It is of key importance for modern printing systems to maintain high standards of efficiency, reliability and print quality. In this regard, the scope of this work is to investigate the applicability of data analysis and machine learning techniques to improve the performances of industrial printers manufactured at Océ Technologies. Two critical aspects of the considered printing process are analysed and multiple algorithms are developed. Temporary failure of printhead nozzles, a well-known issue in inkjet printing, is first addressed. Available data are used to real-time estimate the health status of each nozzle. This allows for a prompt identification of problematic scenarios and lays the foundations for the introduction of condition-based maintenance. The analysis is further enhanced by the application of machine learning. Gaussian process regression is used to predict the evolution of nozzle failures. The designed solutions show to be precious tools for nozzle prognostics, providing great accuracy and a high level of flexibility. Problematic nozzles can be restored by performing automatic cleaning actions. However, because of the costs and limited efficiency of these, their appropriateness is highly questionable. Such delicate matter is tackled by designing an autoregressive model that enables to define an advanced cleaning strategy. The presented method allows to decrease the cleaning costs up to 5-10%, leading to a considerable operational cost reduction. All the proposed solutions are thoroughly evaluated and compared, considering both their efficiency and implementation costs. Therefore they represent valuable proposals, ready to be factually implemented on an Océ printer.

There are many stages that involve humans handling food objects in the processing chains from farms to stores. For some of these tasks it is desirable to look for a robotic solution to either assist the human or even take over that task, e.g. if it is physically demanding, imposes contamination risks or because of economical considerations. Moreover, recently the COVID-19 pandemic revealed even more vulnerabilities in our food processing chains, when seasonal labourers where blocked at borders and some food processing sites turned out to be "corona hotspots" with the result that whole food processing chains were disturbed. A step towards solving some of these problems is studying robotic grasping, since it is a crucial skill for many manipulation tasks. This work focuses on force closure grasps for pick-and-place tasks. Since humans grasp novel objects effortlessly, a human inspired approach is proposed that combines visual and tactile sensory information and learns from its mistakes thanks to reinforcement learning. Visual features obtained from an RGB-D camera in combination with pressure information from sensors on the finger tips of the robotic hand are used to set the desired grasp force adaptively while holding the object. A novel algorithm is proposed for learning to grasp with minimal force: LIFT (Learning of Initial Force and Tuning). The novel grasp approach is evaluated in a Gazebo simulation environment and on a real-world robotic setup. The approach results in successful grasping in both simulation and on the real-world setup in tens or hundreds of learning interactions, depending on the size of the state-action space. Success rates of 96.3 % and 96.7 % are obtained in simulation and on the real-world setup, respectively. The results from the experiments indicate that the approach is successful in grasping with minimal force.

Robots are increasingly deployed in various locations to automate tasks, including in barns. However, in barns cows can obstruct the sensors such as LiDAR or camera, leading to a lack of environmental information. As a result, the robot’s localization system only relies on odometry at those moments, introducing additional uncertainty to the robot’s pose. When the visibility of the environment is restored, the robot may mistakenly believe it is in a location that does not correspond to its actual position. The first contribution of this master thesis is a novel method for improving the self-localization in barns by implementing a line detection algorithm which is called Line Adaptive Monte Carlo Localization (LAMCL). The novelty is that only line segments are used to detect a localization error instead of corners between different line segments. It also retains the robot’s current pose to filter out fault-detected localization errors. In addition, the new approach is applied in a dynamic environment. The proposed method combines the Split-and-Merge line detection algorithm with AMCL. The detected line segments are compared with the walls in the environment to identify localization errors. When an error is detected, the robot’s pose is adjusted by placing a section of the particles at the location of the error. In this way, the robot can find its true location again. The second contribution is a new dataset. This new dataset, called DataCow, consists of four recorded routes in a barn with GT on a handful of spots to evaluate the self-localization. DataCow includes the pseudo-2D LiDAR scans and the odometry of a robot driving through a barn. This dataset is used to evaluate the new self-localization method LAMCL. Through the experiments, it has been discovered that this new method improves the system’s recovery ability, but the accuracy and precision are compromised.

Deterministic predictions of wave induced ship motion are increasingly often used to enlarge the oper- ational envelope of ships that perform complex operations at open sea. The large amount of complex operations performed nowadays, motivates the relevance of motion predictions. The company Next Ocean provides these forecasts, based on linear seakeeping theory. While highly linear degrees of freedom are predicted accurately, roll motion predictions prove to be more demanding and are less accurate. This research aims to improve roll prediction accuracy by adding (non)linear damping to the equations of motion. The prediction optimization is performed on data of the platform supply vessel Acta Auriga. To enable the use of nonlinear forces, the equations of motions are first implemented in the time domain using Cummins’ equations. The linear coefficients in this equation are determined from the frequency dependent coefficients in Acta Auriga’s hydrodynamic database. To ensure the correctness and investigate the limitations of the implemented Cummins equation, the results from the time and frequency domain are extensively compared. Both monochromatic excitations, as well as spectrum (JONSWAP) excitations are considered. In contrast to linear force coefficients, nonlinear force coefficients are generally not known for a given vessel. Implementing these forces into the equations of motion imposes the need for an approximation of these coefficients. Estimates for the linear, quadratic and cubic viscous damping forces are made, using both the measured motions of a vessel operating at sea and the predictions of the corresponding excitation force, made by Next Ocean. A multivariate regression algorithm is used for the identification. The Cummins equation was successfully implemented. However, the translation from frequency to time domain was more error prone than anticipated; the frequency dependent coefficients need to meet requirements that are typically not met by databases meant for frequency domain calculations only. Furthermore, degrees of freedom without a restoring force showed running away behaviour that could not be negated without adding extra damping. The roll motion was susceptible to instabilities. The exact origin of this instability was looked for, but could not be found. Adding damping resolved the instability. Furthermore, the interpolation in the roll response amplitude operator introduced errors in the initialisation of time domain calculations, which led to inaccurate results when spectrum excitations corresponding to more severe sea states were used. Only under specific conditions, these high energy spectra led to accurate roll predictions. Adding damping made for a close match between frequency and time domain calculation for all degrees of freedom. The complexities mentioned made that an easily scalable algorithm could not be obtained using time domain calculations. Easy scalability be- ing important to Next Ocean means that running time domain simulations in Next Ocean’s product is deemed unrealistic. This also means that no nonlinear forces can be used in real time applications. In an attempt to improve linear predictions, coefficients in the linear seakeeping model, including the linear damping coefficient, are identified and vessel motions are re-predicted with the added damping and updated linear force coefficients, using the Cummins equation. None of the identified parameters led to better predictions than were obtained by Next Ocean. Identifying and updating parameters was therefore concluded to be not beneficial to the quality of the motion predictions. Adding a fixed amount of linear roll damping, which was not identified from the field data, did lead to improved prediction quality; correlations between predictions and measurements increased by 9.6%. While the motion prediction could not be improved using the Cummins equation, valuable information on the time domain simulations was obtained. The finding that better predictions were consistently obtained by adding a fixed amount of damping, sparks an opportunity for further research into the ideal amount of damping. This, and more elaborate identification schemes could provide meaningful insight to improve motion predictions.

In recent years, visual Simultaneous Localization and Mapping (SLAM) have gained significant attention and found wide-ranging applications in diverse scenarios. Recent advances in computer vision and deep learning also enrich visual SLAM capabilities in scene understanding and large-scale operation. However, despite remarkable performance in these fields, most visual SLAM frameworks are designed with the static world assumption. Thus, they often confront challenges in dynamic environments, manifesting reduced localization accuracy, tracking failures, and restricted generalization. To investigate the impact of moving objects in dynamic indoor environments, we first benchmark representative visual (dynamic) SLAM approaches, complemented by robustness assessments for preliminary insights. During this process, we adopt challenging sequences from GRADE, an ideal platform for simulating dynamic indoor scenes. Notably, the mainstream of dynamic SLAM methods employs detection or segmentation techniques as solutions. To explore the correlation between detector accuracy and overall SLAM performance, we integrate a series of trained YOLOv5 and Mask R-CNN models, each with varying accuracy levels, into dynamic SLAM systems. Subsequently, we evaluate these configurations on the TUM RGB-D sequences. Contrary to common intuition, the experiments indicate that more accurate object detectors do not necessarily lead to improved visual SLAM performance. This benchmarking process also illuminates several inherent limitations of current dynamic SLAM techniques, underscoring the imperative for further advancements. Building upon these insights, we introduce DynaPix SLAM, an innovative visual SLAM system for dynamic indoor environments, where participation of visual cues (e.g., features) is weighted based on per-pixel motion probability values. Our approach consists of a semantic-free pixel-wise motion estimation module and an improved pose optimization process. In the first stage, our motion probability estimator employs a novel static background differencing method on both images and optical flows to identify moving regions. These probabilities are then incorporated into the map point selection and weighted bundle adjustment for backend optimization. We evaluate our DynaPix SLAM and its variant, DynaPix-D, in comparison with ORB-SLAM2 and DynaSLAM. These assessments are performed on both TUM RGB-D and GRADE sequences, with additional tests on the static versions of the GRADE ones. The results demonstrate that DynaPix SLAM consistently outperforms the other methods, showcasing reduced localization errors and longer tracking durations across various scenarios.

Loop Robots develops and operates the next generation of fully autonomous disinfection robots in hospitals and healthcare settings. Accurate localization is essential in order to navigate reliably and effectively disinfect the tight hallways and corners of a patient room, operating room, or intensive care unit in a hospital. Odometry, or dead-reckoning, is the process of estimating the robots pose with respect to a known initial pose. It forms the backbone of the robots localization strategy, as well as being critical to many other processes that run on the robot, such as control and mapping. The current strategy of generating odometry using the wheel encoders is prone to drift due to the integration of errors into the estimate. Moreover, the estimation takes place on the horizontal plane due to the nature of the wheel encoders. To improve the odometry and extend it to the full 3D space, a solution that makes use of all of the onboard sensors in a tightly-coupled manner is required. In this Thesis, we make the first steps towards this larger goal. The contributions of this thesis include: (1) an Extended Kalman Filter (EKF) algorithm to fuse data from the Inertial Measurement Unit (IMU) and wheel encoders to estimate position and orientation of the robot in 6-DoF; (2) a line feature extraction and tracking methodology to extract primitives from LIDAR data and (3) A Moving Horizon Estimation (MHE) scheme based on a factor-graph formulation to perform pose estimation on the horizontal plane using LIDAR data and wheel encoders. We test the three modules individually using a combination of simulations and real-world data wherever possible. We found that the MHE scheme was able to reduce drift over the long term, but is sensitive to the effects of outliers in feature matching, motion distortion of LIDAR scans, and wheel slip. The EKF scheme is able to reduce the overall drift and correct for wheel slips. Based on these results, promising avenues for the improvement of all the proposed modules are given, along with recommendations on how to combine them all in a tightly-coupled fashion.

Artificial Neural Networks (ANNs) have emerged as a powerful tool for classification tasks due to their ability to outperform traditional methods. Nevertheless, their effectiveness relies heavily on the availability of large, varied, and labeled datasets, which are often not available. To counter this constraint, data augmentation techniques have emerged, leveraging existing data to generate additional, variant data. Extending these techniques to multi-dimensional time series data, such as the transportation mode detection data considered in this thesis, however, introduces challenges. In response, generative models such as Variational Autoencoders (VAEs) have shown promising advancements. In this context, this thesis investigates the application of the Iterative Hierarchical Data Aug- mentation (IHDA) algorithm for ANNs, which represents a VAE-based data augmentation technique. The IHDA method utilizes VAEs not only to generate new data samples but also to map existing data to a lower-dimensional latent space, which is then utilized for identifying samples that might require additional training. The proponents of this method, Khan and Fraz, reported an accuracy elevation for the considered transportation mode detection classifier from 83% to 92%. However, due to the absence of publicly accessible code for this algorithm, the initial step of this thesis involved implementing the IHDA algorithm. Further, this research proposed and incorporated advancements like the σ-VAE, aimed to improve the generative capacity of the VAE and to refine its latent space mapping. Additionally, the Kullback-Leibler (KL) divergence was introduced as a similarity metric, aiming to optimize the identification process of samples that require retraining. Unfortunately, the results reported by Khan and Fraz could not be reproduced in this study. Furthermore, despite the potential shown by the σ-VAE to improve the generative capacity and refine the latent space mapping, along with the enhanced sample identification through the KL divergence, these enhancements did not lead to an overall improvement in the IHDA algorithm. This was primarily attributed to the low generative performance of the VAEs utilized, which also hindered a thorough evaluation of the effectiveness of the IHDA algorithm. Given these outcomes, it is suggested that future work should focus on employing more complex VAE models with the potential to enhance their generative performance, which, in turn, could improve the IHDA algorithm’s overall effectiveness.

Dealing with inherently unmodeled dynamics and large parameter variations or faults, is a challenging task while controlling robot manipulators. Classical control techniques cannot usually provide satisfactory responses, and often external supervision systems have to be designed to handle the faults. Recent research has shown that active inference, a unifying neuroscientific theory of the brain, bares the potential of intrinsically coping with strong uncertainties in the system, mimicking the adaptability capabilities of humans. However, the current state-of-the-art regarding active inference in robotics is very narrow and limited. This thesis presents a novel active inference controller as a general adaptive fault tolerant solution for control of robot manipulators. The goal of this work is threefold. First, we demonstrate the applicability of active inference in robotics, deriving a control scheme which is computationally efficient and with high performance. Second, we verify the claimed adaptability properties of active inference against a model reference adaptive controller, in a simulated on-line pick and place task with a 7 degrees-of-freedom robot arm. Third, we propose a method to exploit the controller's structure to perform fault detection, isolation and recovery, without the use of external supervision systems. This work showed that not only active inference is applicable to robotics, but it also outperforms the model reference adaptive controller, and it allows to efficiently deal with sensory faults. This thesis represents a leap forward with respect to the current state-of-the-art of active inference for robotics, and it lays the foundations for further research in this direction.

Problem: Due to high rejection rates regarding prostheses’ use, the assessment of the amputee’s use of the prosthesis has become more critical. Today’s prosthesis research is limited to assessing a users’ performance to perform tasks in a controlled environment. Therefore, these studies cannot wholly assess how the prosthesis is used in the daily lives of amputees. Purpose: The purpose of this thesis project is to create, test, and examine the performance of methods that can be used to enhance prostheses' research and evaluation. Results: We have created several enhancements and extensions regarding prosthesis research and evaluation for three sensor scenarios. In the first scenario, we only use an accelerometer, which allowed us to create a new method for creating a vector magnitude (VM) that can show us what type of arm movement is made; additionally, we created a novel scoring system to determine the intensity of movements that are performed. In our second scenario, we used an additional gyroscope, which opened up possibilities for using more advanced Sensor Fusion (SF) techniques. We created an accurate tilt estimation algorithm that remains robust against high levels of gyroscope noise, accelerometer outliers, and measurement model violations. With the gyroscope, we were also able to create a VM from rotational velocity, which displays information about the arms’ rotational movement. Additionally, we created a novel scoring system that also shows the intensity of the performed movements. Our last scenario considers the use of two Inertial Measurement Units (IMUs); we present a novel algorithm for estimating the relative sensor orientation. This algorithm uses a joint kinematic estimation method that incorporates the connection between adjacent segments within a SF algorithm that remains accurate in the vicinity of common real-world disturbances, among which are; high levels of gyroscope noise, accelerometer outliers, and Soft-Tissue-Artifacts (STA). Conclusion: The best way to enhance the analysis for prostheses research and evaluation is to start incorporating gyroscopes into the research process. This can either be in the form of the single sensor case or the double sensor system. The additional gyroscope(s) will enable the use the SF techniques and methods as discussed in this report.

Inertial Measurement Units (IMUs) have become increasingly popular human motion estimation due to their portability, self-contained features, and cost-effectiveness compared to marker-based sensing systems which rely on external cameras to observe the position of the markers. Over the years, many studies have proposed various models and algorithms based on IMUs and the kinematics of the human body for motion estimation, neglecting the fact that IMUs measure acceleration, which can be directly related to joint torque with known inertial parameters. In turn, by estimating joint torque and incorporating kinetics into the model, it becomes possible to address a long-standing problem in the field of biomechanics: the marker-based sensing system’s inability to provide a reliable estimation of kinetics due to the need for numerical differentiation. In a previous study, a kinetics model was proposed for estimating the motion but validated only on a robotic arm. In our study, we further explore the performance of using IMUs to estimate wheelchair user motion data based on Extended Kalman Filter (EKF) and the kinetics model, with marker-based Inverse Kinematics (IK)/Inverse Dynamics (ID) as the benchmark. Compared to Marker-based IK, the method leveraging kinetics achieves a Root Mean Squared Difference (RMSD) below 16◦ for three out of four tasks across all joints throughout the trial. After analyzing the only task with degradation in estimation, we conclude that the erroneous IMUs measurements results from Soft Tissue Artefacts (STA) is the most likely reason. For joint torque estimation, the RMSD for joint torque estimation is below 3.05Nm for the tasks less affected STA. Through fine-tuning the EKF, we can achieve fast and responsive estimation results without being affected by numerical differentiation, enabling us to capture both sudden and subtle changes in joint torque estimation. The kinetics model performs better than the kinematics-based model in estimating both kinematics and kinetics and also reduced drifting behavior. Compared to OpenSense, which depends on magnetometer measurements, kinetics model estimation shows comparable kinematics estimation accuracy while excluding the use of heading information. The results show that including the kinetics model for human motion estimation can improve estimation accuracy and robustness encouraging further studies to include kinetics for human motion estimation.

This thesis proposes a novel algorithm, Wi-Closure, to improve computational efficiency and robustness of map matching in multi-robot SLAM. Current state-of-the-art techniques connect maps with inter-robot loop closures, that are usually found through place recognition. Wi-Closure decreases the computational overhead of these approaches by pruning the search space of potential loop closures, prior to evaluation by a typical place recognition algorithm. Wi-Closure achieves this by identifying where trajectories are close to each other through sensing spatial information directly from the wireless communication signal. Then, place recognition is only performed on scans taken at locations close to each other. Wireless sensing provides information even when operating in non-line-of-sight or without existing communication infrastructure. The validity of Wi-Closure is demonstrated in simulation and hardware experiments. Results show that using Wi-closure greatly reduces computation time, by 54% in simulation and by 77% in hardware, compared with a multi-robot SLAM baseline. Importantly, this is achieved without sacrificing accuracy. Using Wi-Closure reduces absolute trajectory estimation error by 99% in simulation and 89% in hardware experiments. This improvement is due in part to Wi-Closure’s ability to avoid catastrophic optimization failure that typically occurs with classical approaches in challenging repetitive environments.

Many recent robot learning problems, real and simulated, were addressed using deep reinforcement learning. The developed policies can deal with high-dimensional, continuous state and action spaces, and can also incorporate machine-generated or human demonstration data. A great number of them depend on state-action value estimates, especially the ones in the actor-critic framework. Deriving unbiased estimates for these values is still an open research question, mostly since the connection between accurate value estimates and system performance is not yet well-understood. This thesis work has three main research contributions. Firstly, it analyzes the connection between value estimates and performance for the TD3 algorithm. Secondly, it derives theoretical bounds for the true value function when dealing with environments where a reward is only given for successful completion of a task (sparse/binary reward). Lastly, a deliberate underestimation objective is added to the TD3 algorithm together with the theoretical bounds to improve system performance when using human demonstration data that only covers a specific part of the state and action space. All the algorithms are tested and evaluated using simulated robot manipulation tasks in the robosuite environment, where the robot is first trained on the demonstration data and then can gather more experiences in the simulation. Results show that the deliberate underestimation together with the value bounds enable the robot to learn from human demonstration, which was not possible for the standard TD3. Additionally, applying just the value bounds speeds up the learning process when using machine-generated datasets.

To improve Active Vibration control methods like Independent Modal Space Control (IMSC), an estimate of the modal contribution of vibrational eigenmodes of compliant structures is required. In this work, three Recursive Bayesian input and state estimation algorithms previously introduced in civil engineering are evaluated for use on on high-tech compliant mechanisms to estimate modal contributions for use in Active Vibration Control. The difference in environment when applying these algorithms from civil engineering into high-tech compliant motion stages allows for a significantly different sensor configuration possibly improving estimation performance. The three algorithms, namely, the Augmented Kalman Filter (AKF), Dual Kalman Filter (DKF) and Gilijns de Moor Filter (GDF) are implemented in simulation and on an experimental compliant motion stage setup. The modal contributions of the guidance flexures are estimated through a noisy acceleration measurement at the stage and strain measurement at the flexure base. For validation, the filters are evaluated on overall fit quality and system acceleration dependency through the quality of estimation of the flexure tip location. The DKF is shown worst performance with lowest fit scores overall and high acceleration dependency while the AKF and GDF perform comparably well. The GDF is shown to transmit less noise into the estimate however, and thus performed overall best.

Though NanoSats are becoming increasingly capable of featuring active payloads, and acquiring inertial data at arcsec level precision; the payload operational duty-cycle largely remains limited by its data downlink capabilities. NanoSat laser communication terminal (LCT) promises 1 Gbps data downlink capabilities. However, LEO to ground free space optical data transmission requires precision attitude knowledge under agile ground target tracking conditions. Under such stringent operational conditions, the calibration of gyro-stellar misalignment and scale factors is a well-established practice for traditional satellites. Nevertheless, the impact of such non-idealities in the context of MEMS gyro-stellar based NanoSat precision pose estimation is unclear. Furthermore, the gyro-stellar calibration performance that can be achieved with NanoSat ADCS system requires further investigation. Under agile operational conditions, slew rate induced star tracker drop-out is common for NanoSats, which limits the operational envelope of agile precision tracking missions. To increase our understanding of attitude knowledge estimation capability of today's NanoSats, this thesis work is extended at Hyperion Technologies. In collaboration with Hyperion Technologies, gyro-stellar sensors are characterised; and CubeCAT LCT is used to laydown the reference payload and mission requirements of an agile precision terrestrial target tracking NanoSat mission. Based on state steady noise PSD properties, and attitude knowledge requirements set by the LCT, gyro-stellar configurations are selected, and characterised. A rigid body attitude simulator is developed, considering the characterised sensor suite, to generate representative nominal and non-ideal sensor outputs under four different mission phases: 1.) Inertial pointing with no angular rate, 2.) Non-harmonic sinusoidal manoeuvres 3.) Agile precision ground target tracking and 4.) Inertial target tracking with non-zero angular rate. Reaction-wheel time delay and torque limits are evaluated to analyse the feasibility of attitude pointing requirements under agile target tracking conditions. Due to favourable properties of robustness to large initialisation errors, fast convergence, and preservation of attitude quaternion unity norm constraint; UnScented QUaternion Estimator (USQUE) variant of UKF based filter is synthesised for attitude estimation. The USQUE filter is further extended to facilitate the calibration of gyro-stellar misalignment and scale factors. It is demonstrated that the presence of gyro-stellar misalignment and scale factor, and star tracker occultation under agile slew rates significantly deteriorates the attitude estimation performance of USQUE based attitude estimator. Star tracker misalignments and dropouts are observed to have a significantly larger impact on the attitude knowledge estimation performance, when compared against MEMS rate-gyro scale factor and misalignment. Degraded MEMS rate gyro sensors are observed to have little impact on the attitude knowledge estimation performance of 6/7 state USQUE filter. However, it has a considerable impact on the convergence performance of star tracker misalignment calibration. More persistent calibration manoeuvres are observed to improve the star tracker misalignment parameter estimation performance. Gyro misalignment and scale factor calibration objectives were not met. A root-cause analysis showed that the signal distortions introduced by such non-idealities are below the gyro-stellar noise floor. Calibration filter synthesised has the potential to improve autonomy, reduce commissioning and ground calibration times, and enhance NanoSat pose estimation performance.

To enable communication for patients who have lost the ability to speak due to severe neuromuscular diseases, covert speech based brain-computer interfaces (BCIs) might be used. These system use neural signals arising from covert speech and translate them into text or synthesised speech. Covert speech is imagining to speak without moving any of the articulators and therefore does not rely on actual motor activity. As recognizing covert speech from neural signals is extremely challenging, machine learning algorithms are deployed. To make use of the full potential of machine learning approaches in the field of decoding covert speech and to accommodate real-world deployment of a BCI, a large number of training samples is required to train the networks. In this study, a novel database is presented containing EEG and audio data from 20 subjects recorded during the covert and overt pronunciation of 15 Dutch prompts. To validate the recorded data, two speaker-independent classification tasks were performed using a ResNet-50 algorithm as classifier with spatial-spectral-temporal features extracted from the EEG signals. The speaker-independent three-class classification of pre-stimulus (rest) trials versus covert speech trials versus overt speech trials obtained an average accuracy of 70.6% and the speaker-independent five-class classification of five covert vowels (“aa”, “ee”, “oo”, “ie”, “oe”) obtained an average accuracy of 19.6%. Even though the five-class classification task did not reach an above chance level accuracy, the high performance reached by the three-class classification task provides support of the existence of discriminative information in the covert speech segments to decode covert speech in the future. Future research should focus on EMG artifact detection and on determining the performance per subject to improve the dataset. Furthermore, subject normalisation strategies should be investigated to address the challenges of subject-independent covert speech decoding.

Underwater position estimation is challenging due to the absence of Global Navigation Satellite System (GNSS) signals. Underwater vehicles are typically equipped with a Doppler Velocity Log (DVL) that measures the velocity relative to the seafloor. Aside from the velocity, the DVL also measures the range of each of the four beams. When compared against a bathymetry height map, these measured ranges provide additional information enabling improving position estimates. Unfortunately, surveys often occur in areas where detailed bathymetry maps are unavailable. In these cases, bathymetry Simultaneous Localization and Mapping (SLAM) could be used to improve the position estimates compared to the velocity integration position. With SLAM, the map is being estimated during the mission while at the same time using the map for position determination. In this thesis, reduced rank Gaussian processes (GPs) are used as map representation for SLAM. The downside of regular GPs is a time complexity of O(n^3) , reduced rank GPs improve the computation performance. GPs provide Gaussian distributions at any point, leading to a neat integration with probabilistic SLAM algorithms. To the best of the author’s knowledge, this has not been used in bathymetry SLAM. This report investigates how reduced rank approximated GPs, representing the bathymetry, can be integrated into a SLAM algorithm to improve the position estimates of an underwater vehicle equipped with a DVL and low-quality gyroscopes. A squared exponential kernel is used as GPs model of the bathymetry. The reduced rank approximation is vital for real-time SLAM performance. This map representation is integrated with a Rao Blackwellized particle filter (RBPF) that estimates both the underwater vehicle’s trajectory and the bathymetry map. The SLAM algorithm is evaluated using data from an underwater vehicle operated at the surface such that a GNSS reference position is available. Experiments of the SLAM algorithm show a reduced position error compared to the GNSS reference. The resulting algorithm has a computation time of up to 30 times faster than the Autonomous Underwater Vehicle (AUV) collects data while improving position estimates. This concludes that the RBPF using reduced rank GPs is capable of onboard improved position estimation on underwater vehicles.

Single Molecule Localization Microscopy (SMLM) has enabled researchers to breakthrough the diffraction limit and obtain nanometer resolution images of macromolecular structures. But due to the time involved in obtaining ample data for proper image, the technique is venerable to many problem including fluctuations due to thermal gradients from surrounding which cause the frames to drift. SMLM relies on the stochastic blinking of fluorophore probes. Thus drift in SMLM could be explicitly modelled as a stochastic state space process. These models could be used to perform drift correction. Two state space models are proposed relying on different properties of SMLM. The first model utilizes shifting of underlying image structure. The state space model for this property is constructed using shift matrices. A system identification method along with image reconstruction method is also derived to form the drift compensation algorithm for this model. This algorithm is further developed to provide adequate performance within low computational time. The second model relies on the position of emitter molecules and utilizes linking or pairing of fluorophore probes in succeeding frame to obtain the output data. Drift compensation algorithm for this model is constructed using Prediction Error Methods (PEM) and Kalman (RTS) smoother. The drift correction algorithm for these two models are also bench-marked with existing algorithms to obtain insight into performance. Furthermore, other properties of these algorithms are explored using simulation dataset and recommendation are provided for improvement and further research.

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.

This research proposes a new differentiator for estimating higher order derivatives of an input signal. The main reason why higher order derivatives are necessary is that Active Inference makes use of generalized coordinates. This means that it keeps internally track of higher order temporal derivatives of states, inputs and measurements. The difficult part of generalized coordinates are the generalized measurements, because these should be measured from a physical system. However, it is not always possible to measure all states of a system and it is definitely not possible to measure all higher order derivatives. A solution to this is to estimate the derivatives of states by using real time differentiators. A short literature study about differentiators is conducted and found that the state of the art differentiator is the algebraic estimation approach differentiator (AEAD). However, this and other differentiators have the problem that when they are converted to discrete time, they cannot track polynomial signals anymore. For that reason, this thesis proposes a new differentiator: the recurrent low pass algebraic differentiator (RLPAD). The proposed differentiator is compared with the (AEAD) in two experiments. The first experiment evaluates the performance based on three analytical inputs with known higher order derivatives. The three inputs are: a polynomial, a sine and a combination of the two. Additionally these three inputs are corrupted by Gaussian white noise. This experiment concludes that the proposed method outperforms the (AEAD) significantly when a polynomial or polynomial combined with sine input is used. They perform similar for sine inputs, although (RLPAD) is considered slightly better. The second experiment evaluates how the two differentiators perform when real sensor data is used. In order to conduct the experiment, the raw data is smoothed by a Savitzky-Golay filter to create a ground truth about the derivatives. This experiment concludes that the differentiators have an optimal set of parameters, where either the one or the other performs better. The proposed method performs a lot better when few derivatives are estimated. On the other hand (AEAD) performs better when there are more derivatives estimated. However, this is not true when the noise gets stronger. (AEAD) is more sensitive to noise than the proposed method. Based on the experiments, the proposed method (RLPAD) is the better differentiator for creating generalized measurements in Active Inference for three reasons. The first is that it can track polynomial signals. The second is that it has stronger noise attenuation. And the third reason is that it is computationally faster.

Inertial Measurement Unit (IMU)-based motion capture has gained interest over the years due to its potential to measure human movement in the clinic and on the sports field at low cost. Still, IMU-based motion reconstruction remains a challenging task as these IMU measurements are corrupted by noise and bias. There have been many filtering and sensor fusion algorithms developed to address this problem, but limited attention has been paid to including the system dynamics of the subject to estimate kinematics and kinetics from multiple IMUs. I propose a novel motion reconstruction algorithm for capturing 3D motions in a markerless and unconstrained environment using gyroscope and accelerometer measurements. The algorithm is based upon an Iterated Extended Kalman Filter for state estimation and the 3D dynamical model of the subject created in OpenSim (IEKF-OS). The IEKF-OS algorithm consists of two stages. The first stage incorporates the dynamics of the subject to predict the motion. The second stage tracks measurements from multiple IMUs to improve model estimates of joint angles and speeds. The goal of this thesis is threefold. Firstly, the IEKF-OS is derived and verified using simulations performed on a six-link robot manipulator including sensor placement errors. Secondly, experiments are performed to validate the algorithm’s estimations against the robot’s ground truth joint encoder values. Thirdly, the algorithm is compared against an inverse kinematics method to track IMU estimated orientations (OpenSense) provided by OpenSim. The IEKF-OS algorithm showed lower motion tracking errors compared to OpenSense for various motions performed by the robot manipulator. The joint angle estimations as computed by both methods are compared against the gold-standard ground truth robot encoder values. OpenSense joint angle values were in the range of 0.6-6.4 [deg] RMSE, whereas the IEKF-OS algorithm estimated joint angles and speeds in the range of 0.4-2.8 [deg] and 0.3-2 [deg/s] RMSE, respectively. These results highlight accurate 3D motion reconstruction on a six-link robot manipulator. Contrary to OpenSense, the IEKF-OS is able to accurately reconstruct motions regardless of magnetic disturbances because it does not rely on magnetometer data.