Human-prosthesis cooperation: combining adaptive prosthesis control with visual feedback guided gait.

BACKGROUND: Personalizing prosthesis control is often structured as human-in-the-loop optimization. However, gait performance is influenced by both human control and intelligent prosthesis control. Hence, we need to consider both human and prosthesis control, and their cooperation, to achieve desired gait patterns. In this study, we developed a novel paradigm that engages human gait control via user-fed visual feedback (FB) of stance time to cooperate with automatic prosthesis control tuning. Three initial questions were studied: (1) does user control of gait timing (via visual FB) help the prosthesis tuning algorithm to converge faster? (2) in turn, does the prosthesis control influence the user's ability to reach and maintain the target stance time defined by the feedback? and (3) does the prosthesis control parameters tuned with extended stance time on prosthesis side allow the user to maintain this potentially beneficial behavior even after feedback is removed (short- and long-term retention)? METHODS: A reinforcement learning algorithm was used to achieve prosthesis control to meet normative knee kinematics in walking. A visual FB system cued the user to control prosthesis-side stance time to facilitate the prosthesis tuning goal. Seven individuals without amputation (AB) and four individuals with transfemoral amputation (TFA) walked with a powered knee prosthesis on a treadmill. Participants completed prosthesis auto-tuning with three visual feedback conditions: no FB, self-selected stance time FB (SS FB), and increased stance time FB (Inc FB). The retention of FB effects was studied by comparing the gait performance across three different prosthesis controls, tuned with different visual FB. RESULTS: (1) Human control of gait timing reduced the tuning duration in individuals without amputation, but not for individuals with TFA. (2) The change of prosthesis control did not influence users' ability to reach and maintain the visual FB goal. (3) All participants increased their prosthesis-side stance time with the feedback and maintain it right after feedback was removed. However, in the post-test, the prosthesis control parameters tuned with visual FB only supported a few participants with longer stance time and better stance time symmetry. CONCLUSIONS: The study provides novel insights on human-prosthesis interaction when cooperating in walking, which may guide the future successful adoption of this paradigm in prosthesis control personalization or human-in-the-loop optimization to improve the prosthesis user's gait performance.

Is there a trade-off between economy and task goal variability in transfemoral amputee gait?

BACKGROUND: Energy cost minimization has been widely accepted to regulate gait. Optimization principles have been frequently used to explain how individuals adapt their gait pattern. However, there have been rare attempts to account for the role of variability in this optimization process. Motor redundancy can enable individuals to perform tasks reliably while achieving energy optimization. However, we do not know how the non-goal-equivalent and goal-equivalent variability is regulated. In this study, we investigated how unilateral transfemoral amputees regulate step and stride variability based on the task to achieve energy economy. METHODS: Nine individuals with unilateral transfemoral amputation walked on a treadmill at speeds of 0.6, 0.8, 1.0, 1.2 and 1.4 m/s using their prescribed passive prostheses. We calculated the step-to-step and stride-to-stride variability and applied goal equivalent manifold (GEM) based control to decompose goal-equivalent and non-goal-equivalent manifold. To quantify the energy economy, the energy recovery rate (R) was calculated based on potential energy and kinetic energy. Comparisons were made between GEM variabilities and commonly used standard deviation measurements. A linear regression model was used to investigate the trade-off between R and GEM variabilities. RESULTS: Our analysis shows greater variability along the goal-equivalent manifold compared to the non-goal-equivalent manifold (p < 0.001). Moreover, our analysis shows lower energy recovery rate for amputee gait compared to nonamputee gait (at least 20% less at faster walking speed). We found a negative relationship between energy recovery rate and non-goal-equivalent variability. Compared to the standard deviation measurements, the variability decomposed using GEM reflected the preferred walking speed and the limitation of the passive prosthetic device. CONCLUSION: Individuals with amputation cleverly leverage task redundancy, regulating step and stride variability to the GEM. This result suggests that task redundancy enables unilateral amputees to benefit from motor variability in terms of energy economy. The differences observed between prosthetic step and intact step support the development of prosthetic limbs capable of enhancing positive work during the double support phase and of powered prosthesis controllers that allow for variability along the task space while minimizing variability that interferes with the task goal. This study provides a different perspective on amputee gait analysis and challenges the field to think differently about the role of variability.

Postural coordination and control to the precision demands of light finger touch.

We examine the proposition that information availability and postural facilitation-usually viewed as opposing views in postural control-are intertwined with the effects of one being related to the other. If that is the case, a single control parameter (precision demands) would capture the changes in postural control relating information and postural facilitation. Using the dynamical systems approach, we investigated whether, manipulating touch requirements as to increase precision demands, would induce quantitative and qualitative changes in postural dynamics. Additionally, we tested whether the COM-COP coupling reflects the qualitative dynamics of the system. Seventeen participants were instructed to maintain quiet standing while maintaining or not a light finger force with either precision or no precision. Standard deviation (SD) of the COP decreased with the precision demands and the correlation dimension (CD) of COP showed higher values for the touch conditions. Participants showed reduced synchronization of COP-COM coupling; following changes in CD. These results point out the integrated nature of information availability, task requirements, and the emergent postural organization reflected in COP-COM coupling.

Harnessing Machine Learning and Physiological Knowledge for a Novel EMG-Based Neural-Machine Interface.

OBJECTIVE: In this study, we aimed to develop a novel electromyography (EMG)-based neural machine interface (NMI), called the Neural Network-Musculoskeletal hybrid Model (N2M2), to decode continuous joint angles. Our approach combines the concepts of machine learning and musculoskeletal modeling. METHODS: We compared our novel design with a musculoskeletal model (MM) and 2 continuous EMG decoders based on artificial neural networks (ANNs): multilayer perceptrons (MLPs) and nonlinear autoregressive neural networks with exogenous inputs (NARX networks). EMG and joint kinematics data were collected from 10 non-disabled and 1 transradial amputee subject. The offline performance tested across 3 different conditions (i.e., varied arm postures, shifted electrode locations, and noise-contaminated EMG signals) and online performance for a virtual postural matching task was quantified. Finally, we implemented the N2M2 to operate a prosthetic hand and tested functional task performance. RESULTS: The N2M2 made more accurate predictions than the MLP in all postures and electrode locations (p < 0.003). For estimated MCP joint angles, the N2M2 was less sensitive to noisy EMG signals than the MM or NARX network with respect to error (p < 0.032) as well as the NARX network with respect to correlation (p = 0.007). Additionally, the N2M2 had better online task performance than the NARX network (p ≤ 0.030). CONCLUSION: Overall, we have found that combining the concepts of machine learning and musculoskeletal modeling has resulted in a more robust joint kinematics decoder than either concept individually. SIGNIFICANCE: The outcome of this study may result in a novel, highly reliable controller for powered prosthetic hands.

Investigating the Association of Quantitative Gait Stability Metrics With User Perception of Gait Interruption Due to Control Faults During Human-Prosthesis Interaction.

This study aims to compare the association of different gait stability metrics with the prosthesis users' perception of their own gait stability. Lack of perceived confidence on the device functionality can influence the gait pattern, level of daily activities, and overall quality of life for individuals with lower limb motor deficits. However, the perception of gait stability is subjective and difficult to acquire online. The quantitative gait stability metrics can be objectively measured and monitored using wearable sensors; however, objective measurements of gait stability associated with human's perception of their own gait stability has rarely been reported. By identifying quantitative measurements that associate with users' perceptions, we can gain a more accurate and comprehensive understanding of an individual's perceived functional outcomes of assistive devices such as prostheses. To achieve our research goal, experiments were conducted to artificially apply internal disturbances in the powered prosthesis while the prosthetic users performed level ground walking. We monitored and compared multiple gait stability metrics and a local measurement to the users' reported perception of their own gait stability. The results showed that the center of pressure progression in the sagittal plane and knee momentum (i.e., residual thigh and prosthesis shank angular momentum about prosthetic knee joint) can potentially estimate the users' perceptions of gait stability when experiencing disturbances. The findings of this study can help improve the development and evaluation of gait stability control algorithms in robotic prosthetic devices.

A simplified model for whole-body angular momentum calculation.

The capability to monitor gait stability during everyday life could provide key information to guide clinical intervention to patients with lower limb disabilities. Whole body angular momentum (Lbody) is a convenient stability indicator for wearable motion capture systems. However, Lbody is costly to estimate, because it requires monitoring all major body segment using expensive sensor elements. In this study, we developed a simplified rigid body model by merging connected body segments to reduce the number of body segments, which need to be monitored. We demonstrated that the Lbody could be estimated by a seven-segment model accurately for both people with and without lower extremity amputation.

Offline Evaluation Matters: Investigation of the Influence of Offline Performance of EMG-Based Neural-Machine Interfaces on User Adaptation, Cognitive Load, and Physical Efforts in a Real-Time Application.

There has been controversy about the value of offline evaluation of EMG-based neural-machine interfaces (NMIs) for their real-time application. Often, conclusions have been drawn after studying the correlation of the offline EMG decoding accuracy/error with the NMI user's real-time task performance without further considering other important human performance metrics such as adaptation rate, cognitive load, and physical effort. To fill this gap, this study aimed to investigate the relationship between the offline decoding accuracy of EMG-based NMIs and user adaptation, cognitive load, and physical effort in real-time NMI use. Twelve non-disabled subjects participated in this study. For each subject, we established three EMG decoders that yielded different offline accuracy (low, moderate, and high) in predicting continuous hand and wrist motions. The subject then used each EMG decoder to perform a virtual hand posture matching task in real time with and without a secondary task as the evaluation trials. Results showed that the high-level offline performance decoders yield the fastest adaptation rate and highest posture matching completion rate with the least muscle effort in users during online testing. A secondary task increased the cognitive load and reduced real-time virtual task competition rate for all the decoders; however, the decoder with high offline accuracy still produced the highest task completion rate. These results imply that the offline performance of EMG-based NMIs provide important insight to users' abilities to utilize them and should play an important role in research and development of novel NMI algorithms.

Community Genetics screening in a pandemic: solutions for pre-test education, informed consent, and specimen collection.

A Community Genetics carrier screening program for the Jewish community has operated on-site in high schools in Sydney (Australia) for 25 years. During 2020, in response to the COVID-19 pandemic, government-mandated social-distancing, 'lock-down' public health orders, and laboratory supply-chain shortages prevented the usual operation and delivery of the annual testing program. We describe development of three responses to overcome these challenges: (1) pivoting to online education sufficient to ensure informed consent for both genetic and genomic testing; (2) development of contactless telehealth with remote training and supervision for collecting genetic samples using buccal swabs; and (3) a novel patient and specimen identification 'GeneTrustee' protocol enabling fully identified clinical-grade specimens to be collected and DNA extracted by a research laboratory while maintaining full participant confidentiality and privacy. These telehealth strategies for education, consent, specimen collection and sample processing enabled uninterrupted delivery and operation of complex genetic testing and screening programs even amid pandemic restrictions. These tools remain available for future operation and can be adapted to other programs.

Artificial intelligence meets medical robotics.

Artificial intelligence (AI) applications in medical robots are bringing a new era to medicine. Advanced medical robots can perform diagnostic and surgical procedures, aid rehabilitation, and provide symbiotic prosthetics to replace limbs. The technology used in these devices, including computer vision, medical image analysis, haptics, navigation, precise manipulation, and machine learning (ML) , could allow autonomous robots to carry out diagnostic imaging, remote surgery, surgical subtasks, or even entire surgical procedures. Moreover, AI in rehabilitation devices and advanced prosthetics can provide individualized support, as well as improved functionality and mobility (see the figure). The combination of extraordinary advances in robotics, medicine, materials science, and computing could bring safer, more efficient, and more widely available patient care in the future. -Gemma K. Alderton.

Characterizing Prosthesis Control Fault during Human-Prosthesis Interactive Walking Using Intrinsic Sensors.

The physical interactions between wearable lower limb robots and humans have been investigated to inform effective robot design for walking augmentation. However, human-robot interactions when internal faults occur within robots have not been systematically reported, but it is essential to improve the robustness of robotic devices and ensure the user's safety. This paper aims to (1) present a methodology to characterize the behavior of the robotic transfemoral prosthesis as an effective wearable robot platform while interacting with the users in the presence of internal faults, and (2) identify the potential data sources for accurate detection of the prosthesis fault. We first obtained the human perceived response in terms of their walking stability when the prosthesis control fault (inappropriate intrinsic control output/command) was emulated/applied in level-ground walking. Then the measurements and their features, obtained from the transfemoral prosthesis, were examined for the emulated faults that elicited a sense of instability in human users. The optimal features that contributed the most in separating faulty interaction from the normal walking condition were determined using two machine-learning-based approaches: One-Class Support Vector Machine (OCSVM) and Mahalanobis Distance (MD) classifier. The OCSVM anomaly detector could achieve an average sensitivity of 85.7 % and an average false alarm rate of 1.7 % with a reasonable detecting time of 147.6 ms for detecting emulated control errors among all subjects. The result demonstrates the potential of using machine-learning-based schemes in identifying prosthesis control faults based on intrinsic sensors on the prosthesis. This study presents a procedure to study human-robot fault tolerance and inform the future design of robust prosthesis control.

Toward Safe Wearer-Prosthesis Interaction: Evaluation of Gait Stability and Human Compensation Strategy Under Faults in Robotic Transfemoral Prostheses.

Although advanced wearable robots can assist human wearers, their internal faults (i.e., sensors or control errors) also pose a challenge. To ensure safe wearer-robot interactions, how internal errors by the prosthesis limb affect the stability of the user-prosthesis system, and how users react and compensate for the instability elicited by internal errors are imperative. The goals of this study were to 1) systematically investigate the biomechanics of a wearer-robot system reacting to internal errors induced by a powered knee prosthesis (PKP), and 2) quantify the error tolerable bound that does not affect the user's gait stability. Eight non-disabled participants and two unilateral transfemoral amputees walked on a pathway wearing a PKP, as the controller randomly switched the control parameters to disturbance parameters to mimic the errors caused by locomotion mode misrecognition. The size of prosthesis control errors was systematically varied to determine the error tolerable bound that disrupted gait stability. The effect of the error was quantified based on the 1) mechanical change described by the angular impulse applied by the PKP, and 2) overall gait instability quantified using human perception, angular momentum, and compensatory stepping. The results showed that the error tolerable bound is dependent on the gait phase and the direction of torque change. Two balance recovery strategies were also observed to allow participants to successful respond to the induced errors. The outcomes of this study may assist the future design of an auto-tuning algorithm, volitionally-controlled powered prosthetic legs, and training of gait stability.

Direct Myoelectric Control Modifies Lower Limb Functional Connectivity: A Case Study.

Prostheses with direct EMG control could restore amputee's biomechanics structure and residual muscle functions by using efferent signals to drive prosthetic ankle joint movements. Because only feedforward control is restored, it is unclear 1) what neuromuscular control mechanisms are used in coordinating residual and intact muscle activities and 2) how this mechanism changes over guided training with the prosthetic ankle. To address these questions, we applied functional connectivity analysis to an individual with unilateral lower-limb amputation during postural sway task. We built functional connectivity networks of surface EMGs from eleven lower-limb muscles during three sessions to investigate the coupling among different function modules. We observed that functional network was reshaped by training and we identified a stronger connection between residual and intact below knee modules with improved bilateral symmetry after amputee acquired skills to better control the powered prosthetic ankle. The evaluation session showed that functional connectivity was largely preserved even after nine months interval. This preliminary study might inform a unique way to unveil the potential neuromechanic changes that occur after extended training with direct EMG control of a powered prosthetic ankle.

Multifractal roots of suprapostural dexterity.

Visually guided postural control emerges in response to task constraints. Task constraints generate physiological fluctuations that foster the exploration of available sensory information at many scales. Temporally correlated fluctuations quantified using fractal and multifractal metrics have been shown to carry perceptual information across the body. The risk of temporally correlated fluctuations is that stable sway appears to depend on a healthy balance of standard deviation (SD): too much or too little SD entails destabilization of posture. This study presses on the visual guidance of posture by prompting participants to quietly stand and fixate at distances within, less than, and beyond comfortable viewing distance. Manipulations of the visual precision demands associated with fixating nearer and farther than comfortable viewing distance reveals an adaptive relationship between SD and temporal correlations in postural fluctuations. Changing the viewing distance of the fixation target shows that increases in temporal correlations and SD predict subsequent reductions in each other. These findings indicate that the balance of SD within stable bounds may depend on a tendency for temporal correlations to self-correct across time. Notably, these relationships became stronger with greater distance from the most comfortable viewing and reaching distance, suggesting that this self-correcting relationship allows the visual layout to press the postural system into a poise for engaging with objects and events. Incorporating multifractal analysis showed that all effects attributable to monofractal evidence were better attributed to multifractal evidence of nonlinear interactions across scales. These results offer a glimpse of how current nonlinear dynamical models of self-correction may play out in biological goal-oriented behavior. We interpret these findings as part of the growing evidence that multifractal nonlinearity is a modeling strategy that resonates strongly with ecological-psychological approaches to perception and action.

Visual effort moderates postural cascade dynamics.

Standing still and focusing on a visible target in front of us is a preamble to many coordinated behaviors (e.g., reaching an object). Hiding behind its apparent simplicity is a deep layering of texture at many scales. The task of standing still laces together activities at multiple scales: from ensuring that a few photoreceptors on the retina cover the target in the visual field on an extremely fine scale to synergies spanning the limbs and joints at smaller scales to the mechanical layout of the ground underfoot and optic flow in the visual field on the coarser scales. Here, we used multiscale probability density function (PDF) analysis to show that postural fluctuations exhibit similar statistical signatures of cascade dynamics as found in fluid flow. In participants asked to stand quietly, the oculomotor strain of visually fixating at different distances moderated postural cascade dynamics. Visually fixating at a comfortable viewing distance elicited posture with a similar cascade dynamics as posture with eyes closed. Greater viewing distances known to stabilize posture showed more diminished cascade dynamics. In contrast, nearest and farthest viewing distances requiring greater oculomotor strain to focus on targets elicited a dramatic strengthening of postural cascade dynamics, reflecting active postural adjustments. Critically, these findings suggest that vision stabilizes posture by reconfiguring the prestressed poise that prepares the body to interact with different spatial layouts.

Y2 and Y4 receptor signaling synergistically act on energy expenditure and physical activity.

Neuropeptide Y receptors are critical regulators of energy homeostasis and are well known for their powerful influence on feeding, but their roles in other important aspects of energy homeostasis, such as energy expenditure and their functional interactions in these processes, are largely unknown. Here we show that mice lacking both Y2 and Y4 receptors exhibited a reduction in adiposity, more prominent in intra-abdominal vs. subcutaneous fat, and an increase in lean mass as determined by dual-energy X-ray absorptiometry. These changes were more pronounced than those seen in mice with Y2 or Y4 receptor single deletion, demonstrating the important roles and synergy of Y2 and Y4 signaling in the regulation of body composition. These changes in body composition occurred without significant changes in food intake, but energy expenditure and physical activity were significantly increased in Y4(-/-) and particularly in Y2(-/-)Y4(-/-) but not in Y2(-/-) mice, suggesting a critical role of Y4 signaling and synergistic interactions with Y2 signaling in the regulation of energy expenditure and physical activity. Y2(-/-) and Y4(-/-) mice also exhibited a decrease in respiratory exchange ratio with no further synergistic decrease in Y2(-/-)Y4(-/-) mice, suggesting that Y2 and Y4 signaling each play important and independent roles in the regulation of substrate utilization. The synergy between Y2 and Y4 signaling in regulating fat mass may be related to differences in mitochondrial oxidative capacity, since Y2(-/-)Y4(-/-) but not Y2(-/-) or Y4(-/-) mice showed significant increases in muscle protein levels of peroxisome proliferator-activated receptor (PPAR)γ coactivator (PGC)-1α, and mitochondrial respiratory chain complexes I and III. Taken together, this work demonstrates the critical roles of Y2 and Y4 receptors in the regulation of body composition and energy metabolism, highlighting dual antagonism of Y2 and Y4 receptors as a potentially effective anti-obesity treatment.

Perceiving amputee gait from biological motion: kinematics cues and effect of experience level.

Physical therapists (PT) and clinicians must be skilled in identifying gait features through observation to assess motor deficits in patients and intervene appropriately. Inconsistent results in the literature have led researchers to question how clinical experience influences PT's gait perception and to seek the key kinematic features that should be trained to enhance PT's skill. Thus, this study investigated (1) what are the informative kinematic features that allow gait-deviation perception in amputee gait and (2) whether there are differences in observational gait skills between PT and individuals with less clinical experience (PT students [PTS] and Novices). We introduced a new method that combines biological motion and principal component analysis to gradually mesh amputee and typical walking patterns. Our analysis showed that on average the accuracy rate in identifying gait deviations between PT and PTS was similar and better than Novices. Also, we found that PT's experience was demonstrated by their better perception of gait asymmetry. The extracted principal components demonstrated that the major gait deviation of amputees was the medial-lateral body sway and spatial gait asymmetry.

Human-prosthesis coordination: A preliminary study exploring coordination with a powered ankle-foot prosthesis.

BACKGROUND: Powered ankle-foot prostheses were developed to replicate the mechanics of the biological ankle by providing positive work during the push-off phase of gait. However, the benefits of powered prostheses on improving overall human gait efficiency (usually quantified by metabolic cost) have not been consistently shown. Here, we have focused on the mechanical work produced at the prosthetic ankle and its interaction with the amputee's movement. METHODS: Five unilateral transtibial amputees walked on a treadmill using 1) a powered ankle-foot prosthesis and 2) their daily passive device. We determined the net ankle work and ankle work loops on the prosthesis-side to quantify the efficiency of the human-prosthesis physical interaction. We further studied peak propulsion timing and the posture of the amputee's lower limb and prosthesis as indicators of the human-prosthesis coordination. Comparisons were made between the passive and powered prosthesis conditions for each participant. FINDINGS: The powered prosthesis did not consistently increase net ankle work compared to each participant's passive device. For participants that lacked efficiency in interacting with the powered prosthesis, we observed 1) early prosthesis-side peak propulsion timing (≥ 4% earlier) and 2) a more vertical residual shank at the time of peak propulsion (> 2° more vertical) indicating that the human's limb movement and the prosthesis control during push-off were not well coordinated. INTERPRETATION: Results from this preliminary study highlight the need for future work to systematically quantify the coordination between the human and powered prosthesis and understand how such coordination at the joint level influences overall gait efficiency.

Identify Kinematic Features for Powered Prosthesis Tuning.

To maximize the benefits of the newly developed powered prosthetic legs, amputees must rely on tuning experts (TE) from manufacturers to tune these devices based on their specific physical conditions. Because TEs are hard to train, it is difficult to access the TEs and the cost of customization is high. If the knowledge used by the TEs could be extracted, it is possible to reduce the tuning cost by automating the tuning procedure or developing efficient TE training programs. In this paper, we preliminarily identified kinematic features that are sensitive to the control parameter change of the powered prosthetic leg. Using data collected from three transtibial amputee subjects with four levels of push-off power, we tested whether a change of push-off power could generate a significant difference on 13 preselected kinematic features during level ground walking at self-selected walking speed. Six features across three joints on the prosthesis side were demonstrated to be sensitive to the change of push-off power.

The precision demands of viewing distance modulate postural coordination and control.

There are contrasting views on the role of vision in modifying postural organization (information-driven and postural facilitation) and limited direct tests of the underlying postural mechanisms. Here, we examined whether the distinction between the two views is appropriate given that both are interrelated parts of task constraints modulating postural coordination and control. The study investigated whether changes in the organization of the postural system are a function of the visual precision demands of a task and, in addition, whether such organization could be described as reflecting an intermittent controller. Sixteen participants were instructed to maintain quiet postural stance while fixating a point at different viewing distances (25, 50, 135, 220, 305 cm) or standing with eyes closed. The 25-cm condition showed the lowest standard deviation of the center of pressure (COP) and the highest correlation dimension (CD) in the anterior posterior direction. Analyses revealed that, contrary to the intermittent controller hypothesis, adaptations in the continuous COP and center of mass (COM) coupling characterized the observed changes in CD. The findings show that the natural act of looking to the same feature in the environment as a function of visual viewing distance can lead to quantitative and qualitative changes in the dynamics of posture. This is consistent with the view that postural facilitation and information availability are integrated in the perceptual-motor dynamics.

Gaze Fixation Comparisons Between Amputees and Able-bodied Individuals in Approaching Stairs and Level-ground Transitions: A Pilot Study.

This paper aims to investigate the visual strategy of transtibial amputees while they are approaching the transition between level-ground and stairs and compare it with that of able-bodied individuals. To this end, we conducted a pilot study where two transtibial amputee subjects and two able-bodied subjects transitioned from level-ground to stairs and vice versa while wearing eye tracking glasses to record gaze fixations. To investigate how vision functioned to both populations for preparing locomotion on new terrains, gaze fixation behavior before the new terrains were analyzed and compared between two populations across all transition cases in the study. Our results presented that, unlike the able-bodied population, amputees had most of their fixations directed on the transition region prior to new terrains. Furthermore, amputees showed an increased need for visual information during transition regions before navigation on stairs than that before navigation onto level-ground. The insights about amputees' visual behavior gained by the study may lead the future development of technologies related to the intention prediction and the locomotion recognition for amputees.