Vehicle modeling and control for emergency maneuvering of automated highway vehicles

Dirk Edward Smith, Purdue University

Abstract

Intelligent Vehicle Highway Systems (IVHS) will have a significant impact on transportation by reducing highway overcrowding and increasing safety. A system of automatically controlled vehicles would result in closer vehicle spacing, higher speeds, and the elimination of driver error, the cause of most accidents. Recent research in Advanced Vehicle Control Systems (AVCS) has focused on automatic control of acceleration, braking, and steering of highway vehicles up to 0.3 g's lateral acceleration. Since fully automatic vehicles must be able to respond to high-g emergencies, the focus of this work is on the development of vehicle models and controllers for automated emergency maneuvers. Several vehicle and tire models are developed and their responses compared to determine the lowest order model that accurately predicts vehicle behavior during emergency maneuvers. Based on the response comparisons, the eight degree-of-freedom vehicle model using nonlinear tires with side force tire lag (8D-NL-L) is needed for accurate responses during severe steering maneuvers. This model is validated by comparisons with published field tests. Once validated, the 8D-NL-L model was used as the basis for controller studies. First, low order controllers were built using the current state-of-the-art techniques (LQR controllers optimized using linear vehicle models). Then, improved controllers were developed by optimizing the feedback gains using the 8D-NL-L model. The improved control strategy was also applied to a four-wheel-steer vehicle. The results of the study showed that the current control strategies are not adequate for emergency maneuvers. Control gains optimized for the linear model caused the 8D-NL-L model to overshoot a lane change target by nearly 1 meter, and under estimated the distance required for the lane change by 37%. The improved control strategy presented here used the response of the 8D-NL-L model to optimize the feedback control gains, and reduced the overshoot to 7 cm. A continuous gain scheduling technique is presented and tested, which accounts for speed variations and possible braking during emergency maneuvers. Adding four wheel steering to the vehicle caused emergency lane change distances to drop 19.4% at high speeds (30 m/s) while maintaining minimal overshoot (0.07 m).

Degree

Ph.D.

Advisors

Starkey, Purdue University.

Subject Area

Mechanical engineering|Automotive materials

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS