RICERCA ITALIANA     

CONTROL OF ADVANCED SYSTEMS OF TRANSMISSION, SUSPENSION, STEERING AND BRAKING FOR THE MANAGEMENT OF THE VEHICLE DYNAMICS.

Università degli Studi del Sannio di Benevento

Abstract

This project wants to analyse, design, develop and validate experimentally new active control systems for the vehicle dynamics.
An automotive vehicle is a complex system, composed by the interconnection of various mechanical components. Each component is equipped with one or more electro-mechanical control device, which are commanded by a digital control system embedded in one of the electronic control units of the vehicle.

The driveline transfers the motion from the driving shaft to the wheels. Its main components are: the driving shaft, the clutch, the gear, the transmission, the differential unit, the steering and the wheels. The forces transmitted by the driveline to the wheels, the forces of the brakes, the interaction between the tyres and the street and the forces due to the suspensions, all contribute to determine the vehicle dynamical behavior. The control of the vehicle dynamics is achieved by controlling each of the components cited above and taking into account their interconnections.

The control systems projected for a vehicle have to drive its subsystems (or, if we want to consider the three dynamics in which the motion can be divided) such that it works as it is desired. This control problem can be solved in a decoupled way, by considering the dynamics as independent among them, or by considering the existing relation among them. This second way of work is important particularly when some dangerous situations have to be considered, as for example the interactions among a vehicle and vulnerable road users (cyclists and pedestrians), to reduce the possible accidents (for example think to the famous "elk test" solved by an automaker with the ESP system).

The goals of this project are:
• the analysis of vehicle behavior in order to identify the models which can adequately describe the interconnection between the different components. We will focus on models of: the driver, the vehicle lateral dynamics, the semi-active suspensions, the longitudinal dynamics, the transmission system (clutch, actuators and gear), the vulnerable road user;
• the design of the control systems for the lateral, vertical and longitudinal dynamics;
• the testing of the designed control systems by means of real-time "Software in the Loop" and "Hardware in the Loop" simulations.
•the spreading of the competencies among the different work units; the spreading of the new knowledge acquired during the project, both in academia and industry.

The proposed project is also a good chance for the participants to cooperate in a field where each of them has qualified knowledge about the various themes. Further, several experimental facilities are available to all the research groups: a car with an "open" electronic control unit for the longitudinal dynamics control algorithm and three systems for the real time SIL and HIL simulations of all control strategies. These experimental facilities have the same dSPACE software/hardware platform, which will make relatively straightforward the integration of models and algorithms among the units.

The project focuses on very interesting problems both for the automotive industry (whose crisis in Italy can be partially ascribed to insufficient innovation efforts) and for the academia. Technological growth in the automotive sector can be made adequate only by transferring new methodologies from academia to industry as happens in the most technologically advanced countries in the automotive field (USA, Germany and Japan) where the tight interaction between university and automotive industry is one of the key element of their success.
Our project is a step towards this goal and its funding will give academia the possibility of advancing on the topic independently of the short term policies of the industry. <<<

Principal Investigator

Luigi GLIELMO Università degli Studi del SANNIO di BENEVENTO

Research Objectives

Automotive vehicles are complex systems composed by the interconnection of various mechanical components. Each component is equipped with one or more electro-mechanical control devices, which are commanded by a digital control system embedded in one of the electronic control units of the vehicle.
The driveline is one of the most important vehicle subsystems. It transfers the motion from the driving shaft to the wheels. Its main components are: the driving shaft, the clutch, the gear, the transmission, the differential unit, the steering and the wheels. The vehicle behaviour depends from the complex interconnection among the constituting sybsystems. The control of the vehicle dynamic is achieved by controlling each component and taking into account their interconnections, so as to guarantee the driver good performances, the comfort, and a certain level of safety, during the different conditions of motion, such as on highways but also on crowded roads, where a higher probability of collision with vulnerable road users is expected.

The vehicle dynamic can be decomposed into three main components:
• the lateral dynamics, excited in curve, are characterized by the rolling and yaw motion, and influenced by lateral wind and tyres adherence. The analysis of these dynamics is interesting for the safety of the driver and for the passengers comfort. The existing systems for the stability recover, VDC and ESP, are actuated by steering actions or by braking on each tyre. Recent academic research pointed out the better performances obtained by using the combined effect of braking and steering.

• the vertical dynamics are characterized by the heave motion due to the interaction between the four suspensions and the road profile. Suspension control systems automatically control the vehicle vertical dynamic. Suspensions control is a hot-topic both in academia and industry. Automatic suspension systems can both counteract the acceleration, deceleration, and steering forces and at the same time guarantee the passenger comfort by isolating the internal part of the vehicle from the road irregularities. Nowadays, semiactive suspensions are a good compromise in term of price/performance relationship.

• the longitudinal dynamics are characterized by the longitudinal motion of the vehicle and by the pitch. They are controlled by the engine, the brakes, the gear and the clutch.

The goals of this project, and the phases necessary to realize them, are:
• MODELING. Analysis of vehicle behavior in order to identify the models which can adequately describe the interconnection between the different components. The aim is the improving of the detailed dynamical models jet available in the literature and in the background of the single units. The main objective is to design a library of mathematical models which can be easily interconnected.
We will focus on models of:
the vehicle lateral dynamics, the semi-active suspensions, the longitudinal dynamics, the transmission system (clutch, actuators and gear) and the driver. The driver model will be used in order to understand the relationship between driver and passengers comforts and vehicle accelerations. Moreover, we will analyse a model of vulnerable road users, to design a supervisory system to avoid collisions. We will focus both on detailed and simplified dynamical models. The simplified models will capture the main behavior of the single components and will be used for control design purposes. It must be pointed out that automotive libraries are available on a commercial basis, but they have a different flavour because of cost, computational load and software platform. Here instead we aim at developing tools open to the participating units and, partially, the the whole scientific community and available on the project web site.
The modeling part and the development of the corresponding software library will be coordinated by the group of Pavia.

• CONTROL. The design of the control systems for the previous modelled subsystems, as described in the Models B of the units. Detailed models will be used for a first validation, and then simplified models will be used for the control design.

We will focus on semi-active suspension for the vertical dynamics; braking and steering control for the lateral dynamics (Turin); clutch engagement (Sannio) and torque profile control for the longitudinal dynamics (Modena); supervisory and automated control system for braking in presence of obstacles for crash avoidance (Pavia). The group of Turin will coordinate the control design for the lateral and vertical dynamics. The group of Modena will coordinate the control design for the longitudinal dynamics.

• TESTING. We will perform "Software in the Loop" and "Hardware in the Loop" (on a dSpace platform) simulations of the vehicles dynamics equipped with the controllers previously developed. Afterwards, the longitudinal control algorithms will be tested on a Maserati vehicle, available at the Bologna University. The group of Turin has access to an instrumented vehicle to test the vertical and lateral controllers. The group of Bologna will coordinate this part of the project.

•TRANSFER OF KNOWLEDGE, DISSEMINATION AND EXPLOITATION OF THE RESULTS.
Transfer of the competencies among the different work units will be one of the main objective of this proposal and will be facilitated through researchers short-term exchanges. Workshops and invited session at international conferences will be organized in order to spread the results obtained during the project. A web site of the project will be published including links, reports, software tools library. The group of Sannio will coordinate this part of the project.

This project evolves from previous projects submitted in the last two years. The version submitted in 2003 regarded only the engagement control, and the interested groups were Sannio, Bologna and Modena. Last year it was extended to the vertical and lateral control, and to the use of robust identification and control, with the inclusion of the group of Turin and of a unit of mechanical engineering of the University of Napoli Federico II. These two projects were positively judged for funding but eventually not funded (due to insufficient financial resources at MIUR).
All the research groups involved in this project have been active for many years in the automotive field, with national automakers) and are well known both nationally and internationally. Indeed, the goal of this project is to research on new methods and methodologies applied to the automotive problems which go beyond the short-term need of the automotive industry, because it could be give to the groups a scientific *national* acknowledgment on the quality of the research in this crucial sector.

The following graph illustrates the collaborations, about the specific application themes, among the units of the project.
<<<

Timescale

24 months

National and international background

In the last decades, modelling, identification and control of automotive systems has become an interesting research area. A great number of difficult problems appear from this multidisciplinary area [1].
One of these problems, very complex and significant, is the management of the vehicle dynamics on the road, and it can be studied by using a structured approach: the whole system is considered as the connection of different subsystems; first each subsystem is studied, and then all the subsystems are connected. Hence, the control of the whole system passes through the control of the single subsystems. Roughly speaking, the vehicle dynamics can be decomposed into the three directions of motion: lateral, vertical and longitudinal dynamics.
The suspensions are used for the control of the vertical dynamics; the suspensions, the brakes and steer are used for the control of the lateral dynamics, by using of ESP or automatic tracking systems (in curve); the brakes, the ABS system, the traction control systems, the clutch engagement and the gear box are used for the control of the longitudinal dynamics; all these control systems, among the other goals, have to guarantee comfort and drive safety.
From the methodological point of view, it is necessary to highlight that most of the problems above are characterized by nonlinearities, time varying parameters, different time-scale dynamics, and the intrinsically hybrid structure of various subsystems, e.g. the transmission.
Thus, it is clear that these types of complex problems can greatly benefit from the academic research which can provide innovative methodologies, often not yet utilized within the industry, which prefers to use more traditional methods, even if they are not optimal, because of pressing manufacturing constraints.
In the following the state of the art about the three vehicle dynamics will be described.

Vertical dynamics.
The design of controlled suspension systems for road vehicles aims at enhancing the vehicle performances for comfort and road handling. Such performance requirements have received, in the last two decades, a growing interest witnessed by an intense research activity developed from both industrial and academic groups [2]. Three main categories of suspensions can be considered: passive, active and semi-active. Comprehensive studies have been published that evaluate and compare the performance of active and semi-active suspensions [2]-[3], and it seems clear that semi-active suspensions provide a good compromise between costs (energy-consumption and actuators hardware) and performances.
These suspensions, realized by dampers with controllable damping coefficient, are relatively inexpensive and require negligible additional power; they are mounted on many newly-designed top-cars, and are the subject of intense academic and industrial research [2]-[4]. At present, the most used control techniques are: On-Off Skyhook and ''clipped''.
In On-Off skyhook control the damper can be adjusted at two possible states, maximum (high-state) or minimum (low-state). The determination of the damper state depends on the product of the relative velocity of the suspension damper and the absolute vertical velocity of the car. If the product is positive or zero, the damper is adjusted at its high state; otherwise, the damper is set to the low state. A "clipped"controller is generally designed in two steps: the first step consists in designing an active control law assuming an ideal active suspension is present; in the second step the computed control is clipped if the "passivity constraint" of the damper is violated.
An effective semi-active suspension control strategy requires to balance a set of comfort and handling specifications that can be formulated through a suitable performance index, subject to the above cited passivity constraint on the damper. Model Predictive Control techniques appear to be the more appropriate to handle the control design accounting for the constraints and the dynamic evolution of involved variables. In this context, researchers of the unit of Turin showed [5], experimentally too [4], that the use of predictive techniques allows to reduce significantly the accelerations peaks on the car body mass with respect to the described Skyhook and clipped techniques, leading to a higher comfort level.
As regards the vertical dynamics control, the project aims at obtaining innovative results with respect to the state of the art, also through integration of the methodological competencies on the predictive control of the units of Turin and Sannio.

Lateral dynamics.
The aim of a lateral dynamics control system is mainly to enhancing driving safety and reduce driver's workload. Automated Highway Systems (AHS), extensively studied since the sixties, are receiving renewed attention due to fast developments in hardware/software technology. Since mid-eighties a larger effort is being conducted mainly in the California PATH program. The paper [8] is an extensive report of results obtained about both lateral and longitudinal control. Most of the contributors rely on buried magnet or electrified wires placed along the path for the detection of the vehicle lateral position (look-down scheme) , or on vision sensors placed on board of the vehicle (look-ahead scheme).
In [9]-[11] the results obtained by applying some vision-based control strategies are shown. The methodological approach used for the lateral dynamics control is the robust control theory [12]-[13]. The group of Turin has been particularly interested to the lateral dynamics control on highway paths: on a instrumented car equipped with a vision system, the steer was used as an actuator in a look-ahead scheme [6]-[7].

Vehicle dynamics control (VDC) is an active safety system introduced by Bosch in 1995 for the control of the yaw motion of the vehicle in emergency situations [14]. In the original control strategy, the steering angle, the accelerator pedal position and the brake pressure are measured in order to determine the vehicle motion desired by the driver, while the actual yaw motion of the car is computed from the measurements of the yaw rate and the lateral acceleration. The aim of the control system is to minimize the difference between the actual and the desired vehicle behaviour through a corrective yaw moment generated by means of both engine torque regulation and wheel brake pressures control. Since 1995 a great deal of research has been made in this field [15]-[16], leading to a number of different control strategies exploiting either the steering command or the differential braking system as well as combinations of those two devices.
The positive effects on the lateral dynamics due to the torque distribution on the differential are also well known [38]. The distribution torque strategies on the differential gear is a new theme for the vehicles. The simplest solutions used are based on tables describing feed forward action, or are an imitation of the behaviour of the self-lock mechanical differential. These two solutions are useful for their simplicity and reliability, but they are not optimized in view of the dynamical behaviour of the vehicle. An optimal torque distribution that should stabilyze the vehicle in curve will be studied during this project. This distribution will be actuated by working on the differential, by using robust control techniques, as [40]-[41], or physical based techniques [39].
Innovative results on this topic are expected from the activities of the project that will involve the lateral dynamics competencies of Turin and Modena units.

Longitudinal dynamics
A vehicle advances thanks to the movement transmitted from the crankshaft to the wheels by the driveline system, which is composed by the crankshaft, clutch, gears, differential and wheels [1]. A smooth motion, comfortable for both passengers and driver, is related to the control of the driveline, obtained controlling its subsystems. The design of control algorithms requires a dynamic model for the whole transmission system, relatively simple but accurate enough to describe all its characteristic dynamic phenomena. The related literature proposes different models for the transmission, each of them emphasizing only some specific aspects, e.g. engine torque control in [17], an automatic transmission use in [20] and the clutch engaged transmission behaviour in [18]. Due to the transmission systems complexity, many authors introduced a dynamic model for each operative condition and interfaced the resulting sub-models using the definition of hybrid systems (e.g., [19]). Once the vehicle transmission dynamic model is known, a management strategy during the gearshift must be chosen in order to guarantee, besides the operation correct execution, low stresses on the mechanical components and a good level of driveability [21]. Often this amounts to a unique dynamic requirement: to keep the derivative of the vehicle acceleration (jerk) as low as possible [17], with respect to the gearshift timing . Recently, Modena unit has begun to refine that analysis introducing the driver too into the system: in [22], [23] a passenger head and neck model enables evaluation of the human reaction to the gearshifts torque transients.

Another important element of the transmission is the clutch-gears system. In the United States such system is almost always replaced by the automatic gear, which enjoys less fortune in Europe and Latin countries. A valid alternative, under the marketing viewpoint, is the semi-automatic gears (or automated manual transmission), composed by servo-commanded clutch and gears. Different problems related to these gears have been considered in the
literature: for example, the gear shift [24], the actuators [25]-[26], the driveline vibrations and the so called "judder" [27]-[28], the wear [29], the clutch engagement control strategies [30]-[35]. Particularly, the dry clutch engagement control is crucial in order to reduce the wear losses and keep driveline performances acceptable. The engagement has to be controlled in order to meet different, and sometimes opposing, needs: low friction losses, minimum engagement time, driveability, driveline oscillations reduction. These objectives must be reached by applying a suitable force normally to the disks, and regulating the torque produced by the engine during the engagement phase, so that different constraints are simultaneously satisfied (no engine killing or wheel slipping, reduction of transmission oscillations). An important sub-system is the engagement/disengagement clutch actuator [25]-[26]. The Bologna unit has worked on the hydraulic actuator [26], but its competences in electric actuators control ([36]-[37]) will allow to consider electric actuators too, whose use however requires an integrated design of electric motor, kinematic chain and controller. As regards the clutch system, the most interesting operating conditions are related to the start-up phase and low gear-ratios gearshifts. Indeed at lower gear ratio (low vehicle speeds), the transmitted torque will be larger, and hence larger will be the torsional effects on the crankshaft and the driveline [27]-[28]. Various authors studied the driveline control problem and in particular, the clutch engagement manoeuvre by considering model-based control strategies as [29]-[35]. In particular the Sannio unit used the finite horizon linear quadratic approach [32], the Model Predictive Control (MPC) [33] and the Multivariable Control [34], explicitly including in the design the physical constraints and the minimization of friction losses.
In this project the analysis of driveline components as hybrid systems
(Sannio) and the use of tracking control techniques (Modena) and constrained control techniques (Bologna and Sannio) will, on one side, contribute to the improvement of the performances of automated manual transmission systems (also through experimental validation) and, on the other side, inspire methodological advancements taking into account real-life constraints.
A problem that crosses the lateral and longitudinal dynamics regards the interaction among the vehicle and the vulnerable road users (VRU). This matter has interested in the last years the European Commission, which financed a lot of projects, as the IST PROTECTOR [42] and the PReVENT [43], that have to develop advanced driver assistance systems (ADAS) for increasing VRUs safety. An ADAS will have to produce on-board warnings that advise the driver as soon as the presence of a VRU is detected, and when the risk of collision is greater than a prefixed value [42].
Some sensors on-board detect the presence of a VRU, by measuring the relative position and speed among the VRU and the vehicle, or by using front and lateral sensors (radar, laser or stereo vision systems). Some tests have shown that it is possible the reduction of the number of collisions if the driver is alerted with an advance sufficient to react in a right way. While in the long range case a warning system is sufficient to reduce the number of accidents, an active intervention system is mandatory whenever the difference between the VRU detection time instant and the predicted collision time instant is close to the average driver reaction time. There are two types of automatic actions that a driver
assistance system can accomplish so as to attain collision avoidance or injury severity mitigation: an emergency braking or a collision avoidance manoeuvre. In this project a supervisory ADAS system will be realized by using innovative control techniques (Pavia). <<<