Collision avoidance design plays an essential role in autonomous vehicle technology. It's an attractive research area that will need much experimentation in the future. This research area is very important for providing the maximum safety to automated vehicles, which have to be tested several times under diFFerent circumstances for safety before use in real life. This thesis proposes a method for designing and presenting a collision avoidance maneuver by using a model predictive controller with a moving obstacle for automated vehicles. It consists of a plant model, an adaptive MPC controller, and a reference trajectory. The proposed strategy applies a dynamic bicycle model as the plant model, adaptive model predictive controller for the lateral control, and a custom reference trajectory for the scenario design. The model was developed using the Model Predictive Control Toolbox and Automated Driving Toolbox in Matlab. Builtin tools available in Matlab/Simulink were used to verify the modeling approach and analyze the performance of the system. The major contribution of this thesis work was implementing a novel dynamic obstacle avoidance control method for automated vehicles. The study used validated parameters obtained from previous research. The novelty of this research was performing the studies using a MPC based controller instead of a sliding mode controller, that was primarily used in other studies. The results obtained from the study are compared with the validated models. The comparisons consisted of the lateral overlap, lateral error, and steering angle simulation results between the models. Additionally, this study also included outcomes for the yaw angle. The comparisons and other outcomes obtained in this study indicated that the developed control model produced reasonably acceptable results and recommendations for future studies.

This report summarizes the research results of TO4233, "Fault Tolerant Autonomous Lateral Control for Heavy Vehicles". This project represents a continuing effort of PATH's research on Automated Highway Systems (AHS) and more specifically in the area of heavy vehicles. Research on the lateral control of heavy vehicles for AHS has been going on at PATH since 1993. MOU129, "Steering and Braking Control of Heavy Duty Vehicles" was the first project and it was followed by MOU242, "Lateral Control of Commercial Heavy Duty Vehicle". Both projects were concerned mostly with the theoretical portion of the problem, i.e. model development, analysis of the dynamic model from the lateral control point of view, and the lateral controller designs. The first experimental results were shown in MOU289 (MOU313), "Lateral Control of Heavy Duty Vehicles for Automated Highway Systems", where the theoretical model was validated and calibrated to the dynamic behavior of an actual tractor-semitrailer vehicle, which was obtained and instrumented. In addition, preliminary closed-loop experiments were performed. A more comprehensive study on a large variety of control strategies was presented in MOU385 and TO4201, "Robust Lateral Control of Heavy Duty Vehicles". More specifically, three types of nonlinear and adaptive controllers for lateral control of heavy vehicles were analyzed theoretically and compared experimentally. All the research efforts mentioned above have been extremely valuable for the development of automated highway vehicles; however they assume the existence of a fully operational magnet-magnetometer scheme. To be more specific, all the results are based on the assumption that each heavy vehicle is equipped with two banks of magnetic sensors, one mounted on the front bumper and the other mounted on the rear bumper of the trailer. The road is also implanted with equally spaced magnets whose magnetic field is used to measure the vehicle's lateral deviation from the road centerline ("lane-keeping control"). Up to now, no heavy-vehicle-related report has discussed the case of vehicle lateral performance under the existence of faults. This problem is very important, since safety and reliability are the primary requirements for the success of AHS. This report addresses the problem of fault tolerant control of heavy vehicles by proposing a secondary system that implements "autonomous vehicle following" instead of "lanekeeping".

Passivity-based Model Predictive Control for Mobile Vehicle Navigation represents a complete theoretical approach to the adoption of passivity-based model predictive control (MPC) for autonomous vehicle navigation in both indoor and outdoor environments. The brief also introduces analysis of the worst-case scenario that might occur during the task execution. Some of the questions answered in the text include: • how to use an MPC optimization framework for the mobile vehicle navigation approach; • how to guarantee safe task completion even in complex environments including obstacle avoidance and sideslip and rollover avoidance; and • what to expect in the worst-case scenario in which the roughness of the terrain leads the algorithm to generate the longest possible path to the goal. The passivity-based MPC approach provides a framework in which a wide range of complex vehicles can be accommodated to obtain a safer and more realizable tool during the path-planning stage. During task execution, the optimization step is continuously repeated to take into account new local sensor measurements. These ongoing changes make the path generated rather robust in comparison with techniques that fix the entire path prior to task execution. In addition to researchers working in MPC, engineers interested in vehicle path planning for a number of purposes: rescued mission in hazardous environments; humanitarian demining; agriculture; and even planetary exploration, will find this SpringerBrief to be instructive and helpful.

This thesis presents disturbance estimators and controllers for autonomous vehicles. In particular, it focuses on a longitudinal distance controller and a lateral lane keeping controller. First, in order to estimate road bank angle as a disturbance term in the lane keeping controller, a kinematic relationship between road shape and sensor measurements was proposed. Utilizing longitudinal and lateral vehicle dynamics, longitudinal road gradient and lateral road bank angle were estimated simultaneously using the Unscented Kalman Filter (UKF) approach. Second, a lane keeping controller associated with the road bank angle estimator was proposed. For the controller, a steady state dynamic vehicle model was derived to describe lateral vehicle dynamics. A Receding Horizon Sliding Control (RHSC) approach was implemented to guarantee simple formulation and easy constraint consideration for the receding horizon technique. For the longitudinal control systems, the front vehicle's future motion was considered as a disturbance term in a longitudinal distance controller for the ego vehicle. To predict the motion, a new car-following model was proposed. To extract the current front vehicle driver's driving style, a driver aggressivity factor was derived and estimated in real-time through the UKF approach. Adopting a base car-following model and an aggressivity factor estimator on the front vehicle, the front vehicle's future motion sequence was propagated. Furthermore, as a distance controller associated with the front vehicle's future motion, a Fuel Eciency Adaptive Cruise Control (ACC) was presented. A new fuel consumption model was included in the optimization problem in order to improve fuel eciency. The nonlinear Model Predictive Control approach was applied to the controller, and the front vehicle's future motion was considered in the prediction horizon. Two disturbance estimators for longitudinal and lateral motion were veried under simulation and real vehicle tests in real-time. The lane keeping controller was proven to have better performance with the bank angle estimator on public roads. Furthermore, for a distance controller, fuel economy using a Fuel Eciency ACC has been veried in simulation.

Each year in the United States, thousands of lives are lost as a result of loss of control crashes. Production driver assistance systems such as electronic stability control (ESC) have been shown to be highly effective in preventing many of these automotive crashes, yet these systems rely on a sensor suite that yields limited information about the road conditions and vehicle motion. Furthermore, ESC systems rely on gains and thresholds that are tuned to yield good performance without feeling overly restrictive to the driver. This dissertation presents an alternative approach to providing stabilization assistance to the driver which leverages additional information about the vehicle and road that may be obtained with advanced estimation techniques. This new approach is based on well-known and robust vehicle models and utilizes phase plane analysis techniques to describe the limits of stable vehicle handling, alleviating the need for hand tuning of gains and thresholds. The resulting state space within the computed handling boundaries is referred to as a safe handling envelope. In addition to the boundaries being straightforward to calculate, this approach has the benefit of offering a way for the designer of the system to directly adjust the controller to accomodate the preferences of different drivers. A model predictive control structure capable of keeping the vehicle within the safe handling boundaries is the final component of the envelope control system. This dissertation presents the design of a controller that is capable of smoothly and progressively augmenting the driver steering input to enforce the boundaries of the envelope. The model predictive control formulation provides a method for making trade-offs between enforcing the boundaries of the envelope, minimizing disruptive interventions, and tracking the driver's intended trajectory. Experiments with a steer-by-wire test vehicle demonstrate that the model predictive envelope control system is capable of operating in conjunction with a human driver to prevent loss of control of the vehicle while yielding a predictable vehicle trajectory. These experiments considered both the ideal case of state information from a GPS/INS system and an a priori friction estimate as well as a real-world implementation estimating the vehicle states and friction coefficient from steering effort and inertial sensors. Results from the experiments demonstrated a controller that is tolerant of vehicle and tire parameterization errors and works well over a wide range of conditions. When real time sensing of the states and friction properties is enabled, the results show that coupling of the controller and estimator is possible and the model predictive control structure provides a mechanism for minimizing undesirable coupled dynamics through tuning of intuitive controller parameters. The model predictive control structure presented in this dissertation may also be considered as a general framework for vehicle control in conjunction with a human driver. The structure utilized for envelope control may also be used to restrict other vehicle states for safety and stability. Results are presented in this dissertation to show that a model predictive controller can coordinate a secondary actuator to alter the planar states and reduce the energy transferred into the roll modes of the vehicle. The systematic approach to vehicle stabilization presented in this dissertation has the potential to improve the design methodology for future systems and form the basis for the inclusion of more advanced functions as sensing and computing capabilities improve. The envelope control system presented here offers the opportunity to advance the state of the art in stabilization assistance and provides a way to help drivers of all skill levels maintain control of their vehicle.

This thesis addresses the problem of controlling the lateral motion of an autonomous vehicle in the presence of parametric uncertainties, disturbances, and hard nonlinearities in the steering system, such as backlash in gears, stiction, hysteresis, and dead zones. The lateral motion of an autonomous vehicle is controlled by two cascaded controllers, the trajectory tracking controller and the steering angle controller. This thesis focuses on the development of both controllers using robust and adaptive control techniques. Two control strategies are developed to control the electric power steering angle, sliding mode control and adaptive backstepping control. The limitation of sliding mode control is first addressed, which is the chattering phenomena, and then a proposed methodology is presented to solve it using variable gain sliding mode control. Self-aligning moment actas as disturbance on the steering system that the controller has to compensate for. A model-based approach to estimate it is first developed and its limitations are addressed, which is tire parameters dependence. Two other approaches are then developed to overcome these limitations, the first one is a sliding mode observer, and the second one is part of a backstepping controller. Two approaches are developed to control the vehicle lateral trajectory, non-adaptive backstepping and adaptive backstepping. The extended matching design procedure is used in the adaptive backstepping controller to avoid the overestimation problem. Road curvature must be accurately know by the controller to follow the planned trajectory. It is usually measured by a camera, but the quality of the measurement is affected by environmental factors. An adaptive law is developed to estimate the road curvature online as part of an adaptive backstepping controller. Two feedforward approaches are presented to compensate for road curvature, one is derived from steady state vehicle lateral dynamics, and another is based on estimating the transfer function dynamics from road curvature to steering angle. Road bank angle is a significant disturbance in vehicle lateral control systems. A vehicle lateral state and disturbance observer is developed to estimate the road bank angle and the vehicle side slip angle, which are expensive to measure in current road vehicles, using extended Kalman filter. The observer combines a dynamical vehicle model with two measurements from inexpensive sensors.