The automotive industry is foreseeing a future where driver and car will form a synergy, by allowing the car to predict the driver’s intent. Nissan has already successfully predicted the intent to make lane changes in a driving simulator with just six electroencephalography (EEG) electrodes on the movement related cortical potential using Machine Learning. This thesis assesses if the prediction can be improved with EEG layouts containing more electrodes. A large data set was recorded with a similar lane change paradigm in a driving simulator as Nissan’s original study. Seven predefined EEG layouts were assessed: including 4, 9, 18, 25, 32, 48 and 64 electrodes. The linear discriminant analysis classifier was used offline to classify two classes: driving straight or steering. Secondly, to assess if a different selection of electrodes could have performed better, partial least squares (PLS) regression was used to select electrodes based on importance. Lastly, it was also assessed if the classifier could predict the direction of lane change as well as intention, in other words, if the data set consisted of three classes: steering left, steering right and driving straight. There was a statistical difference between the performance of the seven layouts with two classes, but from Layout 3 (with 18 electrodes) and up this difference was no longer significant. The selection of electrodes based on PLS did not prove to be better than the predefined layouts. When the data is divided into three classes, for the prediction of direction, similar results were seen: there was a statistical difference between the layouts and Layout 3 proved to be sufficient. Performance improves with more electrodes, but this effect plateaus. Layout 3 contained a satisfactory amount of electrodes, namely 18, to predict lane change intent. Moreover, the direction of the intended lane change can also be predicted sufficiently with this layout.

Of all reported cases, muscular pain in the arms is the second main cause of work-related musculoskeletal disorders (WMSDs) in European workers. The most popular prevention method for shoulder-related WMSDs are passive shoulder exoskeletons, as of their light weight and ease of use. Yet, due to their static force-angle characteristic, passive shoulder exoskeletons show a limited load reduction effect for dynamic tasks. Active shoulder exoskeletons show better performance in the reduction of loads for dynamic tasks, but are not accepted in industrial environments due to their high weight and complexity. In this study, the advantages of both passive and active shoulder exoskeletons have been combined in the design of a mechanism that can actively alter the force-angle characteristic of the Skelex 360-XFR passive shoulder exoskeleton, increasing its load reduction performance, and hence making it a more effective prevention method to WMSDs. By experiments using a prototype, it has been shown that a drivable parallelogram linkage capable of inducing angular changes between an in-going lever arm link and an out-going link to the user’s arm, can effectively induce proportional phase changes of the conventional force-angle characteristic, and that the model predicting the optimal stiffness of the spring used in the implemented force-balancing mechanism is reliable. With the experimental findings, recommendations have been made regarding the minimum capacity in output torque of the actuator, 0.38 Nm, and the stiffness of the balancing spring, 1752 N/m, to end up with optimal weight, size, and power consumption of the semi-passive shoulder exoskeleton.

This paper investigates the effect of intrinsic muscle stiffness on neural control parameters in biological musculoskeletal control of stabilisation or reaching tasks. Current model implementations of intrinsic muscle properties are highly simplified, limiting their accuracy in replicating experimental short-range stiffness (SRS) behaviour, which appears to be important for stabilisation tasks. The Hill model, often used in musculoskeletal simulations, cannot account for SRS, while the Huxley model, which can account for non-linear muscle phenomena such as SRS , has a higher computational burden. The study compares a simplified Huxley-type model to two Hill-type models and determines the effect of intrinsic SRS on the control parameters of stabilizing 1- and 2-Degree of Freedom musculoskeletal models over various positive and negative stiffness positions in the force-length curve. Furthermore, the effect of the intrinsic muscle stiffness on the robustness of the feedback parameters of simple individual muscle feedback systems is determined in reaching experiments similar to classic experiments. The study finds that the Huxley model shows positive SRS in the negative flank of the force-length curve, achieves stabilisation through only co-contraction using a lower level of required muscle excitation than both Hill-type models and stabilises both musculoskeletal systems at a larger muscle range than the Hill-type models, including in the negative stiffness flank. The feedback parameters dominantly responsible for muscle activation patterns are also more robust to change in the Huxley model. These findings suggest that intrinsic muscle stiffness impacts neural control parameters in stabilisation and reaching tasks, and further musculoskeletal modelling should consider using more complex muscle stiffness calculations for improved accuracy.

The human brain, with its intricate web of billions of neurons and trillions of synaptic connections, is a remarkable organ responsible for performing complex cognitive processes. While brain imaging techniques like fMRI and EEG provide insights into neural activity, there is no broadly accepted mathematical framework for the collaborative activity of neuronal populations and their communication. In the interdisciplinary fields of neuroscience and computational modelling, this research introduces a data-driven mathematical framework for modelling the brain's cortical network, derived from EEG data, while incorporating an exogenous input. This framework captures brain dynamics, forming a state-space model using subspace identification, and evaluates statistical dependencies among brain sources, known as functional connectivity. Employed on a dataset mimicking a sensorimotor task, it proves effective in characterizing brain dynamics, with the PO-MOESP method outperforming N4SID. Additionally, the framework is competent to assess functional connectivity between brain sources, resulting in a network connectivity diagram that offers valuable insights into the statistical relationships between distinct brain regions. Altogether, it is concluded the designed algorithm can determine the intricate interactions within the brain during a simple passively performed task stimulating the brain. Consequently, this achievement forms a robust foundation for advancing Brain-Computer Interfaces (BCIs) and enhancing the diagnosis of neurological disorders.

Bimanual coordination is essential for the performance of daily activities, but the underlying motor control mechanisms are not yet fully understood. The goal of the present study is to identify the contribution of contralateral responses in the wrist joints to the performance of a bimanual task. Contralateral responses could possibly be used in rehabilitation therapy to activate hand functions that are affected by a neuromuscular medical condition. In our experiment, participants had to balance a tray in a virtual environment, while either the left or right hand was perturbed. Two identical robotic wrist manipulators intermittently applied force perturbations in flexion or extension direction. Following a perturbation, contralateral responses were present and operated towards stabilization of the tray, for example by allowing an overall faster correction of the perturbation. Notably, flexion perturbations resulted in much larger contralateral responses than extension perturbations. Contralateral responses occurred mainly in the time window for voluntary responses. Results were consistent with our hypothesis for discrete bimanual movements based on optimal feedback control theory: when both hands share one goal, functional coupling occurs in the wrist joints.

Feedback error learning (FEL) is a classical computational model that describes human motor learning. It consists of forward and inverse models representing internal dynamics and environmental disturbances. Such models can be used as controllers that represent the function of the motor cortex. On top of FEL, a model has been built with jointly trained feedforward and feedback controllers using a neural network. The controllers that actuated six muscles driving a two-degrees-of-freedom arm model were trained offline. This model successfully simulated human learning of point-to-point reaching movements in a horizontal plane when it was tasked to adapt itself in a null field (NF) and a velocity-dependent force field (VF). In this study, we further tested this model in the divergent force fields (DF) and a channelled force field (CF) to observe its performance. The comparison between the simulation results and the experimental evidence suggests that this model can predict some of the key features of the learning process, such as the kinematics, the muscle dynamics, and the impedance profiles. The learning decay was hindered when the lateral error was artificially eliminated by the CF, as reported in the literature. Overall, the model could converge towards realistic

Stroke patients can have spastic paresis of the lower leg, impeding an ankle which hinders gait. A novel orthosis has been developed which counteracts this impediment to the ankle. It is expected that gait training will improve stroke patients' use of the orthosis by increasing their ankle dynamics. Gait training with biofeedback, which is based on physiological signal, has been shown to be effective for stroke patients. The main design requirement for the biofeedback is that it facilitates learning of an increased active range of motion of the ankle. To fulfill this requirement the biofeedback is based on the maximum angle in plantar- and dorsiflexion during the swing phase of the impeded leg. In this research, the biofeedback is validated on healthy participants with an impeded right ankle performing five gait trials. First an unimpeded reference trial was conducted capturing normal gait. After which the participant's ankle was impeded. During the initial impeded trial the participant got accustomed to the impediment during gait. Then two trials with feedback were conducted followed by a retention trial without feedback. During the retention trial the effects the biofeedback has on the ankle dynamics are determined. The outcome measures were chosen to validate whether the biofeedback facilitated learning of an increased active range of motion of the ankle. The outcome measures were the increase of active range of motion of the ankle from the initial impeded trial to the retention trial and the error quotient. The error quotient is a measure showing to what extent the angles making up the active range of motion during a trial were the same as during the reference trial. The active range of motion of the ankle of participants increased (p < 0.001) from the initial impeded trial to the retention trial. Moreover a significant decrease in the error quotient of participants was found between the initial impeded trial and the first feedback trial (p = 0.033), second feedback trial (p = 0.013) and retention trial (p = 0.020). Therefore, the biofeedback facilitated learning of an increased active range of motion of the ankle to participants. Further research is required to determine how to best adapt the biofeedback such that it is suitable for use by stroke patients in daily life.

Humans continuously adapt to new sensorimotor environments, wherein we create motor commands to execute movements properly, such as picking up the telephone or riding a bike. Generalization of motor commands means that a learned movement, such as grasping a bottle, transfers to similar but new environments, such as grasping a carton of milk. However, it remains unclear to what extent we are able to learn an uncertain variable environment and to what extent these learned movements transfer to a new environment. We compared generalization to a new environment after learning a fixed (F) or trial-by-trial variable (V) velocity-dependent clockwise force field. Selecting the damping coefficients from a predefined distribution on a trial-by-trial basis introduced the variability for the variable group during the training trials, while the fixed group had a fixed damping coefficient during training. After training, both groups made reaching movements in a new force field that was fixed over all trials. Due to the uncertainty in the force field strengths, we hypothesized that the variable group would have more difficulties with adapting compared to the fixed group. However, because the variable group would experience more different force fields during the training, we hypothesized that they would perform better in the new environment. Based on the force field compensation factors, the adaptation to the training fields was: 79.8% (F) and 79.7% (V), p = 0.982. The generalization learning was similar for both groups: 78.8% (F) and 78.9%, p = 0.986. Interestingly, the trajectory measurements showed that the variable group was less accurate than the fixed group in the training fields, but more accurate in the generalization field. Therefore, our results show that variable force field perturbations can be learned and generalized in a similar manner as fixed perturbations and that the variable group, despite a more inaccurate performance in the training block, reached more accurately than the fixed group in the new environment.

Biologically inspired neural networks are a promising approach to understand the causes and improve the treatments of brain damage. Parkinson's disease is a progressive nervous system disorder that affects mainly movements, speech and cognitive problems. It symptoms cannot be cured, though medications can significantly improve the condition. Among the symptoms, tremor is the only one which remains unaffected by medications and is only responsive to deep-brain stimulation. A simplified, cortico-thalamo-cerebellar model will be simulated with spiking neural networks to evaluate the disease effects under dopamine depletion and connectivity weight changes. Confirming previous findings, striatal dopamine depletion was not found to cause tremor, nor its injection to affect tremor severity. The model showed evidence that parkinsonian weight changes in the pallidal inner feedback loop (GPi-GPe) are responsible of creating a suitable environment for the PD tremor oscillations to rise in the thalamus. Furthermore, both the GPi and the GPe present enhanced maximal activity coherent with muscular co-contraction onsets showing evidence of abnormal basal ganglia firing during re-emergent tremor. These findings may connect abnormal basal ganglia activity to the main parkinsonian motor impairments and may help explaining the beneficial effects of deep-brain stimulation on tremor severity.

Half of the long-termed disabled stroke survivors experience increased hyper-resistance of the wrist. Discrimination between the two components of joint hyper-resistance, i.e. the neural reflexive and intrinsic tissue component, is important since the components require a different treatment method. To discriminate between the two components, objective methods are developed that make use of bio-mechanical modeling. This research aimed to address the agreement between a clinically easy applicable modeling method, the NeuroFlexor method, and a more comprehensive optimization method, which both objectively obtain the neural and intrinsic components of joint hyper-resistance. Furthermore, this research study addressed the agreement between the neural and intrinsic components obtained with the NeuroFlexor and ptimization method, and the external validation of the two components with clinical rating scales. Method NeuroFlexor based assessments and instrumented positional wrist perturbations were applied to chronic stroke survivors (n = 49) and healthy volunteers (n = 11). The neural and intrinsic components were estimated using the NeuroFlexor method, whose method is a force-relationship method, and a nonlinear electromyography driven wrist optimization model. The Modied Ashworth scale (MAS) was rated to all stroke survivors as clinical scale. Correlation analysis was conducted to nd the agreement between the components of both methods, and to nd the agreement between the MAS and the components. To analyze how well the neural and intrinsic components of both methods were able to predict the MAS, multiple regression analysis with a backward selection procedure was conducted for both methods separately and both methods together. On the healthy subjects, the optimization model was applied on the NeuroFlexor data to check differences in model structure. Results The neural components of both the NeuroFlexor method and optimization method had a strong correlation (r = 0.656), as well as the intrinsic components (r = 0.648). For both methods, the neural and intrinsic components were signicant estimators of the MAS, and the NeuroFlexor method and the optimization method were approximately equivalently able to predict the MAS (r^2 = 0.466 and r^2 = 0.519, respectively). For the multiple regression analysis with the intrinsic and neural component of both methods together, the neural component of the optimization method together with the intrinsic component of the NeuroFlexor method were more able to describe the MAS (r^2 = 0.605) than the two components of the methods separately. For healthy patients, the optimization model was not able to reliably estimate the two components from the NeuroFlexor data. Conclusion This study found evidence to support the use of the NeuroFlexor device for quantication of the neural and intrinsic components of wrist hyper-resistance post stroke. Further research is needed to establish the validity of the neural component of the NeuroFlexor and the intrinsic component of the optimization method.

During a brain tumour resection, a neurosurgeon is constantly navigating a delicate balance between resecting as much of the tumour as possible, while avoiding any damage to healthy brain tissue. This challenge is particularly difficult when the tumour is located in a critical functional area, involved in for example language or motor function. For these types of tumours, the awake craniotomy was developed. During this surgery the patient wakes up to perform language and motor tasks, to enable the surgeon to localize these functions inside the brain. In this thesis, we investigate and develop a new quantitative method to monitoring motor function that could potentially improve intraoperative decision making and enables neuroscientific and neurosurgical research. Chapter 1 provides a background about surgical strategies and technologies that have been developed to aid surgeons’ decisions during complex brain tumour resections. We will explain the complexity of robust research in the neurosurgical environment and the need for a dedicated Research Operating Room to create an environment to improve neurosurgical and neuroscientific research. In Chapter 2 we make an overview of the possible solutions to quantify motor function before, during and after awake craniotomies and discuss the best solution for the Erasmus MC. In Chapter 3 we present a new frame to create a standardized environment inside the operating room for good quality data collection of patient functionality. To design this frame, we identified and interviewed all the important stakeholders and designed three prototypes. The two most promising prototypes were developed. The final prototype was implemented during three awake craniotomies. This newly developed frame was used in Chapter 4 to explore video tracking as a new tool to quantify hand motor function. Three patients were followed one day prior to the surgery, during the awake craniotomy, and one day postoperatively. During these three cases, we identified several prerequisites for a reliable recording set-up and explored the potential to detect clinically relevant events during fingertapping and direct electrical stimulation (DES). This showed promising results and underscores the potential for video tracking to be further investigated for quantification of hand motor function. In Chapter 5 we put the discussed work into context, discussing it’s clinical and scientific relevance and future perspectives. In this thesis, we have demonstrated that it is possible to implement a new quantitative measurement method to monitor hand function in the challenging environment of an operating room. Quantification of visual observations has shown to be low-cost, easily available and implementable in clinical context, because of the fast technological advancements in this field. Video tracking can be used for future research to investigate the relation between intraoperative findings and long-term outcomes, and has the potential to add valuable information for neurosurgical and neuroscientific research.

There is a distinction between environmental and social interactions as humans interpret the world. Environmental interactions are based on weighted sensory estimates formed on integrated redundant information. Trust is a heuristic used in social interactions. As social interactions and technology interrelate, social factors become influences on the environmental estimates. Hence trust could influence our sensory estimates. The optimal sensory weights for the sensory estimates are determined with a Maximum Likelihood Estimate based on the reliability of the sensory information. The objective of this research is to find evidence that trust influences sensory weighting on the level of sensory integration, believing that human-machine interaction will benefit from a better understanding of sensory weighting. We hypothesised that information on the reliabilities of the sensory cues, while not changing these, will cause participants to reweigh the sensory information. Reweighing sensory information when the reliability of the sensory cue has not changed contradicts the Maximum Likelihood Estimation prediction. An experiment was conducted in which participants performed a target-hitting task with a haptic device. The target was hidden, yet participants were presented with a visual and haptic cue to deduct the location of the target. Although both cues were not of equal reliability, the participants created the optimal sensory estimate by sensory integration. The results showed that trust influenced sensory weighting. Since the outcome of the Maximum Likelihood Estimation model was challenged by the occurrence of sensory reweighting because of trust, it is recommended to extend the model with a factor trust.

Autonomous driving is a development that has gained a lot of attention lately, because it can lead to major improvements in the mobility sector. One example of a research project that aims to develop vehicles that are capable of reaching the highest level of autonomy in driving, is the WEpods project. The goal of this research is in line with this aim, having the thesis objective defined as follows: let the WEpods continue driving in autonomous mode more often than is currently the case. The WEpod shuttles are not yet completely able to drive autonomously due to their inability to handle unexpected behavior (terminology: faults). Currently, such faults need to be detected and solved by a steward, who will manually initiate a safe stop if necessary. The localization module, which is responsible for localizing the vehicle on a map, sometimes generates unreliable location estimates. This poses two challenges. First, the fact that there is a mismatch between reality and the sensor outcomes of the localization module that needs to be detected. Second, the question of how to prevent the system from showing behavior that is different from what is desired (terminology: failure) in case such a fault is present (terminology: fault tolerant contol). Fault tolerant control can be performed in either a passive or an active manner. The passive approach ensures that either the faults are prevented or the system is able to mitigate them by anticipation in the design. The approach evolves from passive to active fault tolerant control when an on-line adaptation of the system control is made. For applications in autonomous driving, it is apparent that it is important to handle not only anticipated faults, but also to be able to deal with unexpected faults in an on-line manner. This on-line fault tolerant control approach involves two fault diagnosis steps that lead to solving the first challenge: detection and isolation. A so-called model-based fault diagnosis approach turned out to be most suitable, as it has been used for similar applications in the past. However, a model-based fault diagnosis approach has not yet been implemented for detecting and isolating faults in a localization module of autonomous driving, indicating the scientific relevance of this research. In the model-based approach, kinematic and dynamic equations of the research vehicle (WEpod) are used to build a computational model. This model is then subjected to an observer, that is able to compare the model outcomes with the actual measurements in an off-line way. A residual is drawn up by taking the difference between the model outcomes and the measurements. A threshold is computed based on noise on the measurements to compare the residual with. When the residual exceeds the threshold, an alarm is raised. This way, the system itself has been enabled to detect faults when they occur internally. Inclusion of the suggested fault diagnosis approach in an on-line manner into the system is a big step towards fully autonomous driving of the WEpods, and therefore the goal of this research is met.

Parkinson’s Disease (PD), Essential tremor (ET), and dystonia are movement disorders often misdiagnosed as one another and commonly present tremor as one of their motor symptoms. Rates of misdiagnosis between 30 and 50% of ET patients have been reported, where dystonia and PD are the most common missed diagnoses. Additionally, up to 50% of dystonia cases are misdiagnosed/under-diagnosed at their first encounter. Misdiagnosis rates up to 34% are reported for PD. Although similar tremor behaviors between the mentioned disorders lead to substantial misdiagnosis rates and, consequently, subpar care, tremorous signal acquired via wearable sensors can be used to discriminate between PD, ET, and dystonia patients. This study aims to develop three proofs-of-concept, accelerometer-based algorithms to assist medical doctors with decision-making in unclear diagnostic scenarios.

Musculoskeletal models often use Hill-type models to study and simulate muscle behaviour. Due to fast simulation time and ability to simulate large and slow movements Hill-type models have remained largely unchanged throughout recent years. Large and slow movements spend a large part in steady state behaviour and thus experience limited influence of transitional behaviour. However, during small and fast movements, transitional behaviour has more influence and causes inaccuracies in Hill-type models, which can cause an overestimation of muscle force. One characteristic of transitional behaviour is short-range stiffness (SRS). This property is a result of crossbridge dynamics and causes an increase in stiffness when muscle velocity changes. Huxley-type models are capable of simulating transitional behaviour, but are computationally expensive. The goal of this article is to identify the optimal parameter values for two Hill- and one Huxley-type model using a surrogate optimization algorithm and determine if these models can simulate general behaviour and SRS. The parameters are fitted to one soleus and medial gastrocnemius muscle of a cat. All models were able to simulate the experimental data with an average RMS of 6.6N and 4.8N for the Hill models and 5.5N for the Huxley model. However, all three models were not capable of predicting SRS during isometric contractions and the method of determining SRS during non-isometric contractions proved unusable. Thus, the Huxley model that was used had no advantage over the used Hill-type muscle models. Furthermore, it was concluded that the simplest Hill was the only model viable for real-time application.

System identification techniques to analyse movement disorders are in development, but concrete clinical evidence to receive broad support from the clinical world is lacking. This master thesis proposes two ways to accelerate their development as part of the master’s programmes in Mechanical Engineering and Science Communication. The Mechanical Engineering part focuses on the development of system identification techniques themselves. For this study, an ensemble-based algorithm is developed to identify intrinsic and reflexive joint stiffness during voluntary movements. The algorithm combines two methods from previous research: the Parallel-Cascade method to separate intrinsic and reflexive stiffness and an instrumental variable approach to allow subjects to carry out voluntary movements and apply closed-loop identification. A simulation and experimental study are conducted with the human wrist joint in which a sinusoidal angle reference trajectory is followed using multiple realisations of the same trial. To elicit the modulation of intrinsic parameters, a torque field is applied that is linearly dependent on the wrist angle with a greater flexion resulting in a higher torque. In the simulation study, the intrinsic and reflexive pathways are separated and over 98% of the variance is explained. In the experimental study, the modulation of intrinsic stiffness is estimated with an explained variance of over 90%. Intrinsic stiffness is highest during the peaks of the trajectory and lower during the transition phase from one peak to another. However, the torque response of the experimental study shows that no significant extra torque is added during the reflexive EMG peak. The obtained results underline the promising potential of the algorithm and open up new possibilities once the algorithm is tested on an experimental task where more reflexes are elicited. The Science Communication part focuses on developing a shared mental model to transfer knowledge on the processes behind cross-disciplinary collaborations to mechanical engineers, as devoting time and effort to the metacognitive process helps in coping with it. However, most theories on cross-disciplinarity are written from a social sciences paradigm and are inherently difficult to interpret from a mechanical engineering paradigm. An analogy is developed to link concepts from both epistemologies together. This is done by assembling a theoretical framework containing concepts on the processes behind cross-disciplinary collaborations. These concepts are linked to analogous counterparts from mechanical engineering by relying on conceptual similarities between mechanical stiffness and collaborative pliability. As a result, knowledge of cross-disciplinary collaborations can be immersed with existing conceptual knowledge of mechanical stiffness. Collaboration theories become part of the primary discipline instead of being treated as a separate discipline. The model, called the Collaborative Pliability Model, is assessed by a group of six mechanical engineering students. The students considered the model to be an effective tool for learning about cross-disciplinary collaborations. By using the model, they came up with numerous new factors affecting the success of their collaborative process. This gives a promising outlook for the future and opens the door for new applications for developing an analogy used to transfer knowledge between specific target groups.

Ultrasound gives the opportunity to look at muscles and observe their change in length. This tool has increased the knowledge about muscle-tendon dynamics and sometimes revealed surprising muscle stretch behaviour. System identification experiments use robots to disturb the ankle and measure its torque and angle. Muscle movement is derived from the measured joint angle and their assumed relationship. This study uses plane-wave ultrasound imaging to investigate the relation between ankle angle and muscle length during system identification experiments. The first goal is to determine the feasibility of using ultrasound measurements for system identification.  The second goal is to investigate the validity of the assumption that muscle stretch is proportional to ankle angle and the effect of this assumption on the prediction of reflex size. Transient and continuous disturbances were applied to the ankle, while images of the soleus and gastrocnemius were recorded with ultrasound and processed with an image tracking algorithm. For small (1° SD) continuous perturbations ankle angle and muscle length can be assumed proportional during a relax task. However, a conclusion cannot be drawn for the position task due to the low coherence of the muscle length measurements. For transient perturbations with a high velocity (>90°/s) the muscle length showed oscillations that were not present in the ankle angle, demonstrating a non-proportional relationship. The gastrocnemius velocity predicted the size of the short latency reflex better than the perturbation velocity of the ankle robot. Using plane-wave ultrasound imaging for system identification experiments was feasible.