Adaptive motor control and seamless coordination of muscle actions in response to external perturbations are crucial to maintaining balance during bipedal locomotion. There is an ongoing debate about the specific roles of individual muscles and underlying neural control circuitry that humans employ to maintain balance in different perturbation scenarios. To advance our understanding of human motor control in perturbation recovery, we conducted a study using a portable Angular Momentum Perturbator (AMP). Unlike other push/pull perturbation systems, the AMP can generate perturbation torques on the upper body while minimizing the perturbing forces at the center of mass. In this study, ten participants experienced trunk perturbations during either the mid-stance or touchdown phase in two frontal plane directions (ipsilateral and contralateral). We recorded and analyzed the electromyography (EMG) activity of eight lower-limb muscles from both legs to examine muscular responses in different phases and directions. Based on our findings, individuals primarily employ long-latency hip strategies to effectively counteract perturbation torques, with the occasional use of ankle strategies. Furthermore, it was found that proximal muscles, particularly the biarticular Rectus Femoris, consistently exhibited higher activation levels than other muscles. Additionally, in instances where a statistically significant difference was noted, we observed that the fastest reactions generally stem from muscles in close proximity to the perturbation site. However, the temporal sequence of muscles’ activation depends on the timing and direction of the perturbation. These findings enhance reflex response modeling, aiding the development of simulation tools for accurately predicting exogenous disturbances. Additionally, they hold the potential to shape the development of assistive devices, with implications for clinical interventions, particularly for the elderly.

Wind energy is one of the main renewable energy sources in the current energy transition. Due to ever more and ever larger wind turbines (WT), the requirements for WT operation become more complex. Model predictive control (MPC) for WTs shows the potential to handle these requirements and conflicting control objectives in a single optimization-based controller. Recent research has widely investigated MPC for WT in simulation, but mostly lacks experimental validation. This work aims to experimentally validate MPC on a full-scale WT under real conditions. To this end, we combine an extended Kalman filter for nonlinear state estimation with robust linear time-varying MPC. We evaluate the proposed control algorithm in terms of time-domain performance and power curve in simulation. However, the main contribution of this work is the experimental validation on a 3MW WT in Northern Germany with a total duration of 3-h continuous full access of the controller. We were able to demonstrate stable operation of the proposed MPC in the upper partial load regime, transition regime, and lower full load regime, at measured wind speeds between 4.76 and 13.06 m/s, inside and outside the wake shadow of another WT. The power curve determined in simulation shows comparable results to a reference feedback controller. The MPC formulation combines several control objectives in a single optimization problem, yet the tuning effort still remains complex. In future work, we plan to reduce the complexity of the control loop based on this experimentally validated MPC.

Manual tuning of performance-critical controller parameters can be tedious and sub-optimal. Bayesian Optimization (BO) is an increasingly popular practical alternative to automatically optimize controller parameters from few experiments. Standard BO practice is to evaluate the closed-loop performance of parameters proposed during optimization on an episode with a fixed length. However, fixed-length episodes can be wasteful. For example, continuing an episode where already the start shows undesirable behavior such as strong oscillations seems pointless. Therefore, we propose a BO method that stops an episode early if suboptimality becomes apparent before an episode is completed. Such early stopping results in partial observations of the controller’s performance, which cannot directly be included in standard BO. We propose three heuristics to facilitate partially observed episodes in BO. Through five numerical and one hardware experiment, we demonstrate that early stopping BO can substantially reduce the time needed for optimization.

Traditional wheelchairs are pushed from behind the occupant, which hinders eye contact and communication. It was proposed that the wheelchair be pushed from the side using a push bar to place the caregiver beside the occupant. However, this method requires the caregiver to exert continuous effort to maintain a straight trajectory because the force at the lateral push location generates a moment. This study explores simple modifications to the front castor wheel of the wheelchair that allow pushing it from the side without additional effort. We used a three-dimensional dynamic model of the castor wheel to predict the effects of altering its dimensions and rake, cant, and bank angle and present a simplified steady-state solution. Experimental results support the model’s predictions, and a proof-of-concept experiment with a wheelchair showed that it is possible to push a wheelchair from the side without increasing forces or moments. The results also confirmed that the lateral ground reaction force generated on the castor wheel is proportional to the normal force and the cant angle, which can counter the moment caused by the lateral push location. The implications of this model extend beyond wheelchair design and can be applied to other applications that use castor wheels, such as robotics, trolleys, and walkers.

People fall more often when their gait stability is reduced. Gait stability can be directly manipulated by exerting forces or moments onto a person, ranging from simple walking sticks to complex wearable robotics. A systematic review of the literature was performed to determine: What is the level of evidence for different types of mechanical manipulations on improving gait stability? The study was registered at PROSPERO (CRD42020180631). Databases Embase, Medline All, Web of Science Core Collection, Cochrane Central Register of Controlled Trials, and Google Scholar were searched. The final search was conducted on the 1st of December, 2022. The included studies contained mechanical devices that influence gait stability for both impaired and non-impaired subjects. Studies performed with prosthetic devices, passive orthoses, and analysing post-training effects were excluded. An adapted NIH quality assessment tool was used to assess the study quality and risk of bias. Studies were grouped based on the type of device, point of application, and direction of forces and moments. For each device type, a best-evidence synthesis was performed to quantify the level of evidence based on the type of validity of the reported outcome measures and the study quality assessment score. Impaired and non-impaired study participants were considered separately. From a total of 4701 papers, 53 were included in our analysis. For impaired subjects, indicative evidence was found for medio-lateral pelvis stabilisation for improving gait stability, while limited evidence was found for hip joint assistance and canes. For non-impaired subjects, moderate evidence was found for medio-lateral pelvis stabilisation and limited evidence for body weight support. For all other device types, either indicative or insufficient evidence was found for improving gait stability. Our findings also highlight the lack of consensus on outcome measures amongst studies of devices focused on manipulating gait.

Maintaining balance during human walking hinges on the exquisite orchestration of whole-body angular momentum (WBAM). This study delves into the regulation of WBAM during gait by examining balance strategies in response to upper-body moment perturbations in the frontal plane. A portable Angular Momentum Perturbator (AMP) was utilized in this work, capable of generating perturbation torques on the upper body while minimizing the impact on the center of mass (CoM) excursions. Ten participants underwent upper-body perturbations during either the mid-stance or touch-down moment in both ipsilateral and contralateral directions in the frontal plane. Our findings emphasize the predominant role of the hip strategy and foot placement as primary mechanisms for recovering from WBAM perturbations, regardless of the perturbation’s timing or direction. Specifically, hip add/abduction torque and step width were significantly modulated following perturbations during the stance and swing phases, respectively, to reject frontal-plane balance threats. The knee and ankle torque modulation were not found to be effective in the recovery process. Additionally, we observed that recovery from WBAM perturbations occurs promptly within the same stride in which the perturbation occurs, unlike other perturbation scenarios, such as platform translation. These insights have the potential to enhance the development of assistive devices and more robust controllers for bipedal robots.

Accurate and robust vehicle localization in highly urbanized areas is challenging. Sensors are often corrupted in those complicated and large-scale environments. This article introduces gnssFGO, a global and online trajectory estimator that fuses global navigation satellite systems (GNSS) observations alongside multiple sensor measurements for robust vehicle localization. In gnssFGO, we fuse asynchronous sensor measurements into the graph with a continuous-time trajectory representation using Gaussian process (GP) regression. This enables querying states at arbitrary timestamps without strict state and measurement synchronization. Thus, the proposed method presents a generalized factor graph for multisensor fusion. To evaluate and study different GNSS fusion strategies, we fuse GNSS measurements in loose and tight coupling with a speed sensor, inertial measurement unit, and LiDAR-odometry. We employed datasets from measurement campaigns in Aachen, Düsseldorf, and Cologne and presented comprehensive discussions on sensor observations, smoother types, and hyperparameter tuning. Our results show that the proposed approach enables robust trajectory estimation in dense urban areas where a classic multisensor fusion method fails due to sensor degradation. In a test sequence containing a 17-km route through Aachen, the proposed method results in a mean 2-D positioning error 0.48 m while fusing raw GNSS observations with LiDAR odometry in a tight coupling.

Angular momentum, kinetics, and energetics, including total mechanical energy and its rate of change in relation to power exchange, are important quantities when analyzing human motion in sports, physical labor, and rehabilitation. Inertial measurement units (IMU)-based motion capture (MOCAP) systems provide a portable solution for the ambulatory analysis of these quantities which optical MOCAP systems do not offer. Yet, evaluating IMU-based estimates of these quantities by referencing optical systems is limited by the fact that these systems only measure positions, not kinetic and energetic quantities. To evaluate the accuracy of an IMU-based method for estimating kinetic and energetic quantities without using any external reference, firstly, we propose an estimation method only using angular velocity and acceleration signals supplied by an IMU, and apply this to a single rigid body with known mass and inertia. Then, we propose a novel experimental validation method against physical conservation and action/reaction laws that apply during ballistic movements, using a suitably designed and reconfigurable rigid body with a structure of three orthogonal dumb-bells. The results demonstrated that we could estimate the angular momentum, kinetics, and energetics of a rigid body by only using angular velocity and acceleration signals of an IMU, and the estimation accuracy was well evaluated by the proposed validation method. However, the results showed that the errors in original IMU measurements under dynamic conditions especially concerning angular velocity, uncertainties in calculating rigid body parameters, and vibration propagation due to limited rigidity of tubes of the rigid body influenced the estimation accuracy.

Background: Pneumatic actuators are widely used in applications like (medical) robots, or prostheses. Pneumatic actuators require a complex manufacturing process and are produced in standardized dimensions to reduce costs. Over the last decade 3D-printing has emerged as a cost-effective and efficient production method in medical applications. 3D-printing can also function as a cost-efficient alternative production method for pneumatic actuators. Objective: The goal of this research is to study the possibility of creating a pneumatic linear actuator with 3D-printing. Furthermore, the aim is to use the advantage of 3D-printing to create pneumatic actuators with non-circular cross-sections. Methodology: To evaluate the performance of a 3D-printed pneumatic actuator, a test setup was designed and built to measure the leakage and sliding friction force. Furthermore, two pneumatic actuators with a non-conventional cross-sectional shape were designed and their performance was tested and compared with a 3D-printed cylindrical pneumatic actuator, since these tests only ran once, the results are more a guideline. During the manufacturing of the cylinders, no post-processing techniques were used. Results: The functioning of a 3D-printed circular pneumatic actuator was proven with low static leakage rates of 2.5%, low dynamic leakage rates of approximately 1%, and a maximum friction force of [Formula presented]. Furthermore, the results show that it is possible to print functioning pneumatic cylinders with a non-cylindrical concave cross-section. The non-conventional cylinders were tested up to [Formula presented] with maximum dynamic leakage of [Formula presented]. Conclusion: This study demonstrates a method to create functional pneumatic linear actuators with 3D-printing. It was possible to create 3D-printed actuators with a conventional shape, e.g. circular and unconventional shapes e.g. stadium/oval shape and a kidney shape. The leak rates for conventional and unconventional shapes were in the same range. This opens up the world for more design freedom in pneumatic actuators.

Balance recovery after tripping often requires an active adaptation of foot placement. Thus far, few attempts have been made to actively assist forward foot placement for balance recovery employing wearable devices. This study aims to explore the possibilities of active forward foot placement through two paradigms of actuation: assistive moments exerted with the reaction moments either internal or external to the human body, namely 'joint' moments and 'free' moments, respectively. Both paradigms can be applied to manipulate the motion of segments of the body (e.g., the shank or thigh), but joint actuators also exert opposing reaction moments on neighbouring body segments, altering posture and potentially inhibiting tripping recovery. We therefore hypothesised that a free moment paradigm is more effective in assisting balance recovery following tripping. The simulation software SCONE was used to simulate gait and tripping over various ground-fixed obstacles during the early swing phase. To aid forward foot placement, joint moments and free moments were applied either on the thigh to augment hip flexion or on the shank to augment knee extension. Two realizations of joint moments on the hip were simulated, with the reaction moment applied to either the pelvis or the contralateral thigh. The simulation results show that assisting hip flexion with either actuation paradigm on the thigh can result in full recovery of gait with a margin of stability and leg kinematics closely matching the unperturbed case. However, when assisting knee extension with moments on the shank, free moment effectively assist balance but joint moments with the reaction moment on the thigh do not. For joint moments assisting hip flexion, placement of the reaction moment on the contralateral thigh was more effective in achieving the desired limb dynamics than placing the reaction on the pelvis. Poor choice of placement of reaction moments may therefore have detrimental consequences for balance recovery, and removing them entirely (i.e., free moment) could be a more effective and reliable alternative. These results challenge conventional assumptions and may inform the design and development of a new generation of minimalistic wearable devices to promote balance during gait.

Climate-controlled cabins have for decades been standard in vehicles. Model Predictive Controllers (MPCs) have shown promising results in achieving temperature tracking in vehicle cabins and may improve upon model-free control performance. However, for the multi-zone climate control case, proper controller tuning is challenging, as externally, e.g., passenger-triggered changes in compressor setting and thus mass flow lead to degraded control performance. This paper presents a tuning method to automatically determine robust MPC parameters, as a function of the blower mass flow. Constrained contextual Bayesian optimization (BO) is used to derive policies minimizing a high-level cost function subject to constraints in a defined scenario. The proposed method leverages random disturbances and model-plant mismatch within the training episodes to generate controller parameters achieving robust disturbance rejection. The method contains a postprocessing step to achieve smooth policies that can be utilized in real-world applications. First, simulation results show that the mass flow-dependent policy outperforms a constant parametrization, while achieving the desired closed-loop behavior. Second, the robust tuning method greatly reduces worst-case overshoot and produces consistent closed-loop behavior under varying operating conditions.

Three-dimensional (3D) cameras used for gait assessment obviate the need for bodily markers or sensors, making them particularly interesting for clinical applications. Due to their limited field of view, their application has predominantly focused on evaluating gait patterns within short walking distances. However, assessment of gait consistency requires testing over a longer walking distance. The aim of this study is to validate the accuracy for gait assessment of a previously developed method that determines walking spatiotemporal parameters and kinematics measured with a 3D camera mounted on a mobile robot base (ROBOGait). Walking parameters measured with this system were compared with measurements with Xsens IMUs. The experiments were performed on a non-linear corridor of approximately 50 m, resembling the environment of a conventional rehabilitation facility. Eleven individuals exhibiting normal motor function were recruited to walk and to simulate gait patterns representative of common neurological conditions: Cerebral Palsy, Multiple Sclerosis, and Cerebellar Ataxia. Generalized estimating equations were used to determine statistical differences between the measurement systems and between walking conditions. When comparing walking parameters between paired measures of the systems, significant differences were found for eight out of 18 descriptors: range of motion (ROM) of trunk and pelvis tilt, maximum knee flexion in loading response, knee position at toe-off, stride length, step time, cadence; and stance duration. When analyzing how ROBOGait can distinguish simulated pathological gait from physiological gait, a mean accuracy of 70.4%, a sensitivity of 49.3%, and a specificity of 74.4% were found when compared with the Xsens system. The most important gait abnormalities related to the clinical conditions were successfully detected by ROBOGait. The descriptors that best distinguished simulated pathological walking from normal walking in both systems were step width and stride length. This study underscores the promising potential of 3D cameras and encourages exploring their use in clinical gait analysis.

Standing up using one leg is a challenging task for those with a transfemoral amputation, particularly for elderly users with a low activity level. Active prostheses are generally not accessible to this group and available passive prostheses do not support standing up. This article presents the design and evaluation of the “Energy Restoring Intelligent Knee” (ERiK), which stores energy during sit-down in a pneumatic cylinder and returns it during stand-up. We hypothesized that the system would reduce the time needed to perform transitions and also enable higher load sharing by the prosthetic leg. However, the results of an experimental study with seven participants with transfemoral amputation contradict these hypotheses: the participants could neither move faster nor make more use of the prosthetic leg to share their body weight during transitions. We observed that a major obstacle to the useful functionality of the leg was the absence of ankle dorsiflexion – the foot tended to slip during stand-up initiation, such that only low pre-pressures and therefore support levels could be set. The rather binary action of the pneumatics also complicated movement initiation. The lessons learned from this study may be helpful to those seeking to create better designs in the future.

The global navigation satellite systems (GNSS) play a vital role in transport systems for accurate and consistent vehicle localization. However, GNSS observations can be distorted due to multipath effects and non-line-of-sight (NLOS) receptions in challenging environments such as urban canyons. In such cases, traditional methods to classify and exclude faulty GNSS observations may fail, leading to unreliable state estimation and unsafe system operations. This work proposes a deep-learning-based method to detect NLOS receptions and predict GNSS pseudorange errors by analyzing GNSS observations as a spatio-temporal modeling problem. Compared to previous works, we construct a transformer-like attention mechanism to enhance the long short-term memory (LSTM) networks, improving model performance and generalization. For the training and evaluation of the proposed network, we used labeled datasets from the cities of Hong Kong and Aachen. We also introduce a dataset generation process to label the GNSS observations using lidar maps. In experimental studies, we compare the proposed network with a deep-learning-based model and classical machine-learning models. Furthermore, we conduct ablation studies of our network components and integrate the NLOS detection with data out-of-distribution in a state estimator. As a result, our network presents improved precision and recall ratios compared to other models. Additionally, we show that the proposed method avoids trajectory divergence in real-world vehicle localization by classifying and excluding NLOS observations.

Falling is a major cause of morbidity, and is often caused by a decrease in postural stability. A key component of postural stability is whole-body centroidal angular momentum, which can be influenced by control moment gyroscopes. In this proof-of-concept study, we explore the influence of our wearable robotic gyroscopic actuator “GyroPack” on the balance performance and gait characteristics of non-impaired individuals (seven female/eight male, 30 ± 7 years, 68.8 ± 8.4 kg). Participants performed a series of balance and walking tasks with and without wearing the GyroPack. The device displayed various control modes, which were hypothesised to positively, negatively, or neutrally impact postural control. When configured as a damper, the GyroPack increased mediolateral standing time and walking distance, on a balance beam, and decreased trunk angular velocity variability, while walking on a treadmill. When configured as a negative damper, both peak trunk angular rate and trunk angular velocity variability increased during treadmill walking. This exploratory study shows that gyroscopic actuators can influence balance and gait kinematics. Our results mirror the findings of our earlier studies; though, with more than 50% mass reduction of the device, practical and clinical applicability now appears within reach.

Regaining the ability to walk overground, to climb stairs and to perform other functional tasks such as standing up and sitting down are important rehabilitation goals following neurological injury or disease. However, these activities are often difficult to practice safely for patients with severe impairments due to the risk of injury, not only to the patient but also to therapists. The emergence of various technologies that provide a degree of body weight support can play a role in rehabilitation focused on recovering overground gait and balance functions. These can greatly reduce the risk of falls and thus allow more intense and longer training sessions. Therefore, the systems empower individuals with the ability to practice the types of activities and functions they need in order to return home and to be reintegrated into the community as much as possible. This chapter explores the origin of body weight supported devices and considers which groups could derive benefit from the training. An overview of the main training platforms available today—which comprise both robotic and non-robotic technologies—is then provided, followed by a discussion regarding outcomes of the devices thus far and possible future directions of the technology.

Background: Falls are a common complication experienced after a stroke and can cause serious detriments to physical health and social mobility, necessitating a dire need for intervention. Among recent advancements, wearable airbag technology has been designed to detect and mitigate fall impact. However, these devices have not been designed nor validated for the stroke population and thus, may inadequately detect falls in individuals with stroke-related motor impairments. To address this gap, we investigated whether population-specific training data and modeling parameters are required to pre-detect falls in a chronic stroke population. Methods: We collected data from a wearable airbag’s inertial measurement units (IMUs) from individuals with (n = 20 stroke) and without (n = 15 control) history of stroke while performing a series of falls (842 falls total) and non-falls (961 non-falls total) in a laboratory setting. A leave-one-subject-out crossvalidation was used to compare the performance of two identical machine learned models (adaptive boosting classifier) trained on cohort-dependent data (control or stroke) to pre-detect falls in the stroke cohort. Results: The average performance of the model trained on stroke data (recall = 0.905, precision = 0.900) had statistically significantly better recall (P = 0.0035) than the model trained on control data (recall = 0.800, precision = 0.944), while precision was not statistically significantly different. Stratifying models trained on specific fall types revealed differences in pre-detecting anterior–posterior (AP) falls (stroke-trained model’s F₁-score was 35% higher, P = 0.019). Using activities of daily living as non-falls training data (compared to near-falls) significantly increased the AUC (Area under the receiver operating characteristic) for classifying AP falls for both models (P < 0.04). Preliminary analysis suggests that users with more severe stroke impairments benefit further from a stroke-trained model. The optimal lead time (time interval pre-impact to detect falls) differed between control- and stroke-trained models. Conclusions: These results demonstrate the importance of population sensitivity, non-falls data, and optimal lead time for machine learned pre-impact fall detection specific to stroke. Existing fall mitigation technologies should be challenged to include data of neurologically impaired individuals in model development to adequately detect falls in other high fall risk populations. Trial registrationhttps://clinicaltrials.gov/ct2/show/NCT05076565; Unique Identifier: NCT05076565. Retrospectively registered on 13 October 2021

Trunk motor control is essential for the proper functioning of the upper extremities and is an important predictor of gait capacity in children with delayed development. Early diagnosis and intervention could increase the trunk motor capabilities in later life, but current tools used to assess the level of trunk motor control are largely subjective and many lack the sensitivity to accurately monitor development and the effects of therapy. Inertial measurement units could yield an objective quantitative assessment that is inexpensive and easy-to-implement. We hypothesized that root mean square of jerk, a proxy for movement smoothness, could be used to distinguish age and thereby presumed motor development. We attached a sensor to the trunks of six young children with no known developmental deficits. Root mean square of jerk decreases with age, up to 24 months, and is correlated to a more established method, i.e., center-of-pressure velocity, as well as other standard inertial measurement unit outputs. This metric therefore shows potential as a method to differentiate trunk motor control levels.

Gait analysis has traditionally been carried out in a laboratory environment using expensive equipment, but, recently, reliable, affordable, and wearable sensors have enabled integration into clinical applications as well as use during activities of daily living. Real-time gait analysis is key to the development of gait rehabilitation techniques and assistive devices such as neuroprostheses. This article presents a systematic review of wearable sensors and techniques used in real-time gait analysis, and their application to pathological gait. From four major scientific databases, we identified 1262 articles of which 113 were analyzed in full-text. We found that heel strike and toe off are the most sought-after gait events. Inertial measurement units (IMU) are the most widely used wearable sensors and the shank and foot are the preferred placements. Insole pressure sensors are the most common sensors for ground-truth validation for IMU-based gait detection. Rule-based techniques relying on threshold or peak detection are the most widely used gait detection method. The heterogeneity of evaluation criteria prevented quantitative performance comparison of all methods. Although most studies predicted that the proposed methods would work on pathological gait, less than one third were validated on such data. Clinical applications of gait detection algorithms were considered, and we recommend a combination of IMU and rule-based methods as an optimal solution

Sensory-motor impairments due to age or neurological diseases can influence a person's ability to maintain balance, and increase the risk of falls. Recently, wearable Control Moment Gyroscopes (CMGs) have proven to provide effective balance support. Here, we show a new design of a Series-Elastic Control Moment Gyroscope (SECMG) enhanced by an additional passive degree of freedom, namely a second, orthogonal gimbal that is supported by a (visco)elastic element. The design mainly aims to reject disturbances originating from human movement and render a low remaining impedance, as well as to provide more accurate torque sensing, based on angular deflection of the compliant element. Evaluation of the torque tracking performance with regards to a classic rigid Single-Gimbal Control Moment Gyroscope (SGCMG) showed that the device equally exceeds the bandwidth requirements for its application in human augmentation. However, characterization of our current compliant construction also revealed some backlash occluding the torque-deflection relation. In the future, the SECMG could be evaluated in experiments with humans, to validate its predicted low remaining impedance.