|Ph.D Student||Rubin Daniel Yitzhak|
|Subject||Vehicle Active Suspension Applications for Lateral|
Disturbances Attenuation and Safety
|Department||Department of Autonomous Systems and Robotics||Supervisor||Professor Emeritus Per-Olof Gutman|
Active suspensions allow force generation between the vehicle body and wheels, and were originally aimed at isolating the body heave, pitch and roll responses, from road excitation and vehicle longitudinal and lateral accelerations. However, mechanical and geometrical properties of the suspensions enable warp steering - steering the vehicle by suspension warp.
This research focusses on novel lateral disturbances attenuation and safety applications for active suspensions. Throughout this research, warp steering is used, together with active front steering to compensate for external lateral disturbances such as crosswind and road banking, which act to divert the vehicle from its path. This multiple-input, multiple-output (MIMO) compensation is a novel vehicle disturbance attenuation scheme, and its modeling and formulation are discussed in detail.
First, a Quantitative Feedback Theory (QFT) based solution is proposed. The design is based on an identified frequency response model of the vehicle. The QFT compensator is proven robustly stable and performs according to specifications in the linear region. However, the solution is not guaranteed to work when the system reaches its input and state constraints. Hence, novel interpolation-based control methods are attempted to handle the constraints.
A polyhedral-set based Interpolating Control (IC), extended by the author and colleagues to fit the problem, was used. The vision was to combine IC with our high performance linear QFT based compensator, thus achieving robust stability and performance in the presence of uncertainty, disturbances, and constraints. This attempt was however unsuccessful due to failure in computing the required polyhedral invariant sets. A suboptimal IC scheme, developed by the author, was implemented as an alternative. However, its performance was inadequate, and motivated the development of novel ellipsoidal-set based IC algorithms: A simple algorithm with no stability guarantee, and two line-search based methods which guarantee robust stability, and recursive feasibility.
In addition to the control design effort and theoretical results, the research included implementation of side-slip observers. This includes a two-mode Interacting Multiple Model estimator which uses a limited sensor set, and ultimately a camera-aided multi-rate Kalman filter.
A concluding simulative assessment of our MIMO algorithm shows its superiority to the state-of-the-art method. All controllers and estimators were validated by CarSim simulations. At the end of the day, the QFT based solution outperforms our new IC implementations in simulations, even at saturation. %and the implementation of our new IC methods are inferior. The reason for this is mostly numerical difficulties, which currently prevent the use of high order controllers and design models in IC based implementations.