Vehicle collision avoidance strategy based on real-time estimation of tire-road friction coefficient

Abstract

The number of vehicles around the world continues to gradually increase, but on the contrary, the fatality rate has been decreasing. This is because various technologies, such as in-vehicle sensors and improved computing performance, have been developing. With the help of this kind of technology, while moving away from passive safety systems such as airbags and seat belts in the past, more active safety systems such as automatic emergency braking and lane departure warning are being newly developed and applied. In particular, a collision avoidance system that reduces an accident rate by controlling both braking and steering together has recently emerged as a major field of active safety systems. Unfortunately, however, many related studies have limitations that it is difficult to properly cope with changes in road surface conditions that can frequently occur in real world. It is because those studies assume the tire-road friction coefficient, which plays a decisive role in collision avoidance, as a known constant value. Therefore, to overcome these limitations, a comprehensive collision avoidance system that estimates and applies tire-road friction coefficient in real-time is proposed in this dissertation. First of all, a path planner and tracking controller capable of applying the friction coefficient change within the algorithm and providing real-time feedback of the result (Updated state variables of a vehicle) are proposed. In order to overcome the disadvantage that the spline based path planning method is light in calculation but difficult to reflect tire-road friction information, a path that induces the maximum physically possible lateral acceleration on the corresponding road surface is created by applying the curvature optimization algorithm proposed in this study. As for the tracking controller, the use of longitudinal tire force is optimized based on the currently generated lateral tire force, away from the traditional method of optimizing the use of tire force by taking into account the tire model and future steering references. This makes it easier to maximize the use of tire force and quickly feedback updated state variables to the friction coefficient estimator. As the proposed path planner and tracking controller are designed to maximize the use of the tire-road friction force, estimating the tire- road friction coefficient in real-time plays an important role in ensuring the performance of the overall collision avoidance system. Therefore, a tire-road friction coefficient estimator capable of estimating the tire-road friction with high accuracy even during collision avoidance is proposed. In general, in addition to the fact that accurate estimation of the friction coefficient must be made in a short time, there is a limitation that sufficient excitation required for estimation is not continuously performed during collision avoidance. However, this problem is solved by combining the advantages of tire-road friction coefficient estimators using linear and nonlinear tire models. Finally, the collision avoidance performance of the overall system combining the proposed collision avoidance system and the tire-road friction coefficient estimator is demonstrated. Looking at related studies on collision avoidance system and tire-road friction coefficient estimation, a detailed analysis of the interaction between the two has not been included, even though the two systems are closely related. However, in this study, the virtuous cycle of interaction between the collision avoidance system and the tire-road friction coefficient estimator is demonstrated by applying the estimated friction coefficient in real-time to the collision avoidance system and using the response of the system to estimate the friction coefficient again.