Florent Coffin

Lateral guidance of an autonomous vehicle by a fuzzy logic controller

This study is the fruit of cooperative research between Renault and the Fuzzy Logic Group at the University of Technology of Compiegne. Emphasizing the drivers comfort, our main objective was to design and tune an active driver assistance enabling and ensuring automatic guidance, while at any time allowing the driver to take over control of his vehicle. A fuzzy logic controller (FLC) has been developed and implemented on a Renault Safrane, ensuring transversal control of the car. An on-board CCD camera, connected to an image processing unit, delivers the specific data required for the guidance controller. The FLC processes these data and generates a set point of steering wheel torque which is delivered to a classic PI regulator controlling an electric motor implanted in the steering column of the prototype. Simulation results show the convergence, over a large range of longitudinal speeds, between the trajectory output of a vehicle model-controlled by the FLC-and the reference trajectory to be followed: the controller is also robust to any perturbation such as gusts of wind or violent swerves of the wheel. Currently, experimental tests on road yield satisfactory results: the FLC generates a nonoscillating smooth and comfortable trajectory in hands-off driving situations, while it arouses a real feeling of trust and helpfulness when the driver keeps his hands on the steering wheel. An ergonomics study carried out concurrently will enable us to refine our copiloting system.

BUS STOP DESIGN AND AUTOMATED GUIDANCE FOR LOW-FLOOR BUSES: EVALUATION OF PROTOTYPES WITH INVESTIGATION OF HUMAN FACTORS

Low-floor vehicle design improves the accessibility of urban buses to people with reduced mobility. In France, on the basis of the experience gained from low-floor tramways, a system approach soon appeared necessary and the low-floor bus within the urban environment was considered. Described are the results of the research and development project initiated by the Grenoble network to reach efficient guidance solutions that would ensure minimal gaps at bus stops. Operating conditions of accessibility equipment on the vehicle (kneeling, access ramp) as well as driver capability to dock at the bus stop have been investigated. From the results, an improved bus stop was designed and two prototype systems—GIBUS visual aid and VISÉE guidance system—were conceived and tested. Evaluation of these during operation on a bus route in the city of Grenoble concentrated on assessing system performance and human factors so that recommendations could be drawn before implementation.

FUZZY LANE-TRACK CONTROL FOR THE AUTOMATIC GUIDANCE OF AN AUTOMOTIVE VEHICLE

Abstract A Fuzzy Lane-track Controller called RPV is presented in this paper. Putting the emphasis as much on safety as on driving comfort, the aim of this study was the design of a driver assistance that would ensure a guidance function along a “virtual rail”, while any time allowing the driver to take over the controls. A fuzzy controller was designed and embedded on a Renault Safrane to perform the lateral control of the vehicle. An on-board CCD camera and image processing software and a gyroscope deliver the necessary information to the controller (lateral displacement, lateral speed, yaw rate). From these data, the fuzzy controller outputs a value of torque variation, which is applied to the steering column by means of a coaxial electric motor. Full-size field experiment yield encouraging results: on one hand, the lane-track controller induces a nonoscillating, smooth and comfortable trajectory in situations where the driver does not steer the test vehicle (hands free); on the other hand, RPV generates a noticeable feeling of confidence and assistance when the driver keeps his hands on the steering wheel.

A Vehicle Path Planning Algorithm Using Polar Polynomials Optimized Through Fuzzy Logic

Abstract The Navigation Aided Intelligent Cruise Control (N.A.I.C.C.) system acts in the field of the lateral and longitudinal control of a car. This paper only describes the lateral part of this new knowledge-based predictive driver-aid system. Once the vehicle is located on the route, this copilot detennines the distance to the next bend and predicts the optimal path to negotiate this curve, considering the road profile, the vehicle characteristics and the constraints given by the driver. The use of polar polynomial curves for the description of a turn eliminates steering-function discontinuities by providing continuous curvature along the path. A fuzzy-logic method is used for the determination of the characteristics of the end-points. In order to compute paths matching with a drivers style (novice, experienced driver... ), real experiments were carried out with the instrumented laboratory test car and drivers of different types.