human-robot interaction since they provide an objective estimate of the users psychological state. Unfortunately, their usefulness in rehabilitation robotics is uncertain since they are influenced by both physical activity and pathological conditions such as stroke. We performed psychophysiological measurements in subacute and chronic stroke patients as well as healthy controls during a reaching and grasping exercise task performed in a multimodal virtual environment. Furthermore, we evaluated the differences in kinematic and static parameters between the three groups of subjects. The results of the observed kinematic and static evaluation parameters showed significant differences when different assistive modes enabled the subject to focus on a particular function of the exercise, like reaching or grasping, or coordinated actions that combine reaching and grasping, reflecting the motor abilities of the individual. The analysis of psychophysiological responses suggests that both chronic and subacute stroke subjects have weaker psychophysiological responses than healthy subjects, though the responses of chronic patients have recovered somewhat. This certainly indicates that further studies are needed before psychophysiological responses can be used in clinical practice.

Challenges in biocooperative rehabilitation robotics

Psychological states such as mood, motivation and engagement are known to be critical for the success of rehabilitation, and encouraging unmotivated stroke patients improves the likelihood of their eventual recovery. Psychological factors can be incorporated into the closed-loop control of biocooperative rehabilitation systems, augmenting the device with critical information about the patient state. However, in rehabilitation robotics, interpretation of psychophysiological measurements is made complex by the multi-task environment, the presence of strenuous physical activity and patients damage to the central and autonomic nervous systems. The study examines these challenges and proposes possible solutions for implementation in biocooperative control of rehabilitation robots.

River multimodal scenario for rehabilitation robotics

This paper presents the novel “River” multimodal rehabilitation robotics scenario that includes video, audio and haptic modalities. Elements contributing to intrinsic motivation are carefully joined in the three modalities to increase motivation of the user. The user first needs to perform a motor action, then receives a cognitive challenge that is solved with adequate motor activity. Audio includes environmental sounds, music and spoken instructions or encouraging statements. Sounds and music were classified according to the arousal-valence space. The haptic modality can provide catching, grasping, tunnel or adaptive assistance, all depending on the users needs. The scenario was evaluated in 16 stroke users, who responded to it favourably according to the Intrinsic Motivation Inventory questionnaire. Additionally, the river multimodal environment seems to elicit higher motivation than a simpler apple pick-and-place multimodal task.

The comparison of stepping responses following perturbations applied to pelvis during overground and treadmill walking

BACKGROUND Treadmills are used frequently in rehabilitation enabling neurologically impaired subjects to train walking while being assisted by therapists. Numerous studies compared walking on treadmill and overground for unperturbed but not also perturbed conditions. OBJECTIVE The objective of this study was to compare stepping responses (step length, step width and step time) during overground and treadmill walking in a group of healthy subjects where balance assessment robots applied perturbing pushes to the subjects pelvis in sagittal and frontal planes. METHODS During walking in both balance assessment robots (overground and treadmill-based) with applied perturbations the stepping responses of a group of seven healthy subjects were assessed with a motion tracking camera. RESULTS The results show high degree of similarity of stepping responses between overground and treadmill walking for all perturbation directions. Both devices reproduced similar experimental conditions with relatively small standard deviations in the unperturbed walking as well as in perturbed walking. CONCLUSIONS Based on these results we may conclude that stepping responses following perturbations can be studied on an instrumented treadmill where ground reaction forces can be readily assessed which is not the case during perturbed overground walking.

An effective balancing response to lateral perturbations at pelvis level during slow walking requires control in all three planes of motion

In this study we investigated balancing responses to lateral perturbations during slow walking (0.85m/s). A group of seven healthy individuals walked on an instrumented treadmill while being perturbed at the level of waist at left heel strike in outward and inward lateral directions. Centre of mass (COM) and centre of pressure (COP), rotation of pelvis around vertical axis, step lengths, step widths and step times were assessed. The results have shown that beside control of COP in lateral direction, facilitated by adequate step widths, control of COP in sagittal direction, slowing down movement of COM was present after commencement of lateral perturbations. Sagittal component of COM was significantly retarded as compared to unperturbed walking for both inward (4.32±1.29cm) and outward (9.75±2.17cm) perturbations. This was necessary since after an inward perturbation first step length (0.29±0.04m compared to 0.52±0.02m in unperturbed walking) and step time (0.45±0.05s compared to 0.61±0.04s in unperturbed walking) were shortened while after an outward perturbation first two step lengths (0.36±0.05m and 0.32±0.11m compared to 0.52±0.03m in unperturbed walking) were shortened that needed to be accommodated by the described modulation of COP in sagittal plane. In addition pronounced pelvis rotation assisted in bringing swing leg to new location. The results of this study show that counteracting lateral perturbations at slow walking requires adequate response in all three planes of motion.

Kinematics of turning during walking over ground and on a rotating treadmill

BackgroundAfter neurological injury, gait rehabilitation typically focuses on task oriented training with many repetitions of a particular movement. Modern rehabilitation devices, including treadmills, augment gait rehabilitation. However, they typically provide gait training only in the forward direction of walking, hence the mechanisms associated with changing direction during turning are not practiced. A regular treadmill extended with the addition of rotation around the vertical axis is a simple device that may enable the practice of turning during walking. The objective of this study was to investigate to what extent pelvis and torso rotations in the transversal plane, as well as stride lengths while walking on the proposed rotating treadmill, resemble those in over ground turning.MethodsTen neurologically and orthopedically intact subjects participated in the study. We recorded pelvis and torso rotations in the transversal plane and the stride lengths during over ground turning and while walking on a rotating treadmill in four experimental conditions of turning. The similarity between pelvis and torso rotations in over ground turning and pair-matching walking on the rotating treadmill was assessed using intra-class correlation coefficient (ICC - two-way mixed single measure model). Finally, left and right stride lengths in over ground turning as well as while walking on the rotating treadmill were compared using a paired t-test for each experimental condition.ResultsAn agreement analysis showed average ICC ranging between 0.9405 and 0.9806 for pelvis and torso rotation trajectories respectively, across all experimental conditions and directions of turning. The results of the paired t-tests comparing left and right stride lengths showed that the stride of the outer leg was longer than the stride of the inner leg during over ground turning as well as when walking on the rotating treadmill. In all experimental conditions these differences were statistically significant.ConclusionsIn this study we found that pelvis rotation and torso rotation are similar when turning over ground as compared to walking on a rotating treadmill. Additionally, in both modes of turning, we found that the stride length of the outer leg is significantly longer than the stride length of the inner leg.

Further Steps Toward More Human-like Passive Bipedal Walking Robots

This paper describes two-legged mechanical structure with ankle, knee and hip joints that is capable of walking in naturally stable walking cycle on downhill slope without presence of energy source. Bipedal model has elastic strings attached to the designated places that imitate a part of human muscular-skeletal system. By combining spring-like properties with passive walking principles we accomplish energy exchanges between biped segments during stance and swing phases that also simultaneously facilitate solution to foot clearance problem during swing phase by enabling the biped to perform rocking movement in lateral plane. We designed and examined behavior of the biped when walking with extended and flexible knees. The observed joint trajectories in the configuration with flexed knees were very similar to those observed in normal human walking. The results show new perspectives on future bipedal robots development and for studying involvement of particular muscle in human gait.