Recognising the user’s locomotive intentions is crucial for the correct functionality of exoskeletons and active orthoses. For gait applications, extrapolating control inputs from the arm swing may be worthwhile, since arm oscillations naturally occur during human locomotion. A similar method would be unaffected by severe impairments of the lower limbs, and there is evidence suggesting enhanced results of gait rehabilitation when arms and legs exercise together. In this thesis, we propose a control algorithm to drive online a lower limb exoskeleton through the arm swing. Contrary to a previous EMG-based approach by La Scaleia et al. (2014), our algorithm exploits shoulder kinematic data to mimic “single swinging”, a natural mode of human interlimb coordination which is characterised by each arm swinging in-phase with the contralateral leg. Our proposed control architecture relies on two major modules: an Arm Observer and a Gait Generator. The Arm Observer consists of an adaptive frequency oscillator which extrapolates the frequency and phase of the arm swing by receiving online measurements of the angular shoulder position in the sagittal plane. This data is used by the Gait Generator to compute lower limb trajectories, based on regression models from a previous study by Koopman et al. (2014). We validated our controller through human-subject experiments, involving three participants walking on a treadmill with and without a lower limb exoskeleton, the Lopes II. When feed by data associated with natural walking, our adaptive frequency oscillator could very precisely replicate the arm swing frequency, stride cadence and timing of shoulder flexion peaks when walking faster than 0.5 m/s. When wearing the exoskeleton, our algorithm allowed the participants to cope with constant and variable treadmill velocities in the range of 0.5-1.25 m/s. As such, the results of this thesis show that our proposed approach can extend the applicability of arm-based control to walking speeds suitable for gait rehabilitation and assistance.

Body weight support (BWS) systems are widely used in gait rehabilitation. These systems allow patients to gradually increase training intensity and promote independent walking. Currently, there is no standardized method to assess the performance of BWS systems. This work proposes methods and metrics that can be used to assess and compare any overground, force controlled BWS systems. The RYSEN, a novel, cable-driven, three-dimensional overground BWS system, developed by Motek Medical B.V., Delft University of Technology, GTX Medical and École Polytechnique Fédérale de Lausanne, is used as a case study throughout this work. Analytical methods are proposed that allow to (1) define the non-linearity bounds of static spring systems and (2) select the most important configurations for performance evaluation using sensitivity analyses on weighted stiffness matrices. These analytical methods resulted in testing the system in one direction at a time for a limited number of configurations. Furthermore, experiments are proposed based on three modes of operation: static force tracking, sinusoidal force tracking and disturbance rejection. The input signals for these experiments are based on human walking and support strategies. A test setup, equipped with an external sensor and reciprocal mechanism, was used to objectively obtain the frequency response, tracking and estimation error of the RYSEN. The experiments were successful in characterizing the RYSEN in the chosen configurations.

This thesis presents modeling and control of novel tensegrity joint based on Twisted and Coiled Polymer Muscle (TCPM) that is biologically inspired by the human elbow. The tensegrity principle that is used to construct this joint has made it structurally compliance. Therefore, this joint can be categorized as a compliance actuator. The modeling is performed by constraining the joint movement to 2 Degree of Freedom (DoF). As a result, the joint kinematics and dynamics can be simplified and approximated as rigid body movement. To actuate the joint, the length of the TCPMs can be controlled individually by Joule heating. This joint can be seen as a MIMO system subject to multiple constraints. These constraints are imposed by the TCPM and used to prevent the joint from failure during actuation. Model Predictive Control (MPC) is designed to control the joint movement. This controller is suitable to handle a multivariable system subject to constraints. MPC controller based on linearized model and Hammerstein-Wiener model are designed and their performance on controlling the joint is compared. Finally, trajectory tracking simulation with various cases is provided to assess the controller performance.

Due to neuromuscular capacity decline, sit-to-stand, an essential daily life activity, is increasingly difficult to perform for elderly. In practice, elderly frequently use compensatory arm-strategies in sit-to-stand. However, the specific advantages of these compensatory strategies are unclear. The focus of this research is to study the influence of thigh push-off, a common compensatory strategy, on the stability of sit-to-stand for both the young and elderly. Motion data was retrieved from 50 young and elderly participants who performed sit-to-stand with and without thigh push-off. To mimic realistic daily-life strategies, participants were not restricted in sit-to-stand style in any way but arm-use. The experimental sit-to-stand data was fitted to a pendulum model with feedforward and feedback control to simulate the stability limits of the observed sit-to-stand strategies. The model allows us to explore stability limits without the need of perturbation. For each arm-strategy and participant a stability basin, i.e. all potential trajectories of motion, was formed. We compared the stability basins between arm-strategies and between age-groups. The size of the computed stability basin for thigh push-off was larger than for no arm push-off implicating thigh push-off is a more stable sit-to-stand strategy. The difference between arm strategies was larger for the elderly than for the young participants. Overall, the stability basins of the elderly were larger compared to those of the young participants, for both arm strategies. The resulting stability basins suggest that thigh push-off increases the stability of sit-to-stand and thus could indeed be an effective compensatory strategy. Stability basin shape differences indicate that without thigh push-off elderly compensate for possible early sit-down and a step halfway sit-to-stand, and with the thigh push-off strategy only for early sit-down. The age-groups differences confirm stability basins quantify the stability of the observed movement strategy rather than the stability of the participants. We observed that elderly use more precautious sit-tot-stand movements.

According to the WHO, falls are responsible for over 38 million disability-adjusted life years lost each year, globally. Additionally, an estimated 684.000 individuals die of falling each year, making it the second leading cause of unintentional death. One of the most effective physical activity interventions to reduce the risk of falling, is targeted exercise that safely challenges balance. One tool, that is currently being developed, which can aid physical therapists with these interventions is the GyBAR. A wearable device that uses gyroscopes to apply moments to the patient. This can be used to either provide balance assistance or challenge balance by applying perturbations. The goal of this research was to develop an interface for the GyBAR. The interface should contribute to the acceptance of the GyBAR. This can be achieved by an excellent perceived ease of use, which combined with perceived usefulness are indicators for the Technology Acceptance Model (TAM). The interface inspired by a Voodoo-doll was developed by going through the following design process steps; the creation of a list of requirements; a brainstorm to generate ideas; the development of three concepts; and the selection and development of one concept into a prototype. The Voodoo Doll controls the GyBAR by a handheld model of the patient, which can be manipulated and translates the movements of the model to the actual patient. User tests were performed to validate the design. With a SUS-score of 82.81 (SD=7.48), which is within the 90-95 percentile, it can be concluded that the interface has an excellent perceived ease of use. Further development is encouraged are several recommendations for points of improvement are given.

Inspired by phenomena in the plant world, a meteoro-sensitive rotational actuator is developed. The design uses a hygro-active shell, whose water-based swelling is restricted at selective locations to form a helicoid structure. The influence of geometrical parameters on the performance is investigated using a numerical analysis of various geometries, by looking at resulting rotation and torque during this rotation. Prototypes are built of five key geometries in the design space, to validate the simulations and to investigate the real-life behaviour of the design. These prototypes are submerged in water to investigate their deformation, after which they are placed in a torsion machine to investigate the torque during rotation. The experiments result in similar rotations and torques as the simulations. The designed Hygromorphic Rotational Actuator is capable of passively rotating its own structure, thereby expanding the possibilities of engineers and designers when designing passive autonomous systems.

Passivity-based control is a well-established technique for coordinating groups of fully-actuated systems, but existing methods for underactuated systems are limited to groups of homogeneous systems, coordinate synchronization tasks, and to specific applications. We propose a generic distributed control method that enables heterogeneous groups of underactuated and fully-actuated mechanical systems to cooperatively assume desired task-space formations, with or without leaders with constant task-space references. Extending the method of passivity-based control by interconnection and damping assignment (IDA-PBC) to distributed networks of mechanical systems, we derive matching conditions and control laws to achieve the desired stable group behavior. For a suitable choice of virtual coupling forces between the systems in the task space, we can decouple the matching conditions into three conditions local to each agent, independent of the topology of the undirected and connected network. If these local conditions are satisfied, we show how existing single-system IDA-PBC solutions can be used to construct distributed control laws, thereby enabling distributed control design for a large class of applications. By shaping the input-output behavior of a system in addition to shaping its total energy, we also show how human operators can interact with groups of underactuated mechanical systems using the proposed distributed control scheme. The procedure is illustrated using simulation studies of networks of unmanned aerial vehicles that can assume formations and dock with underactuated flexible-joint manipulators.

Gait event detection allows for insight into one’s gait pattern, an invaluable aid in rehabilitation. Current methods often rely on measured acceleration and rarely on position measurements [3]–[6]. In this paper we propose 4 novel gait detection methods based on the position of the Center of Mass (one approach being causal and thus suitable for real-time use) and compare them to 4 existing state-of-the-art acceleration- based methods. All algorithms are benchmarked on an existing data set (overground walking, 23 participants, 1772 steps), comparing the detection rate, false positive rate and the mean and (intra- and interparticipant) standard deviation of the timing error for Heel Strikes and Toe-Offs. We show that position-based algorithms give well-balanced results and are able to outperform the acceleration-based algorithms in all five metrics. Additionally, we propose and compare several methods for detecting left and right steps, thereby enabling quantification of the full gait cycle.

Falling remains a large source of (traumatic) injuries and healthcare costs. Over the past years, different actuators have been developed in the field of wearable robotics to help prevent injuries from falling. To increase the wearability of these systems, the weight of power storage can be decreased with intermittent instead of continuous control. A fall detector is then needed for these systems to trigger the activation of the actuator to prevent the fall. A proof of concept for a model-based fall detector that is aimed at using only wearable sensor measurements is presented. The algorithm is based on a single inertial measurement unit placed on the lower back. The upper-body orientation and centre of mass velocity are estimated with two separate Kalman filters. The velocity is estimated with a gait model consisting of a spring-damper-legged point mass in three-dimensional space. The balance of the subject is evaluated with the velocity estimates and the extrapolated centre of mass method. The presented model is verified on a non-falling treadmill walking dataset of real humans and shows accurate estimates of the centre of mass velocity. Furthermore, a planar falling simulation is performed to show successful pre-impact fall detection. This resulted in no false alarms outside the initial estimation settling time, and a successful detection with a lead time of 680 ms. This lead time is long enough to provide a trigger for fall mitigation devices.

With the decrease in sensor and actuator costs decentralized control strategies have become increasingly attractive, aiming to use multiple simpler robots for achieving a global objective. The problem of reaching the global objective generally results in a consensus problem requiring communication amongst the agents. The cooperative manipulation problem, where a payload is manipulated using multiple robots, poses an attractive alternative: By using the payload's motion as the means of communication, the agents can reach consensus without using explicit communication. The advantage being that no additional bandwidth is required as the number of participating agents increases and all to all communication is effectively achieved. Whereas previous works considered only the translation dynamics this thesis work considers the use of the full rigid body motion as a means of communication, such that the agents reach consensus on the desired wrench and the payload is stabilized at any desired configuration. As a possible application the towing of a payload by multiple UAV via cables is considered. This brings the additional challenge of underactuation from the perspective of each agent, since only forces can be used to control the full payload's motion. The result is a decentralized nonlinear control law for the forces applied to a payload such that consensus is reached amongst the agents, the leader's control action is amplified and the payload is stabilized at any desired configuration. Proofs are constructed via Lyapunov arguments and the applicability of the control design to the aerial towing problem is validated in simulation.

Part I: Inflicted head injury by shaking trauma (IHI-ST) is often simulated to better understand the injury mechanisms and to analyze whether violent shaking can cause head injury in infants. Computational models are usually subjected to linear and rotational inputs to simulate shaking scenarios. In existing studies, the head’s rotation center is kept fixed over time during shaking. However, the infant’s head is unlikely to rotate around a fixed pivoting point in real life due to the flexibility of the infant’s neck and the external imposed shaking motion by the perpetrator. It is currently unknown how the location of the rotation center changes over time and how this manifests itself in the expression of the injury mechanisms associated with IHI-ST. In this study, the variation of the rotation center of an infant’s head during shaking and its potential effect on injury mechanisms were analyzed. First, dynamics of the infant’s head were obtained in shaking experiments with an infant surrogate. Next, the variation of the rotation center was calculated and relations between characteristics of the participants and shaking variables were analyzed. Key findings: during shaking the location of the head’s rotation center varied in both anterior-posterior and vertical direction with respect to the head, causing the head’s radius of curvature to vary six orders of magnitude. Therefore, head-dynamics and injury mechanisms underlying IHI-ST are possibly simulated incorrectly when using a fixed rotation center. It remains unclear how this affects the validity of IHI-ST injury risk assessments and the injury thresholds on which these assessments are based. Future research should therefore evaluate the performance of head-dynamic simulations regarding IHI-ST. Part II: Computational model simulations are extensively used to analyze inflicted head injury by shaking trauma in infants (IHI-ST). Infant head models are usually excited by dynamic inputs, which are applied to a specific point with respect to the head. In existing studies the load application point is assumed to be fixed over time; thereby neglecting spatiotemporal variation of the rotation center during shaking. Therefore, this assumption may be inappropriate, because the location of the heads’ rotation center is in fact not constant over time during shaking. It is unknown to what extent head dynamics are correctly simulated when using a fixed rotation center, hence simulation results regarding injury thresholds and shaking trauma assessment could be invalid. In this study, loading-methods used in IHI-ST simulations were evaluated for their temporal accuracy in replicating external head-dynamics. First, a mathematical model incorporating spatiotemporal variation of the head’s rotation center was proposed. Secondly, head dynamics were calculated using the proposed mathematical model and existing model-loading methods. Finally, the calculated head dynamics were compared to a reference dataset. Key findings: in all of the 29 cases from the reference dataset, implementation of a time-varying load application point resulted in an improved temporal replication of shaking dynamics compared to existing model-loading methods. Accelerations of the head in x- and z-direction had a two and four times smaller absolute error over a typical shake cycle than any previously existing finite element model (FEM) for IHI-ST. It remains unclear how implementation of a time-varying load application point affects the dynamics of fluids and tissues inside the skull. Future research should therefore focus on re-evaluating the results of IHI-ST assessment studies and injury threshold studies employing FEM head-models.

Complex motions are generally accepted as movements which consists of at least 2 Degrees of Freedom (DoFs) actuated in a coordinated manner, which takes more than one practice session to be mastered, and that are ecologically valid (see Wulf et al. 2002 for a clear statement). The techniques to teach and re-teach (also, (re-)learn) a complex motion are thoroughly investigated by the motor learning research community since decades and up to know. There is certain agreement on the theses of practice intensity and duration as a factor for learning. However, the content of the practice sessions is still, only, partially understood. One point which is still to be confirmed is the "Part-whole transfer" paradigm. This theory promulgates the fact that motions are rather a composition where alphabet items are skillfully combined to produce a dexterous movement and that the better the "alphabet" is known, the faster the learning will be and dexterous the result. The importance of the whole practice of a full motion remains still unknown. Our hypotheses were that inclusion of the full motion is crucial for correctly learning a complex motion pattern. We hypothesized, also, that the way of re-composition of a motion would play a role in the performance after the training and the retention of the learned skill, as well as the transferability of its fundamentals to a new skill. In this thesis, these questions are investigated. For this, an experiment with 16 healthy subjects has been conducted. The subjects where pseudo-randomly assigned to one of the following groups: whole visuo-haptic coordination training, anatomical visuo-haptic coordination training, anatomical visuo-haptic training and anatomical visual-only training. The importance of the haptic coordination training (i.e. practice of the whole motion together) was investigated in the first two groups, while the coordination pattern was shown only visually in the last two groups. Additionally, the way of showing the single components (anatomically or holistically) and the type of feedback (visual-only or visuo-haptic) was investigated in the different groups designed. The results showed that the full trajectory is better learned by subjects who were shown the full motion visuo-haptically (opposed to only visually). The transfer and the consolidation of the motion pattern (also translated into a better retention) are also benefited by adding special emphasis to the single components of a motion first (Group 2). Only the training of the speed profile is benefited by more practicing the full motion visuo-haptically (i.e. group 1). All these aspects have, obviously, an impact in motivation. The groups which were more guided (i.e. visuo-haptic full motion guidance) felt less pressure, committed with less effort but got higher sense of competence and more motivation, not necessarily translating in better performance (somewhat in agreement with the Guidance Hypothesis). The results of this study contribute to the field of motor learning bringing some more insight in the mechanisms of complex motion learning and could have a transfer to the clinical practice to improve re-learning of lost motor skills (e.g. after a stroke).

Gait rehabilitation attempts to alleviate gait and balance impairments caused by spinal cord injury. Conventional gait rehabilitation often includes unloading a portion of the user’s weight with a vertical force. This vertical force enables people to practice stepping but makes forward propulsion more difficult. Users will adopt a compensatory movement strategy during rehabilitation that may not completely transfer to unsupported walking. It is hypothesized that including a small forward force during vertical unloading makes gait rehabilitation more similar to unsupported locomotion. This thesis presents and investigates a novel patent of a device that passively generates a forward to propel the user forward. The forward force is generated using the vertical oscillations that naturally occur during gait. This thesis quantifies the forward force, simulates the impact on human gait and details the construction of a prototype. The prototype is then used to validate the simulations of the device. Analysis indicates that the ratio between the sprocket radius and the wheel radius is the key parameter for determining the amount of forward force. Simulations in which the device is coupled with the spring-loaded inverted pendulum model indicate that the device may generally deteriorate the walking velocity, step size and maximum height difference. Experiments with the prototype indicate that friction has a big effect on the functionality of the device, and that predictions from the simulation may be quantitatively off. The device may have a favorable interaction with the neurological control strategy of human gait, which was not modeled. Human trials will have to give a definite answer.

The Sit-to-Stand (SitTS) and Stand-to-Sit (StandTS) transitions are among the most demanding activities of daily living in terms of required muscle strength. Many elderly struggle with the transitions, which become even harder after a transfemoral amputation. The current state of the art in knee prostheses does not supply any support to reduce the required muscle strength. A new, energy-recovering ankle-knee prosthesis was designed to provide support during SitTS and StandTS, with a natural torque profile. The prosthesis uses pneumatic cylinders to store energy during StandTS, and recover it during SitTS. To make the use of the prosthesis as intuitive as possible, the design was optimized to mimic natural torque profiles during the transition. The addition of an ankle joint enables a more posterior placement of the prosthetic limb, which should reduce the required knee torque. The prototype provides a peak extending knee torque of 21.7 Nm and a peak plantarflexing ankle torque of 10.8 Nm during SitTS, which can be further increased by increasing the pre-pressure in the system. It is concluded that the current prototype has the potential to provide support during the SitTS and StandTS transitions, but that alterations are required to improve the available joint torques. The next phase will be user tests with transfemoral amputees.

The mechanisms that move plants can serve as biological role models for engineers, designers and architects. This practice is slowly being implemented in various engineering fields, with architects often being pioneers. Various methodologies have been written about the subject, but often only from an architectural point of view. The classification presented in this paper provides a different perspective on the subject. It is structured like a toolbox, containing a clear classification of the technical working principles that plants use to generate motion. With the working principles abstracted, it is no longer necessary to dive deep into the inner workings of plants. The Scopus and Web of Science databases have been systematically searched for compliant plant movements. Plants mainly move by reallocating water, either actively via osmosis or passively via hygroscopic tissue. In compliant plant mechanisms, these basic movement initiators bring about deformations of plant parts. These movements are classified according to their goal: does the plant move quasi-static or dynamic? And does the plant only use a mechanism or does it rely on the gradual storage and fast release of elastic energy as well? Quasi-static movements are often only mechanical and reversible, while dynamic movements rely on energy storage and are often irreversible due to their failure-based release. A bilayer structure in one form or another is present in almost all mechanisms, proving its large adaptability to various circumstances. This adaptability is achieved by the various configurations of the two layers, including fibre-orientation and cellular set-up. Existing bio-inspired devices are classified according to the same system. This enables identification of plant mechanisms that are already suited for implementation and exposes plant movements that are not yet used in the human world. Additionally, a lot could be gained from copying mechanisms on a cellular level rather than on a macroscopic level. Most importantly, a change in mindset needs to happen in order to fully benefit the intricate mechanisms that the plant world has to offer.