Mariana Netto

Time to line crossing for lane departure avoidance: a theoretical study and an experimental setting

The main goal of this paper is to develop a distance to line crossing (DLC) based computation of time to line crossing (TLC). Different computation methods with increasing complexity are provided. A discussion develops the influence of assumptions generally assumed for approximation. A sensitivity analysis with respect to vehicle parameters and positioning is performed. For TLC computation, both straight and curved vehicle paths are considered. The road curvature being another important variable considered in the proposed computations, an observer for its estimation is then proposed. An evaluation over a digitalized test track is first performed. Real data are then collected through an experiment carried out in test tracks with the equipped prototype vehicle. Based on these real data, TLC is then computed with the theoretically proposed methods. The obtained results outlined the necessity to take into consideration vehicle dynamics to use the TLC as a lane departure indicator.

Driver Steering Assistance for Lane-Departure Avoidance Based on Hybrid Automata and Composite Lyapunov Function

This paper presents the design and the practical implementation of vehicle steering assistance that helps the driver avoid unintended lane departure. A switching strategy is built to govern the driver-assistance interaction, and the resulting hybrid system is formalized as an input/output (I/O) hybrid automaton. Composite Lyapunov functions, polyhedral-like invariant sets, and linear matrix inequality (LMI) methods constitute the heart of the approach used to design the lane-departure avoidance (LDA) system. The practical implementation of this steering assistance in a prototype vehicle confirms the effectiveness of this approach.

H/sub /spl infin//, adaptive, PID and fuzzy control: a comparison of controllers for vehicle lane keeping

We provide, in this paper an overview as well as a comparison of four controllers for the vehicle lateral control problem. H/sub /spl infin//, adaptive, fuzzy and PID controllers are compared by simulations over a test track circuit. Curvature and wind perturbations as well as variations on the speed and on the adherence coefficient are introduced in the simulations.

Vehicle lateral dynamics estimation using unknown input proportional-integral observers

This paper proposes a robust state and fault estimation of vehicle lateral dynamics by means of proportional integral observers. The observer is able to estimate the state and any constant or varying fault signals even in presence of road bank, sensor noise or offset. A framework for common road adhesion factor estimation and road bank angle estimation is also proposed. Field test results obtained with a prototype vehicle show the effectiveness of the approach for performing in real conditions

A nested PID steering control for lane keeping in vision based autonomous vehicles

In this paper a nested PID steering control for lane keeping in vision based autonomous vehicles is designed to perform path following in the case of roads with an uncertain curvature. The control input is the steering wheel angle: it is designed on the basis of the yaw rate, measured by a gyroscope, and the lateral offset, measured by the vision system as the distance between the road centerline and a virtual point at a fixed distance from the vehicle. No lateral acceleration and no lateral speed measurements are required. A PI active front steering control on the yaw rate tracking error is used to reject constant disturbances and the overall effect of parameter variations while improving vehicle steering dynamics. The yaw rate reference is viewed as the control input in an external control loop: it is designed using a PID control on the lateral offset to reject the disturbances on the curvature which increase linearly with respect to time. The robustness is investigated with respect to speed variations and uncertain vehicle physical parameters: it is shown that the controlled system is asymptotically stable for all perturbations in the range of interest. Several simulations are carried out on a standard big sedan CarSim vehicle model to explore the robustness with respect to unmodelled effects such as combined lateral and longitudinal tire forces, pitch and roll. The simulations show reduced lateral offset and new stable μ-split braking manoeuvres in comparison with the CarSim model predictive steering controller implemented by CarSim.

A new robust control system with optimized use of the lane detection data for vehicle full lateral control under strong curvatures

In this work, we have proposed and implemented in our prototype vehicle a complete lateral control system that works with a good performance even for strong curvatures. The proposed system has important new features with respect to the known systems. While a lookahead distance linear with the speed is often proposed, we have proposed here a quadratic lookahead distance of the speed, that improved considerably the tracking of the lane in curves. A strategy to optimize the use of the information coming from the lane detection module is also proposed. The logics is, if we do not have optimal lane detection exactly in the desired lookahead distance, but in the proximities of this distance we can still control the vehicle with high accuracy. Finally, we have proposed a visual interface that alerts the driver in the case of a degradation in the lane detection. The driver can deactivate the lateral control whenever he wants by simply counteracting the system and an automatic reactivation of the system is provided under strict conditions on the vehicle positioning and on the quality of the lane detection

Dynamic controller for lane keeping and obstacle avoidance assistance system

This paper presents the design and simulation tests of a steering assistance for passenger vehicles based on a dynamic state feedback controller. Its main purpose is to avoid unintended lane departure and collisions. The design of the proposed lane keeping system takes into account the road curvature, considered as an exogenous input, into its internal model. The computation of the control law has been achieved by linking Lyapunov theory of stability to Bilinear Matrix Inequalities which considers bounds in the control input and minimises the reachable set of the vehicle after activation. This control strategy ensures convergence of the lateral offset to zero, even in curvy roads. Simulations show the performance of the controller and an extended application for collision avoidance.

Vehicle Roll and Road Bank Angles Estimation

Driving safety can be enhanced by better understanding of risk situation, which can be achieved by the knowledge of vehicle dynamic states as well as the road geometry. Among the parameters of the road that have an influence on vehicle dynamics, one can find the bank angle, which can not however be measured by low cost onboard sensors. In this paper, a new method of road bank angle and vehicle roll estimation using an unknown input proportional integral (PI) observer is proposed. To reach this goal, first a bicycle vehicle model is chosen. This model is quite simple but well appropriate for the considered application. Thereafter, an Extended Kalman Filter (EKF) is developed in order to estimate the sideslip angle which is also difficult to measure with low cost sensors. This estimate is then used as an input of the PI observer in order to estimate vehicle roll angle and the road bank angle (road attribute). Testing on measurements obtained with a prototype vehicle shows the good behavior the proposed estimation scheme.

Integrated longitudinal and lateral control for vehicle low speed automation

This paper proposes and evaluates an integrated longitudinal and lateral control for vehicles low speed automation. The automation is dedicated to sub-urban congested highways. Using a simplified coupled longitudinal/lateral model, we provide a solution for the vehicle following problem using first and second order sliding mode controls. The performances of the synthesized control laws are highlighted by several simulation tests.

Vehicle yaw rate control based on piecewise affine regions

This paper shows that an active front steering control, that considers the nonlinear behaviour of the tire-road forces, can be designed by parameterizing the vehicle dynamics with respect to the measurable yaw rate and taking into account the steady state behaviour of the vehicle. In order to ensure the tracking of the yaw rate reference signal on the basis of the yaw rate tracking error, despite constant disturbances and parameters uncertainties, the proposed control strategy uses a proportional integral (PI) control, in which the gains depend on the defined parametrized vehicle dynamics. The proposed control system switches depending on the yaw rate as it is a variable measured at low cost. The stability is proved by a piecewise quadratic Lyapunov function using linear matrix inequalities technique. Several simulations, including disturbances rejections and step references, are carried out on a standard nonlinear CarSim D-Class vehicle model to explore the robustness with respect to unmodelled effects such as combined lateral and longitudinal tire forces, pitch, roll and driver dynamics. The simulations confirm that the proposed piecewise linear (PWL) control can greatly improve the vehicle stability and is advantageous in very demanding manoeuvres.