Traffic congestion remains a critical challenge with profound economic, safety, and environmental implications, exacerbated by the continuous growth of global population and vehicle densities. The expansion of road infrastructure is often impractical due to excessive costs and spatial constraints. In response, intelligent traffic management strategies have gained prominence, leveraging advanced control methodologies to optimize flow and mitigate congestion. Among these, Model Predictive Control (MPC) has been extensively applied due to its capacity to handle complex, constrained systems. However, the accuracy of the underlying prediction models inherently determines its effectiveness. Traditional system identification methods can have inherent uncertainties and are often only accurate to a certain confidence level, leading to potential performance degradation in dynamic traffic conditions. To overcome this limitation, learning-based MPC integrates machine learning techniques, particularly Reinforcement Learning (RL), that can enhance adaptability and closed-loop performance. This integration can result in more responsive and robust traffic management systems capable of handling uncertainties and dynamic environments.

The integration of MPC and RL has emerged as a compelling approach to control complex nonlinear systems, capitalizing on the strengths of both methodologies. RL constantly improves control policies based on observed system dynamics and performance feedback, while MPC automatically chooses the best control actions over a limited prediction horizon while still enforcing system constraints. This thesis investigates the integration of RL within the MPC framework for highway traffic flow optimization, specifically focusing on the dynamic regulation of Variable Speed Limits (VSL) and Ramp Metering (RM). The proposed approach is evaluated against three benchmark models: a theoretically optimal MPC model with perfect system knowledge, assuming exact parameters; an MPC controller with an imperfect prediction model, where these parameters deviate from their true values, leading to suboptimal control decisions; and the MPC-RL model, which dynamically adjusts the learnable parameters during training to improve overall system performance.

Optimal control strategies are often combined with safety certificates to ensure both performance and safety in safety-critical systems. A prominent example is combining Model Predictive Control (MPC) with Control Barrier Functions (CBF). Yet, tuning MPC parameters and choosing an appropriate class kappa function in the CBF is challenging and problem dependent. This thesis introduces a safe model-based Reinforcement Learning (RL) framework where a parameterized MPC incorporates a CBF with a parameterized class kappa function and serves as a function approximator to learn improved safe control policies. Three variations are introduced, distinguished by the way the class kappa function is parameterized. The Learnable Optimal Decay CBF (LOPTD-CBF) extends the Optimal Decay CBF by allowing RL to tune the optimal decay parameters, improving performance while enhancing constraint feasibility and preserving safety guarantees. The Neural Network CBF (NN-CBF) parametrizes the decay term of a discrete exponential CBF with a neural network, enabling richer state-dependent safety conditions. Finally, the Recurrent Neural Network CBF (RNN-CBF) extends the NN-CBF with a recurrent architecture to handle time-varying CBF constraints, such as moving obstacles. Numerical experiments on a discrete double-integrator with static and dynamic obstacles demonstrate that the proposed methods improve performance while ensuring safety, each offering distinct trade-offs in performance, feasibility and complexity.

Greenhouses allow production of crops that would otherwise be impossible. Permitting more local, fresher and nutrient richer crop production. Eorts are taken to minimize societal harm due to energy and resource consumption by greenhouse production systems. One way to control such systems is by using model predictive control. Optimal crop yield and resource eciency can, in theory, be achieved by model predictive control. Unfortunately, one major drawback of model predictive control is that it is not well equipped to deal with parametric uncertainty. Significant prediction errors can occur when a mismatch between the model and the real system exists, resulting in deteriorated performance of the system. Strategies exist, such as robust MPC, that are designed to handle uncertainty, but those often result in conservative control policies. This thesis proposes to use model predictive control as a function approximator for RL in order to learn values for model and MPC parameters that can deliver optimal performance in the case of model mismatch.
In this thesis, data-driven economic nonlinear model predictive control using Q-learning is proposed as a method to alter the model parameters. The performance of the system af- ter learning is compared to approaches using robust and nominal model predictive control. Three dierent goals are determined: maximizing economic profit, minimizing the constraint violations and maximizing the economic performance while minimizing constraint violations.
In this work, an ENMPC scheme is used as a function approximator in a Q-learning envi- ronment. The optimization solution from the ENMPC scheme is used as the input to the system, while the Q-learning agent optimizes the parameter values of the ENMPC scheme and model for the environment. The performance of the system after learning is compared to approaches using robust and nominal model predictive control. The simulation results show that the data-driven ENMPC using reinforcement learning is able to decrease constraint vi- olations by up to 94%, but unable to increase economic performance compared to nominal MPC, compared to robust MPC the EPI is increased by almost 10% while keeping constraint violations at a similar level.