Present position: Contract Professor
|Thesis title:||Active braking control systems design for road vehicles|
|Advisor:||Sergio M. Savaresi|
|Research area:||Automotive Control|
Introduction and description of objectives
The control of longitudinal dynamics - constituted by braking and traction control is one of the main areas of vehicle dynamics control. As far as braking control is concerned, electronic Anti-lock Braking Systems (ABS) have recently become a standard for all modern cars. In fact, ABS can greatly improve the safety of a vehicle in extreme circumstances which cannot be handled by human drivers. From the technological point of view, the design of automatic braking control systems is clearly highly dependent on the braking system characteristics and actuator performance. As a matter of fact, ABS systems for wheeled vehicles equipped with traditional hydraulic actuated brakes - that is those commonly available on all passengers cars - pose specific design constraints as they have to cope with an on/off modulation of the brake pressure. On the other hand, the recent technological advances in actuators which have led to both Electro-Hydraulic and Electro-Mechanical braking systems have radically changed the starting point of braking control systems design. In fact, these brake systems enable a continuous modulation of the braking torque, thereby allowing to deal with the design problem employing classical Control Theory tools. It should be also mentioned that a great boost to the research in this field comes directly from the industrial world, which poses challenging problems asking for reliable control systems with the simplest possible architecture, reduced sensors layout and the capability of coping with transmission delays and significant measurements errors and parametric uncertainties. The PhD Thesis has been developed within this interesting and evolving context, with the aim of providing a thorough analysis of active braking control systems and proposing innovative solutions, which are both effective from the application viewpoint and theoretically sound from a methodological perspective. The research work has been organized into two different parts, the first devoted to the design, analysis and test of different active braking control systems and the second to estimation and identification problems which arise in the braking control context.
Methodology and Main Contributions
Braking Control Algorithms
An original contribution to the analysis of standard ABS systems with hydraulic actuated brakes is provided. We propose a hybrid ABS controller, which is proved to give rise to an asymptotically stable limit cycle on the wheel slip. Based on this control system, we first give necessary conditions for the limit cycle existence, prove the limit cycle existence and its asymptotic stability properties and provide a structural stability analysis with respect to different road conditions and to the actuator characteristics. To the best of our knowledge, ABS systems based on hydraulic brakes with on-off dynamics - even though they are a standard on all cars - have never been fully analyzed, nor formal proofs of the stability and robustness properties of the control algorithm have been fully established in the literature so far.
A new control approach named Mixed Slip Deceleration (MSD) control is proposed: the basic idea is that the regulated variable is a convex combination of wheel deceleration and longitudinal slip. This strategy turns out to be very powerful and flexible as it inherits all the attractive features of slip-control, while providing a much lower sensitivity to slip-measurement noise. An activation and deactivation control logic is also devised, which constitutes a very important step toward implementation on a real car. Further, some preliminary experimental results obtained on a prototype by-Wire vehicle assess the industrial suitability of the proposed control strategy.
A detailed analysis of the implications of employing a quarter-car model for the design of braking control systems and its comparison with an half-car model is performed, which leads to highlight the dynamic coupling between front and rear axle and to devise a new control algorithm which offers advantages in terms of coupling effects minimization. Moreover, this control approach allows to remove the force sensors from the rear electro-mechanical brakes, hence having also a relevant technological impact in terms of architectural simplification and cost reduction.
Three different nonlinear control approaches, each of which is a novel approach with specific peculiarities have been studied:
– a sliding mode approach applied within the MSD framework, which guarantees robustness with respect to uncertainties in the actuator dynamics and provides powerful theoretical tools to analyze the MSD control logic and to highlight its advantages with respect to pure slip and deceleration control;
– a robust output feedback control, which allows to overcome the ABS unreliability at low speed. Moreover, the closed-loop system properties are such that the proposed control algorithm allows
to detect if the closed-loop system is operating in the unstable region of the friction curve, thereby allowing to adapt the set-point and greatly enhance both performance and safety;
– a non-local extremum seeking controller, which guarantees global closed-loop stability.
Finally, we carried out an analysis of the braking control design for two-wheeled vehicle tailored to devise an ABS logic which can handle panic brakes on curves, which is not yet available on commercial motorbikes.
Identification and Estimation problems
A new algorithm is proposed for longitudinal vehicle speed estimation which employs a low-cost sensors configuration and whose performance has been successfully assessed on experimental data. Secondly, a new estimation strategy is presented, which enables the on-line detection of the peak–value of the tyre-road friction curve which may be employed as a supervisory control to enhance safety properties and performance of ABS Systems. The algorithm has been tested on an instrumented car. Further, a novel approach to the on-line direct estimation of tyre-road contact forces based on in-tyre sensors has been studied; specifically, the idea is to use a wheel encoder and an accelerometer mounted directly in the tyre. The key innovative concept is to use the phase shift between the wheel encoder and the pulse-like signals provided by the accelerometer as the main regressor for force estimation. The feasibility of the proposed approach has been confirmed by experiments on a test vehicle. Finally, an original on-line roll angle estimation algorithm for two-wheeled vehicles is described, which is the enabling step to move toward active control systems which can handle panic brakes on curves. The estimation method is based on a low-cost sensor configuration, which may be suitable for industrial purposes. The validity of the proposed approach is first assessed in simulation and then also on an instrumented test motorbike.