This thesis is concerned with model predictive control (MPC) within the field of autonomous driving. MPC requires a model of the system to be controlled. Since a vehicle is expected to handle a wide range of driving conditions, it is crucial that the model of the vehicle dynamics is able to account for this. Differences in road grip caused by snowy, icy or muddy roads change the driving dynamics and relying on a single model, based on ideal conditions, could possibly lead to dangerous behaviour. This work investigates the use of Gaussian processes for learning a model that can account for varying road friction coefficients. This model is incorporated as an extension to a nominal vehicle model. A double lane change scenario is considered and the aim is to learn a GP model of the disturbance based on previous driving experiences with a road friction coefficient of 0.4 and 0.6 performed with a regular MPC controller. The data is then used to train a GP model. The GPMPC controller is then compared with the regular MPC controller in the case of trajectory tracking. The results show that the obtained GP models in most cases correctly predict the model error in one prediction step. For multi-step predictions, the results vary more with some cases showing an improved prediction with a GP model compared to the nominal model. In all cases, the GPMPC controller gives a better trajectory tracking than the MPC controller while using less control input.

This paper presents the mathematical modeling and the design of an optimal pressure tracking controller for an often used setup in pneumatic applications. Two pneumatic chambers are connected with a pneumatic tube. The pressure in the second chamber is to be controlled using two switching valves connected to the first chamber and based on the pressure measurement in the first chamber. The optimal control problem is formulated and solved using the MPC framework. The designed controller shows good tracking quality, while fulfilling hard constraints, like maintaining the pressure below a given upper bound.

<p>Ninety-five percent of all roadway crashes are attributed fully or partially to human error, and a multitude of safety-related programs, policies, and initiatives have seen limited success in reducing roadway crashes and their accompanying fatalities, injuries, and property damage. For this reason, safety professionals have lauded the emergence of autonomous vehicles (AVs) as a promising palliative to the persistent problem of road crashes. Such optimism is reflected in recent literature that have argues from a conceptual standpoint, that road safety enhancement will be one of the prospective benefits of AV operations because automation removes humans from vehicle driving operations and therefore criminates or mitigates human error. It can be argued that the safety benefits of AVs will be manifest when AV market penetration reaches 100%. However, it seems clear from a practical standpoint that the transition from a system of exclusively human-driven vehicles (HDVs) to that of exclusively AVs will not only be necessary but also an arduous journey. This transition period will be characterized by heterogeneous traffic, where human-driven vehicles (HDVs) and AVs share the road space, and whence the prospective safety benefits of AVs may not be fully realized due to human error arising from the HDV operations in the mixed traffic space. These traffic conflicts, which may lead to collisions, could arise from any of several contexts of driving maneuvers, one of which is aggressive lane changes, the focus of this thesis. From the literature, it is clear that lane changing is inherently more collision-prone compared to most other maneuvers including car following, and therefore the consequences of errant human driving behavior such as inattention of misjudgment during lane changing, are more severe. To address this problem, this thesis developed a control framework to be used by AVs to help them avoid collision in a mixed traffic stream with human drivers who exhibit aggressive lane-changing behavior. The developed framework, which is based on a Model Predictive Control (MPC) approach, is designed to control the AV’s movements safely by duly accommodating potential human error from the HDVs that could otherwise lead to any of two common collision patterns: rear-end and side-impact. Further, the thesis investigated how connectivity between the HDVs, and AVs could facilitate joint operational decision-making and sharing of real-time information, thereby further enhancing the safety of the entire traffic stream. Finally, the thesis presents the results of driving simulations carried out to test and validate the performance of the control framework under different traffic conditions.</p>

The operation of groups of heavy-duty vehicles at small inter-vehicular distances, known as platoons, lowers the overall aerodynamic drag and, therefore, reduces fuel consumption and greenhouse gas emissions. Experimental tests conducted on a flat road and without traffic have shown that platooning has the potential to reduce the fuel consumption up to 10%. However, platoons are expected to drive on public highways with varying topography and traffic. Due to the large mass and limited engine power of heavy-duty vehicles, road slopes can have a significant impact on feasible and optimal speed profiles. Therefore, maintaining a short inter-vehicular distance without coordination can result in inefficient or even infeasible speed trajectories. Furthermore, external traffic can interfere by affecting fuel-efficiency and threatening the safety of the platooning vehicles. This thesis addresses the problem of safe and fuel-efficient control for heavy-duty vehicle platooning. We propose a hierarchical control architecture that splits this complex control problem into two layers. The layers are responsible for the fuel-optimal control based on look-ahead information on road topography and the real-time vehicle control, respectively. The top layer, denoted the platoon coordinator, relies on a dynamic programming framework that computes the fuel-optimal speed profile for the entire platoon. The bottom layer, denoted the vehicle control layer, uses a distributed model predictive controller to track the optimal speed profile and the desired inter-vehicular spacing policy. Within this layer, constraints on the vehicles' states guarantee the safety of the platoon. The effectiveness of the proposed controller is analyzed by means of simulations of several realistic scenarios. They suggest a possible fuel saving of up to 12% for the follower vehicles compared to the use of existing platoon controllers. Analysis of the simulation results shows how the majority of the fuel saving comes from a reduced usage of vehicles brakes. A second problem addressed in the thesis is model predictive control for obstacle avoidance and lane keeping for a passenger car. We propose a control framework that allows to control the nonlinear vehicle dynamics with linear model predictive control. The controller decouples the longitudinal and lateral vehicle dynamics into two successive stages. First, plausible braking and throttle profiles are generated. Second, for each profile, linear time-varying models of the lateral dynamics are derived and used to formulate a collection of linear model predictive control problems. Their solution provides the optimal control input for the steering and braking actuators. The performance of the proposed controller has been evaluated by means of simulations and real experiments. / <p>QC 20150911</p>

This thesis presents the work of two significant projects. In the first project, a suite of benchmark problems for grid energy management are presented which demonstrate several issues characteristic to the dynamic optimization of these systems. These benchmark problems include load following, cogeneration, tri-generation, and energy storage, and each one assumes perfect foresight of the entire time horizon. The Gekko Python package for dynamic optimization is introduced and two different solution methods are discussed and applied to solving these benchmarks. The simultaneous solve mode out-performs the sequential solve mode in each benchmark problem across a wide range of time horizons with increasing resolution, demonstrating the ability of the simultaneous mode to handle many degrees of freedom across a range of problems of increasing difficulty. In the second project, combined optimization of propulsion system design, flight trajectory planning and battery mass optimization is applied to solar-regenerative high-altitude long-endurance (SR-HALE) aircraft through a sequential iterative approach. This combined optimization approach yields an increase of 20.2% in the end-of-day energy available on the winter solstice at 35°N latitude, resulting in an increase in flight time of 2.36 hours. The optimized flight path is obtained by using nonlinear model predictive control to solve flight and energy system dynamics over a 24 hour period with a 15 second time resolution. The optimization objective is to maximize the total energy in the system while flying a station-keeping mission, staying within a 3 km radius and above 60,000 ft. The propulsion system design optimization minimizes the total energy required to fly the optimal path. It uses a combination of blade element momentum theory, blade composite structures, empirical motor and motor controller mass data, as well as a first order motor performance model. The battery optimization seeks to optimally size the battery for a circular orbit. Fixed point iteration between these optimization frameworks yields a flight path and propulsion system that slightly decreases solar capture, but significantly decreases power expended. Fully coupling the trajectory and design optimizations with this level of accuracy is infeasible with current computing resources. These efforts show the benefits of combining design and trajectory optimization to enable the feasibility of SR-HALE flight.

Research about unmanned ground vehicles (UGVs) has received an increased amount of attention in recent years, partly due to the many applications of UGVs in areas where it is inconvenient or impossible to have human operators, such as in mines or urban search and rescue. Two closely linked problems that arise when developing such vehicles are motion planning and control of the UGV. This thesis explores these subjects for a UGV with an unknown, and possibly time-variant, dynamical model. A framework is developed that includes three components: a machine learning algorithm to estimate the unknown dynamical model of the UGV, a motion planner that plans a feasible path for the vehicle and a controller making the UGV follow the planned path. The motion planner used in the framework is a lattice-based planner based on input sampling. It uses a dynamical model of the UGV together with motion primitives, defined as a sequence of states and control signals, which are concatenated online in order to plan a feasible path between states. Furthermore, the controller that makes the vehicle follow this path is a model predictive control (MPC) controller, capable of taking the time-varying dynamics of the UGV into account as well as imposing constraints on the states and control signals. Since the dynamical model is unknown, the machine learning algorithm Bayesian linear regression (BLR) is used to continuously estimate the model parameters online during a run. The parameter estimates are then used by the MPC controller and the motion planner in order to improve the performance of the UGV. The performance of the proposed motion planning and control framework is evaluated by conducting a series of experiments in a simulation study. Two different simulation environments, containing obstacles, are used in the framework to simulate the UGV, where the performance measures considered are the deviation from the planned path, the average velocity of the UGV and the time to plan the path. The simulations are either performed with a time-invariant model, or a model where the parameters change during the run. The results show that the performance is improved when combining the motion planner and the MPC controller with the estimated model parameters from the BLR algorithm. With an improved model, the vehicle is capable of maintaining a higher average velocity, meaning that the plan can be executed faster. Furthermore, it can also track the path more precisely compared to when using a less accurate model, which is crucial in an environment with many obstacles. Finally, the use of the BLR algorithm to continuously estimate the model parameters allows the vehicle to adapt to changes in its model. This makes it possible for the UGV to stay operational in cases of, e.g., actuator malfunctions.

No description available.

Le travail de thèse décrit dans ce manuscrit concerne les Systèmes Avancés d’Aide à la Conduite (ADAS) qui sont devenus de nos jours un axe de recherche stratégique chez de nombreux constructeurs automobiles. Ce type de systèmes peuvent être considérés comme la première génération de dispositifs de conduite assistée ou semi-autonome et qui ouvrira la voie aux véhicules pleinement autonomes. La première partie de ce manuscrit concerne l’analyse et la commande pour les applications de contrôle de la dynamique latérale du véhicule – autoguidage par suivi de cible et aide au maintien au centre de la voie (LCA). Dans ce cadre, la sécurité joue un rôle clé, mettant en lumière la mise en oeuvre différentes techniques de commande contrainte pour des modèles linéaires à paramètres variants (LPV). La commande prédictive (MPC) et la commande par interpolation (IBC) ont été sélectionnés dans ce travail. De plus, la conception d’un système de commande robuste qui assure un comportement correct malgré la variation des paramètres du système ou la présence d’incertitudes est une caractéristique critique. Les outils de la théorie de l’invariance positive robuste (RPI) sont pris en considération pour la conception de stratégies de commande robustes LPV par rapport aux larges variations de la vitesse véhicule et aux changements de courbure de la route. Le second axe de cette thèse est la planification optimale de trajectoire pour les manouvres de dépassement et de changement de voie sur autoroute, avec réduction des risques de collision. Pour atteindre cet objectif, la description exhaustive des scénarios possible est présentée, permettant de formuler un problème d’optimisation qui maximise le confort du conducteur et assure la satisfaction des contraintes du système. / The thesis work contained in this manuscript is dedicated to the Advanced Driving Assistance Systems, which has become nowadays a strategic research line in many car companies. This kind of systems can be seen as a first generation of assisted or semi-autonomous driving, that will set the way to fully automated vehicles. The first part focuses on the analysis and control of lateral dynamics control applications - Autosteer by target tracking and the Lane Centering Assistance System (LCA). In this framework, safety plays a key role, bringing into focus the application of different constrained control techniques for linear parametervarying (LPV) models. Model Predictive Control (MPC) and Interpolation Based Control (IBC) have been the selected ones in the present work. In addition, it is a critical feature to design robust control systems that ensure a correct behavior under system's variation of parameters or in the presence of uncertainty. Robust Positive Invariance (RPI) theory tools are considered to design robust LPV control strategies with respect to large vehicle speed variations and curvature of the road changes. The second axis of this thesis is the optimization-based trajectory planning for overtaking and lane change in highways with anti-collision enhancements. To achieve this goal, an exhaustive description of the possible scenarios that may arise is presented, allowing to formulate an optimization problem which maximizes passenger comfort and ensures system constraints' satisfaction.

In model predictive control an optimization problem is solved in every time step, which in real-time applications has to be solved within a limited time frame. When applied on embedded hardware in fast changing systems it is important to use efficient solvers and crucial to guarantee that the optimization problem can be solved within the time frame. In this thesis a path following controller which follows a motion plan given by a motion planner is implemented to steer a truck and trailer system. To solve the optimization problems which in this thesis are quadratic programs the three different solvers DAQP, qpOASES and OSQP are employed. The computational time of the active-set solvers DAQP, qpOASES and the operator splitting solver OSQP are compared, where the controller using DAQP was found the fastest and therefore most suited to use in this application of real-time model predictive control.  A certification framework for the active-set method is used to give complexity guarantees on the controller using DAQP. The exact worst-case number of iterations when the truck and trailer system is following a straight path is presented. Furthermore, initial experiments show that given enough computational time/power the exact iteration complexity can be determined for every possible quadratic program that can appear in the controller.

This compilation thesis presents an overarching framework on the utilization of nonlinear model predictive control(NMPC) for various applications in the context of Unmanned Aerial Vehicle (UAV) path planning and collision avoidance. Fast and novel optimization algorithms allow for NMPC formulations with high runtime requirement, as those posed by controlling UAVs, to also have sufficiently large prediction horizons as to in an efficient manner integrate collision avoidance in the form of set-exclusion constraints that constrain the available position-space of the robot. This allows for an elegant merging of set-point reference tracking with the collision avoidance problem, all integrated in the control layer of the UAV. The works included in this thesis presents the UAV modeling, cost functions, constraint definitions, as well as the utilized optimization framework. Additional contributions include the use case on multi-agent systems, how to classify and predict trajectories of moving (dynamic) obstacles, as well as obstacle prioritization when an aerial agent is in the precense of more obstacles, or other aerial agents, than can reasonably be defined in the NMPC formulation. For the cases of dynamic obstacles and for multi-agent distributed collision avoidance this thesis offers extensive experimental validation of the overall NMPC framework. These works push the limits of the State-of-the-Art regarding real-time real-life implementations of NMPC-based collision avoidance. The works also include a novel RRT-based exploration framework that combines path planning with exploration behavior. Here, a multi-path RRT * planner plans paths to multiple pseudo-random goals based on a sensor model and evaluates them based on the potential information gain, distance travelled, and the optimimal actuation along the paths.The actuation is solved for as as the solutions to a NMPC problem, implying that the nonlinear actuator-based and dynamically constrained UAV model is considered as part of the combined exploration plus path planning problem. To the authors best knowledge, this is the first time the optimal actuation has been considered in such a planning problem. For all of these applications, the utilized optimization framework is the Optimization Engine: a code-generation framework that generates a custom Rust-based solver from a specified model, cost function, and constraints. The Optimization Engine solves general nonlinear and nonconvex optimization problems, and in this thesis we offer extensive experimental validation of the utilized Proximal-Averaged Newton-type method for Optimal Control (PANOC) algorithm as well as both the integrated Penalty Method and Augmented Lagrangian Method for handling the nonlinear nonconvex constraints that result from collision avoidance problems.