While CO₂ is one of the factors increasing the global temperature, it is also utilized in the agriculture to enhance crop yield. In the agriculture, a sustainable solution to CO₂ mitigation would potentially lie in the greenhouses. A moisture swing direct air capture system can adsorb CO₂ from the air through changes in humidity levels between the adsorption and the desorption stages. During adsorption dry air is required, while during desorption humid air is required. In a greenhouse, where humid air is typically required for the crops and CO₂ enhancement is often achieved through fossil-fuel combustion, a moisture swing system could be a sustainable innovation.

Insights into the feasibility and potential implementation of a moisture swing column designed to enhance CO₂ concentrations within a greenhouse are obtained through an investigation of the key parameters of the sorbent, the effect of global climate conditions on the sorbent, a parameter sensitivity analysis of the performance of the system, and a preliminary techno-economic viability assessment.

A one-dimensional numerical model was developed to solve the mass balance of a cylindrical adsorption column. Critical key parameters that affect the CO₂ saturation coverage of the sorbent are the gas phase CO₂ concentration, the relative humidity, and the temperature. Investigated is how these conditions vary globally, highlighting optimal arid climate regions with a high saturation coverage. It was found that seasonal and diurnal deviations are significantly higher in regions where a lower saturation coverage can be reached. From the sensitivity analysis of the parameters and conditions, it was that a 1.6-meter-long column achieved a system efficiency of 15.0 micromoles of CO₂ per kilogram sorbent per second, a water loss of 45.4 moles of water per mole of CO₂, and a system productivity of 9.05 moles of CO₂ per second. In the short term, the majority of costs are attributed to the capital expenditures, whereas in the long term, operational expenditures become the dominant expense.

Over nine billion people will have to be fed fresh and healthy food in 2050 according to United Nations. This puts pressure on the horticulture sector, responsible for a large portion of the world’s food production. The literature has shown that automatic optimal control algorithms are able to make better use of resources and can even outperform growers in terms of resource efficiency.

While applications of automatic optimal control methods show promising results, few are made adaptive to the greenhouse-crop system, assuming the system to be time-invariant. That is, the prediction models or techniques used in the controllers are not updated to adapt better to the evolving greenhouse-crop system. This thesis sets a first step towards this on-line system adaptation: it investigates the possibility to use a linear state-space model to estimate the non-linear greenhouse-crop systems dynamics around an operating point, for the purpose of automatic optimal climate control. In this work, a linear state-space model is identified and is implemented in an automatic optimal controller based on Model Predictive Control (MPC). Moreover, a greenhouse-crop simulation is built from state-of-the-art models based on first-principles. This ground-truth simulation is used to generate data and test the proposed control method.

In this thesis, I introduce a novel model identification approach to time series forecasting. For linear stationary processes, such as AR processes, the direction of time is independent of the model parameters. By combining theoretical principles of time-reversibility in time series with conventional modeling approaches such as information criteria, I design a criterion that employs the backwards prediction (backcast) as a proxy for the forecast. Hereby, I aim to adopt a finite sample, data-driven approach to model selection. The novel criterion is named the \textit{backwards validated information criterion (BVIC)}. The \acs{BVIC} quantifies forecasting performance of models by trading off a measure of \mbox{goodness-of-fit} and a models ability to predict backwards. I test the performance of the BVIC by conducting experiments on synthetic and real data. In each experiment, the BVIC is examined in contrast to conventionally employed criteria.

Experimental results suggest that the BVIC has comparable performance to conventional information criteria. Specifically, in most of the experiments performed, I did not find statistically significant differences between the forecast error of the BVIC under certain parameterizations and that of the different information criteria. Nonetheless, it is worth emphasizing that the BVIC guarantees are established by design where the model order penalization term depends on strong mathematical properties of time-reversibility and a finite data assessment. In particular, the penalization term is replaced by a weighted trade-off between functional dimensions pertaining to forecasting.


That said, I observed that the BVIC recovered more accurately the real order of the underlying process than the other criteria, which rely on a static penalization of the model order. Lastly, leveraging the latter property, I perform the assessment of the order model (or, memory) of time series pertaining to epileptic seizures recorded using electrocorticographic data. The results provide converging evidence that the order of the model increases during the epileptic events.

The increase in quantity and electrification of energy in residential areas pushes the existing distribution network to its limits. If left unchanged, the network will suffer from voltage violations caused by the increase of electric vehicle (EV) charging, energy generation by Photovoltaics (PV) panels and the phasing out of natural gas. Typically, cables are reinforced to cope with increased demand, but upgrading large parts of the Dutch electricity network is expensive. By not expanding the network, the remaining capacity of the network needs to be optimally used. This situation encourages the switch to an active distribution network, which opens up a new possibility for voltage management and optimal use of the distribution network.
One of these prospects is the multi-port converter. This converter combines the usually separate converters of EV, PV and battery energy storage system (BESS) to a single entity. Combined with an optimal charging algorithm, the energy management system (EMS) of the multi-port converter decides when to (dis)charge the BESS and EV, using a moving horizon window. By controlling the previously nondispatchable loads, flexibility is generated on the demand side. Using a network of multi-port converters on its own is insufficient to cope with voltage problems in the distribution network due to the overlap of (dis)charging schedules and lack of communication.
A Voltage management system (VMS) is used to coordinate the EMS of all the multi-port converters in the network to stay within the voltage range. This controller aims to prevent and correct voltage problems by varying the input or output of the multi-port converter. The prevention is done by actively reserving energy and changing prices to encourage changes in the (dis)charge schedule of the multi-port converter. The correction of the voltage profile is achieved by active power (Volt-Watt), reactive power (Volt-VAR)and apparent power (Volt-VA) control. 
Due to the stochastic nature of solar energy and load demand, the actual battery capacity and voltage profile may deviate. To combat the change in battery capacity, a reference tracker is installed. For the change in voltage, a real-time controller is implemented, used for rapid response. These controllers are placed inside the EMS under the optimal charging algorithm. 
The main focus of this thesis is to design a control structure, which can be gradually deployed, to prevent and correct voltage problems in a low voltage feeder in various time scales. By using multiple intertwined controllers, the proposed hierarchical control structure achieves fast operation, prevention and correction of the voltage profile over an extended period.
The simulations have been validated on the IEEE European low voltage test feeder. Results demonstrate that a limited number of multi-port converters is needed to keep the voltage profile within limits. The various correction methods have been tested for effectiveness to correct voltage deviations. Both active (Volt-Watt) and apparent (Volt-VA) power show capabilities of voltage control. Furthermore, the real-time controller is capable of removing individual load usage from the network and prevent voltage violations caused by uncontrolled PV and EV.

Multi-robot motion planning without a central coordinator usually relies on the sharing of planned trajectories among the robots via wireless communication in order to achieve predictive collision avoidance. Path planners found in the literature that feature this scheme usually boast levels of performance comparable with their centralized counterparts. However, in practice, communication tends to be unreliable and lead to significant performance degradation, which may eventually result in collisions among robots.

This thesis proposes a data-driven decentralized approach for multi-robot motion planning in dynamic environments. Instead of requiring robots to share their intentions with each other, a model based on recurrent neural networks (RNNs) is used to predict them. This model is trained on data obtained with a centralized sequential model predictive control (MPC)-based motion planner in which intended trajectories of neighboring robots are available while planning. The learned model can be efficiently run online and provide accurate predictions for each robot in the environment based on a horizon of past observations of all robots' states. This model is then incorporated into the MPC planning framework so that it may be run in a decentralized manner. It is finally shown in simulations that the proposed approach achieves a similar level of performance to its centralized counterpart.

Vehicle platooning is often regarded as one of the most promising solutions for solving the road congestion issues. Experiments have successfully proven that longitudinal platooning can be safely achieved in real world applications using a platoon of several vehicles. Lateral control however, has received far less attention. The main difficulty with lateral platooning lies in the fact that the two possible vehicle following methods both have their own limitations. The direct vehicle following approach suffers from corner cutting, whereas with path following using path reconstruction is often unreliable and difficult to achieve in platooning applications. In this thesis work a new method is proposed which combines the strengths of both vehicle following methods, and at the same time negate their drawbacks.

Multi-Agent Systems, often referred to a network of loosely connected autonomous units, are widely used to model the dynamics of crowds, vehicles, robots and swarms in traffic management, biological environment, distributed control and communication technologies. Recently, the study of multi-agent systems is rapidly growing due to the beneficial advantages of using a team of agents in logistics, mapping, search and rescue, etc.

In this thesis, we focus on the problem of decentralized collision avoidance among multiple intelligent moving agents. Imaging each agent to be a human, a moving car, or an aircraft, it is supposed to make its decision independently based on its perception of the local environment through sensors only. Two different collision avoidance protocols are presented to generate updated reference velocity continuously for each agent that leads to no future collision. The first method, the rotation based method, is adopted by a geometric based algorithm introduced for 2-Dimensional space and the second method, the potential fields based method, could be categorized as an example of Harmonic Potential Fields (HPF) discussed by Masoud. Both methods employing position information of obstacles result in collision-free paths for each agent under the assumption that all other agents following similar maneuvers. Through simulations in MATLAB, satisfied performances are achieved for method 2 in all challenging scenarios we set while method 1 faced some difficulties dealing with multi-obstacle simultaneously as well as 3D scenarios.

This research considers using center of mass (CoM) height variation as an input for balance control on a humanoid robot. Traditional balance strategies for humanoid robots are taking a step, control of the center of pressure (CoP) location, a result of the ‘ankle strategy’, and changing the angular momentum about the CoM, for example by a ‘hip strategy’. For humanoid robots, a common assumption behind these strategies is that the CoM height is predefined. However, CoM height changes can be used as an input for balance control, as for example can be observed during the landing of an athlete after a long jump. The first contributions of this work are bounds on the initial states for the variable height inverted pendulum (VHIP) from which convergence is possible to a stopped state, also known as capture regions. First, only a unilateral contact constraint is considered; negative CoM acceleration cannot be smaller than gravitational acceleration. Second, CoM height constraints are added to the model, after which a capture region can still be computed in closed-form. Third, vertical force constraints are added, after which capture regions are computed numerically using a bang-bang control law. The last capture region bridges the transition to the applied part of this work. The second contribution is a control law on vertical acceleration, suitable for application on a humanoid robot using a momentum-based control framework. Push recovery is tested on NASA’s Valkyrie humanoid robot while the robot is standing. In simulation, an increase in recoverable push of 9% can be observed when comparing to a controller that only uses CoP, when pushing the back of the robot. On hardware, an average increase of 7% can be observed for this push direction using a load sensor. Additionally, tests are conducted on hardware on Boston Dynamics’ Atlas using a medicine ball on a rope, but no improvement in recovery is observed. The control method for standing push recovery is also extended for use while the robot is walking. For Valkyrie in simulation, recovery improved the most compared with a predefined height approach for a push applied in the first part of the single support state for rear and frontal push directions. Additionally, a hardware result on Atlas while walking is briefly presented.

A challenging task in synchronization of multi-agent systems is steering the network towards a coherent solution when the dynamics of the constituent systems are heterogeneous and residing in a possibly large uncertainty set. In this situation, synchronization can be achieved via adaptive protocols (with adaptive feedback gains, or adaptive coupling gains, or both). However, as state-of-the-art synchronization methods adopt a distributed observer architecture, they require to communicate extra local observer variables among neighbors, in addition to the neighbors' states (or outputs). The distinguishing feature of this work is to show that synchronization in heterogeneous and uncertain multi-agent systems is possible without the need for any distributed observer. This can be achieved by designing adaptive distributed synchronization protocols, based on homogenization reasoning. Specifically, ideal gains are defined (feedback and coupling gains) that could lead all the heterogeneous agents to a desired homogeneous behavior and thus synchronization. However, since these gains are unknown in view of the unknown dynamics, we design adaptive laws for these gains that lead the agents toward synchronization. In this thesis, different control protocols are designed to address both leaderless and leader-follower synchronization of linear multi-agent systems. The protocols are then extended to achieve synchronization over a special class of nonlinear multi-agent systems, whose units are modeled as Kuramoto-like systems. Throughout this work, convergence of the synchronization error to zero is shown via Lyapunov analysis, and numerical examples demonstrate the effectiveness of the proposed protocols.

Quadrotor UAV's have become extremely popular over the last decade, as they combine great agility with mechanic simplicity. For tasks such as area surveillance or maintaining a communication network, the deployment of multi-agent systems is required. In these scenarios, it often is necessary for the robot teams to navigate in formation while avoiding static and dynamic obstacles. In situations, where team keeping is not crucial it can be beneficial to split and merge robot teams. This could be the case, if multiple targets are supposed to be tracked, or due to obstacle avoidance in order to improve goal progress and help avoiding deadlocks.

Within this thesis a novel approach for splitting and merging of robot teams moving in formation is developed. The method extends an existing geometric method for local formation control, which uses the approximation of typically non-convex workspaces by obstacle free convex regions. The regions are used to constrain a non-linear cost function, which is minimized through Sequential Convex Programming (SCP). This determines the references position for the robots and furthermore allows for the adjustment of the formation shape according to the environment. This thesis presents a goal directed version of the obstacle-free convex regions. These regions are used to determine team splitting and merging based on region intersections. The method takes into account limited sensing and communication range of the agents. Furthermore computations are done in a distributed manner to allow for better scaling with the number of team members.

The presented method is validated through simulations in various scenarios. Furthermore, an experimental framework is recreated and experiments with up to four quadrotor UAVs of the type Parrot Bebop 2 are conducted. The approach shows sufficient real-time capabilities in static and dynamic environments, while maintaining safe navigation of the drones.

One of the most crucial links to building an autonomous system is the task of decision making. The ability of a vehicle to make robust decisions on its own by predicting and assessing future consequences is what makes it intelligent. This task of decision making becomes even more complex due to the fact that the real world is uncertain, continuous and vehicles interact with each other. Sensors that perceive the real world and measure quantities such as position and speed of other traffic objects are inherently noisy and are further susceptible to external conditions. On the other hand, the road users’ intentions are stochastic and not measurable, and the presence of partial or complete vision based occlusions can make any measurements obtained useless. The decision making unit thus has to be aware of these issues and use the limited knowledge available to anticipate future situations that could unfold in an infinite number of ways to then maximize a reward (or minimize a cost).

In this thesis, a method to make automated longitudinal decisions along a predetermined path for autonomous vehicles in unsignalized urban scenarios is proposed. The decision mak- ing problem is formulated as a continuous Partially Observable Markov Decision Process (POMDP) with a discrete Bayesian Network estimating the behavior of other traffic objects. The future evolution of states are predicted using a multi-model Trajectory Estimator along with the digital map of the current driving area. Since continuous spaces make the belief space infinitely large, calculation of the value function for this problem becomes computa- tionally intractable. The presented solver algorithm approximates the value function instead of computing it directly, and additional optimizations reduce the belief space exploration area in order to improve performance. The results show that this single generalized framework is able to handle all the tested urban scenarios with a good safety margin in scenes with multiple traffic objects, even under zero visibility of the other traffic objects due to the presence of occlusions.