Robots have been used to automate repeatable tasks in industries for decades. With the rise of computing power, robots can now be used in a wider range tasks involving finer motion control, even ones which involve sharing a workspace with humans. An example of this is using a dual arm robot for quality control, as was demonstrated by BMW with their RoboCT quality control system. However, such systems require guarantees of safety to demonstrate reliable operation.

This thesis will delve into the question of how to compute and execute a motion plan for a dual robot system which has a guarantee of synchronized and collision-free motion. This thesis proposes using a hierarchical control approach to ensure task execution while satisfying collision avoidance and alignment constraints. The proposed approach is verified through a case study of using two manipulators to pick and place a ball.

Rechargeable Lithium(Li)-Ion Batteries are a ubiquitous element of modern technology, as they pertain to efficient and sustainable energy storage for Electric Vehicles (EVs), as well as wind and solar farms. In the last decades, the production and design of such batteries and their adjacent embedded control, charging, and safety protocols, denoted by Battery Management Systems (BMS), has taken centre stage in the energy transition. A fundamental challenge to be addressed in battery technology, however, is the trade-off between the speed of the charging protocol employed by the BMS and the aging behaviour exhibited by the battery resulting in the loss of capacity in the battery cell, all while maintaining the safe operation of the battery.

This thesis aims to explore electrochemical models describing the charging and aging behaviour of Li-Ion Battery Systems, as well as the current existing charging protocols that aim to maximize charging speed while minimizing the aging effects that result in capacity loss. The proposed approach in this thesis is to adopt a data-driven approach to controller design, implementing improvements on an existent Reinforcement Learning (RL) pipeline to design an aging-aware battery-charging protocol, and extending the work into the field of Formal Methods for Systems and Control. This is done by expressing the closed-loop system resulting from a trained charging policy in the form of a data-driven abstraction capable of verifying the formal system specifications under probabilistic guarantees. Furthermore, a Counterexample-Guided Inductive Synthesis (CEGIS) scheme is proposed to additionally guide the training of the charging policy based on information from the learning results.

The rediscovered interest in space exploration has led to plans to establish outposts on the Moon and beyond. The lunar bases currently planned are to be manned incrementally, with robots performing most work. With new trends in robotics, the use of collaborating swarms has become more abundant. To support lunar operation, a distinct area within swarming, namely foraging, is proposed. Foraging systems have the primary objective of recovering resources. From literature, no foraging framework was found that included system maintenance in its mechanisms. This report aims to answer the question if this is possible while maintaining the benefits of foraging.

The research considers the fundamental act of recharging as its required maintenance task. To evaluate it dynamically, a rudimentary energetics model is included. For the framework of foraging, the work of [Adams] is used as a baseline. The newly proposed system implements an additional recharging region and role to perform recharging activities, both having major implications for role selection and agent operation. Furthermore, to enable navigation based on energy considerations, the experience communicated by mobile agents is amended to include the energy cost of a travelled path. In doing so, additional quality indicators of paths are available making path optimization a more dynamic process resulting in finer population behaviour. Finally, the decaying of beacons is updated and a fallback feature is introduced to maximize agent utilization.

The newly developed foraging system is evaluated using data collected through simulation in Webots. Simulation scenarios included obstacles with impenetrable boundaries and surfaces with increased rolling friction to emulate cost-expensive regions. Qualitative analysis identified all features of the foraging system as expected, both in the exploration and exploitation phase. Quantitative results proved that the system is able to function with the added requirements of recharging, perform path optimization with the additional path quality indicator, and can do so in various types of scenarios. With this, the research statement that foraging functionality is achievable with the practical considerations of robotics is confirmed to hold.

This thesis addresses the detection and prevention of adversarial attacks on self-triggered control (STC) systems and demonstrates how manipulating the sampling times can be used to provide a secure control framework. Secure control is vital to guarantee both the safety and security of critical infrastructure, which has been the target of malicious attacks over the past decade.

First, a novel watermarking scheme is proposed based on the early triggering of a well-designed STC policy, such that stability is guaranteed to be preserved. We show that this watermarking scheme, together with an event-triggered χ² detector we design, is able to detect replay attacks. An online heuristic for obtaining an optimal early triggering policy is provided. If certain assumptions hold we show that the early triggering policy is a discrete uniform one. Through an illustrative example, both a quantitative and qualitative comparison between two other watermarking schemes are provided. We conclude that none of the watermarking schemes can claim absolute superiority, and trade-offs between all considered schemes exist.

Next, we propose a new type of attack called a switched zero dynamic attack (ZDA), and provide an algorithm on how to construct these switched ZDA. We show that certain STC systems are susceptible to such attacks, and demonstrate that by tuning the triggering parameters there exist sufficient conditions such that these attacks are no longer disruptive. The effect of additive perturbations and a non-zero initial condition, as well as the proposed tuning method, are shown in a numerical example. We provide a qualitative comparison between several other countermeasures in the literature, which we extend for aperiodic sampling when needed. Finally, shortcomings and future directions are discussed.

Modern control systems require control strategies that can handle multiple levels of abstraction while providing formal guarantees of system behavior. Traditional hierarchical control approaches often lack formal interfaces between different layers of abstraction, making it challenging to ensure system-wide properties and extend implementations across different platforms.

This thesis presents a contract-based hierarchical control framework that provides formal guarantees across different layers of abstraction. Building upon recent developments in vertical contracts with heterogeneous refinements, we develop a three-layer control architecture for robot surveillance tasks, decomposing the problem into motion planning, trajectory generation, and tracking control. Each layer operates at its own time scale and uses appropriate model abstractions, with contracts ensuring system-wide guarantees while vertical system relationships enable proper interfacing between layers.

We validate our framework through both simulation studies and physical implementation on Elisa3 robots. The simulation results demonstrate the framework's ability to satisfy high-level temporal logic specifications while maintaining guarantees at each layer. The physical implementation reveals practical challenges in transitioning from simulation to hardware, leading to adaptations in the control architecture that maintain theoretical guarantees while accommodating hardware constraints.

With the introduction of autonomous vehicles on public roads, their performance in emergency situations has become a strong focus. Collision Imminent Control (CIC) concerns the planning and control of aggressive evasive maneuvers for collision avoidance of automated vehicles. CIC is implemented using adaptive Nonlinear Model Predictive Control (NMPC), which considers obstacles and road barriers for combined trajectory re-planning and control. To achieve real-time performance, the prediction model complexity is often reduced, which can lead to an under-utilization of the control potential. The aim of CIC is to use as much of the control potential of the vehicle as possible while remaining real-time viable.

In this research, CIC is implemented using objective-based collision avoidance based on the distance to obstacles and road boundaries. Different collision avoidance formulations were derived and compared on accuracy and real-time performance. The control potential of the vehicle was further exploited by a computationally efficient vehicle model that employs differential braking. The NMPC problem is solved using Sequential Quadratic Programming (SQP) with Real Time Iterations (RTI). Different techniques that reduce the computation time were compared. Sparse solvers and variable timesteps were found to be most significant.

The robustness of the controller was improved by friction estimation. The controller is furthermore demonstrated to work on highly curved roads and in scenarios with dynamic obstacles. The controller is implemented on the hardware of a real autonomous vehicle and simulated on a closed-loop embedded system. Combining all these elements results in a CIC controller that can apply more control potential and reach control frequencies upwards of 100 Hz, increasing the level of safety in vehicle collision avoidance.

In recent years, the deployment of ground-based mobile robots has gained more and more interest in various domains. In contrast to other types of mobile robots, legged robots can traverse irregular terrains, climb stairs, and step over obstacles. However, these unique properties intensify the energy demand and require highly advanced perception methods, actuator designs, and motion control algorithms. The most significant challenges in legged robotics lie in robustness, energy efficiency, and agility.

In recent years, the integration of an articulated torso or active spine, inspired by the body motion of high-performance mammals like the cheetah, has shown promising results. Various studies observed higher maximum velocities and lower energy consumption compared to a rigid torso. However, in these studies, the compliant systems were typically controlled using basic control strategies. In recent years, the development of highly dynamic model-based motion optimization strategies has greatly enhanced the overall performance of various legged robots. Therefore, a model-based motion optimization scheme is developed specifically for articulated quadruped robots. This scheme fully exploits the additional degrees of freedom of the torso to enhance the dynamic performance of the legged robot further.

Traditionally, Event-Triggered Control (ETC) methods are sample-and-hold control schemes that implement a triggering condition in order to reduce the number of control updates. Given a decay rate of the Lyapunov function, they focus on minimizing the (average) Inter-Sample Time (IST). In this thesis, we focused on the scheduling of Periodic Event-Triggered Control (PETC) controllers. By dynamically switching between triggering conditions, we
are maximizing the average rate of decay of the Common Lyapunov Function (CLF) given a minimum Average Inter Sample Time (AIST) or burst condition.

Given the physical system, we construct a switched system which captures all possible
scheduling behaviors. The l-complete abstraction of the switched system is constructed by solving a conjunction of quadratic equations. By setting a minimum AIST or burst condition, a set of states in the abstraction is marked and a safety game is played to construct the Maximal Permissive Controller (MPC).

On the safe behaviors inside the MPC, the guaranteed minimum control performance is maximized for the infinite horizon problem, i.e. by maximizing the minimum weighted time average of the primitive cycles in the MPC. First, several energy games are played to estimate
the maximum control performance. Thereafter, a mean-payoff game is played to generate the strategy securing this maximum control performance, which is used to construct the infinite horizon controller.

Reinforcement learning, especially deep reinforcement learning, has made many advances in the last decade. Similarly, great strides have been made in multi-agent reinforcement learning. Systems of cooperative autonomous robots are increasingly being used, for which multi-agent reinforcement learning can be used as a training method. However, the curse of dimensionality remains a problem for the computational speed of the learning algorithm and the bandwidth of communication channels. This research will focus mainly on reducing the problem of overloading communication channels by trying to reduce the number of communications. This is possible since it is usually unnecessary for every agent to communicate with every other agent constantly.

To do this, we use some ideas by Daniel Jarne Ornia. The first is to reduce communications in a multi-agent reinforcement learning system by treating it as an event-triggered control problem. This method uses so-called robustness surrogates as an equivalent to a Lyapunov function to determine if a communication can be skipped without decreasing the performance more than some tolerance. The second is a method to increase the observational robustness of a policy by using lexicographic reinforcement learning.

We aim to combine these ideas and trade the additional observational robustness for decreased communications. We also want to test whether additional observational robustness can help mitigate the sim-to-real gap. We implement this method for the multi-agent deep deterministic policy gradient algorithm and perform tests on a variant of the predator-prey domain in increasingly more realistic simulations.

We found that the combination of this robust policy and the robustness surrogates method does enable the agents to achieve the same return while communicating less. Unfortunately, our research shows that the observational robustness obtained using lexicographic reinforcement learning, does not help mitigate the sim-to-real gap.

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.

Water scarcity is a persistent global issue that requires effective solutions. One increasingly popular approach to address the distribution of freshwater in different regions is the utilization of Water Irrigation Systems (WIS). Due to their vital role in ensuring people's well-being, it is crucial to design a reliable infrastructure that can maintain water levels in channels as close as possible to predefined reference points.
Extensive research papers have been dedicated to discussing various aspects of WIS control, modeling, and identification. Although simulations can yield valuable insights, it is essential to validate the results in real-life systems due to certain factors that cannot be precisely estimated, such as leakages along the channels. To address this need, a scaled-down version of a WIS at TU Delft has been developed, which effectively emulates the behavior of a real water irrigation system. Through continuous refinement and design enhancements, the simulator has proven to be a valuable tool.
Working with the scaled-down WIS has presented significant challenges, but a dedicated controller has been developed, yielding satisfactory results. The approach to controlling the WIS is simple yet effective, employing a decentralized controller that is applied to both the nonlinear simulator and the real setup. Remarkably, this approach successfully meets the predefined reference points, even in the face of applied disturbances to the system.

Localization is one of the most fundamental competencies required by an autonomous robot, providing crucial information about its position for decision-making in indoor environments. In the current literature, an indoor localization system utilizes exteroceptive sensors such as GNSS(Global Navigation Satellite System), a camera, or an ultrasonic sensor and onboard sensors such as odometry or accelerometer to observe its environment. These sensor data are typically fused using Extended Kalman Filter (EKF) techniques. To solve the challenges posed by multiple sensors, different EKF variants such as Multi-Rate EKF and Single-Rate EKF have been proposed. Additionally, localization architectures like Cascade and OWA (Ordered Weighted Averaging) have been introduced to enhance the fusion process.

This thesis aims to develop an indoor localization system for Elisa-3 robots. The proposed localization system is evaluated through simulations and real-world experiments conducted using the Elisa-3 robots platform. Furthermore, the effectiveness of the localization system is assessed in a control task involving the formation of a consensus within a robot swarm.

The simulated and experimental results indicate that using a Single-Rate Extended Kalman Filter under an OWA architecture can generate an accurate and precise trajectory across a wide range of scenarios.

This thesis investigates the potential of state-dependent sampling strategies (SDSS) for the control of heavy-haul trains. Event-triggered control (ETC) is a control approach in which data is only sent when some state-dependent condition, the triggering condition, is satisfied. In this way, the number of communications required to stabilise a system can be drastically reduced. Periodic event-triggered control (PETC) is a variant of ETC in which the triggering condition is checked periodically. By sometimes sampling earlier than a PETC-generated deadline, long-term pay-offs in the average inter-sample times are possible. As searching for formal models of early-triggering controllers quickly becomes computationally infeasible as system dimensions grow, a sample-based approach using reinforcement learning was utilised to find the SDSS controller, which takes the form of a neural network mapping states to the time until the following sample, trained to hopefully yield long-term payoffs in sample efficiency. It was found that, by choosing suitable control parameters, the SDSS controller can outperform the PETC baseline in terms of inter-sample times (IST). Next, a hardware-in-the-loop (HIL) setup was made to evaluate the optimised controller’s performance in a real-time context controlling a non-linear train system subject to noise, disturbances, and tasked with attaining different speed setpoints while minimising the inter-wagon forces. It was found that the optimised controller did not perform well during these scenarios. Since the controller’s robustness was not considered during the training process, performance suffered under these conditions. To improve the robustness of the controller, the system must be trained on realistic (noisy) data. Simulations with more wagons must be performed to assess whether improvements can still be attained when using larger models. Finally, the accuracy of the HIL setup needs to be improved by using an appropriate network stack that would allow each wagon to send its own data and turn off the sensor node radios when otherwise idly listening for incoming data.

In multi-agent systems reaching consensus has been a long-standing problem. A considerable amount of research has been focused on how event-triggered consensus can be used to limit the energy consumption of the communication system while still ensuring convergence to the neighbourhood of a point or a formation. These algorithms rely on having knowledge of the global state information of each agent in the network. Given this information, it is possible to consider using distributed algorithms from the field of Computer Science to achieve the same.

This thesis presents a study on average consensus algorithms for single-integrator multi-agent systems. The focus is on comparing the time required to reach convergence and the energy required by the communication system. To perform this study, three methods are used to compare the algorithms. The first method involves deriving mathematical bounds on the convergence time and the number of transmissions. The second involves a simple simulation to verify the mathematical bounds. Finally, the algorithms are compared in a practical application where robots are subjected to noisy measurements and need to assemble in a formation.

The results of the comparisons show that distributed algorithms are capable of fast convergence and require less energy for communication. Therefore, it is worth considering the use of distributed algorithms when agents have enough memory to temporarily store the global state of all agents in the network. There might still be situations in which the ETC algorithm converges slightly faster but this benefit is overshadowed by the increase in energy consumption.

Event-Triggered Control (ETC) is a control method where the controller is only updated when necessary. The control inputs are kept fixed until a state-dependent event triggers their re-computation. The triggering condition is designed to guarantee the stability and desired performance of the control system. This prevents the excessive use of scarce communication and energy resources. To fully take advantage of these savings when dealing with a (physically) distributed system, it is vital to decentralise the triggering condition. A decentralised triggering condition can be checked locally at the nodes of the system using the locally available states, eliminating the need to send updated sensor measurements to a central location constantly. However, the decentralisation of the triggering condition needs to be optimised for the state of the system to fully take advantage of the possible resource savings. This thesis provides a framework to optimally decentralise quadratic triggering conditions for Periodic Event-Triggered Control (PETC) systems in the absence and presence of bounded disturbances. In addition, a method of constructing a region-based map of such optimisations is provided, making it feasible to implement the optimisation approach on larger systems without needing more powerful hardware.

Since industrialization, world population is constantly on the rise, and so the demand for nutritious and healthy food is increasing as well. Traditional agriculture is steadily extended with and substituted by crop growing in greenhouses, which has better yield, more resistance against weather extremes, and overall more control over the growing parameters than regular agriculture. Greenhouses are traditionally controlled by expert growers, but currently there are not enough experienced growers in the world, and one grower can only handle a limited number of greenhouses. Autonomous greenhouse control solves this problem by taking over the task of defining setpoints for the low-level climate controllers, so that a full crop cycle can be managed with little effort from the grower’s part. Autonomous climate control in greenhouses can be extended by automatic irrigation, which reduces the workload manual calculation of irrigation decisions puts on the growers. Furthermore, it has the possibility of providing better yield than manual control during a crop cycle while reducing the water usage. This thesis introduces a new predictive irrigation control approach, which uses forecast weather data and climate predictions to create irrigation decisions with the use of a Mixed Logical Dynamical (MLD) Model Predictive Control (MPC) algorithm. The behaviour of the controller can be changed through costs and constraints, which can be defined according to the need of greenhouse growers. The water balance model, constituting from a simple Plant Water Uptake (PWU) model from literature as well as a novel drain model, is used to predict the water content of the growing substrate accurately for 24 hours ahead. The MLD MPC exploits the structure of nonlinearities inside the water balance model to create a Mixed Integer Linear Programming (MILP) optimal control problem, which can be solved using efficient algorithms with guarantee on optimality. The validation results on data of multiple greenhouses show, that the created algorithm can be generalized with little effort. The effectiveness of the control algorithm in open-loop was inspected through simulations.

Recent developments in \ac{ML} have paved the way for unprecedented possibilities in the field of data analytics in numerous team sports, such as American football, baseball, and basketball. In more recent years, \ac{ML} techniques have been applied to football as professional teams got inspired to collect enormous quantities of data to evaluate the performance of their players and their applied tactics.
However, research in performance analytics has primarily focused on the development of offline evaluation techniques, while online evaluation of a game has remained largely unexplored.
To attract the attention of executive decision-makers in football, research requires the creation of comprehensive metrics that can give answers to the continuous calls of coaches and trainers to evaluate within-game performance.
This work therefore aimed at finding an online evaluation technique to capture the complex aspects and highly non-linear dynamics of the sport in an interpretable way. \\
To achieve this goal, an online predictive model was developed using an \ac{ANN} approach, which is able to give an instantaneous value to every moment in the game through the computation of the \ac{ED}. The \ac{ED} is a comprehensive metric created in this work, which defines a form of relative, goal-related danger and encapsulates the complex spatio-temporal characteristics of the 22 players on the pitch.
The \acp{ANN} have been trained and evaluated on two newly created datasets of simulated data which contain a combined total of 72 games and include the tracking data of the players and ball as well information on in-game events such as shots, passes, and goals.
The extensiveness of the datasets enabled the creation of a novel \ac{MDP} that describes the complicated dynamics of the system.
The developed \ac{MDP} model is a crucial component of this work and served as the basis for the design of the \ac{ED} and allowed the creation of the \ac{ANN}-based prediction model.
The prediction pipeline was tested over five different prediction horizons, ranging from 5 to 45 minutes. Accurate predictive performance was achieved over all horizons,
thereby showing the \ac{ANN} architecture was able to account for all the nuances within the wide range of games of the datasets.
The predictor functions were also tested on a small, additional dataset of unknown opponents to evaluate online performance and test robustness against unknown opponents. Experiments on this dataset demonstrated that reasonably accurate predictions can be made as long as the prediction horizon is not larger than 15 minutes. \\
Results achieved in this work show that a \ac{ANN}-based approach to online predictive modelling can achieve accurate results. However, future research should focus on validating the performance of the \acp{ANN} on a more extensive dataset of unknown opponents and, ultimately, on a dataset of real-life football games. In addition, the online prediction pipeline of this work can be extended to an online decision-making model with the purpose of changing tactics dynamically during the game.

Control engineering researchers are increasingly embracing data-driven techniques like reinforcement learning for control and optimisation. An example of a case where reinforcement learning could be useful is the synthesis of near-optimal sampling strategies for self-triggered control. Self-triggered control is an aperiodic control method that aims to reduce the number of communications between the controller and sensors in a control loop, by predicting when some triggering condition is met and only transmitting a sample accordingly. Recent research has shown that greedily following the proposed sampling times can result in sub-optimal long-term average inter-sample times. Abstraction-based methods have been able to synthesise sampling strategies that result in better long-term average inter-sample times, by allowing for early sampling and considering the proposed sampling times as deadlines. However, these abstraction-based methods suffer greatly from the curse of dimensionality in the form of combinatorial explosion, which limits their practicality for more complex systems. This thesis proposes a novel deep reinforcement learning tool for finding near-optimal sampling strategies for self-triggered control of LTI systems. The proposed tool is evaluated and compared to a state-of-the-art abstraction-based method. The proposed tool is shown to match the performance of the abstraction-based method for smaller systems, while still achieving good results on more complex systems that prohibit the use of abstraction-based methods.

With over 75,000 attributed deaths in 2020, opioid overdose is now the leading cause of death amongst Americans under the age of 50. With record-breaking increases in fatalities for the past 10 years, fentanyl in particular has many experts worried about the future of the opioid epidemic. To reverse opioid overdose there exists a drug called Naloxone, an opioid antagonist that completely restores the subject’s vitality within minutes. As the US opioid epidemic is surging by the increase of synthetic opioid-related overdoses, current methods of Naloxone administration are unsuccessful in suppressing the rise. One large obstacle in the rate of success for Naloxone is related to the fact that opioid users are not able to administer the antagonist to themselves due to their overdosing symptoms, which includes unconsciousness. Therefore, to save a life with Naloxone, one has to rely on a bystander that is present during the overdose, recognizes it, possesses Naloxone, and is able to administer it in time. The rising death toll of opioid overdose reveals that one of these conditions are often not met. This thesis presents a fully implantable device that is capable of autonomously detecting and treating opioid overdose to provide an alternative solution that takes the said bystander out of the equation. To achieve this, three different sensors (an Inertial Measurement Unit (IMU), Electrocardiography (ECG), and Photoplethysmography (PPG) sensor) are used for the monitoring of heart rate and respiratory rate. These vital sings are continuously fed into an advanced decision-making system that is capable of detecting opioid overdose within 81 seconds. Once detected, instructions are sent to a novel drug delivery pump that is capable of delivering high volumes of Naloxone within only 3.25 seconds, while outperforming similar technologies in terms of energy efficiency. The proposed subcutaneous implantable device can be refilled through a refill port, be wirelessly charged, and has a battery capacity that lasts up to 78 days. It is concluded that the proposed closed-loop drug delivery system is a feasible and effective autonomous tool that could complement existing Naloxone administration solutions in the ongoing fight against the opioid epidemic.

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.