162_A Review of Yaw Rate and Sideslip Controllers for Passenger Vehicles

http://tim.sagepub.com/Measurement and Control

Transactions of the Institute of

http://tim.sagepub.com/content/29/2/117The online version of this article can be found at:

DOI: 10.1177/0142331207072989 2007 29: 117Transactions of the Institute of Measurement and Control

W.J. Manning and D.A. CrollaA review of yaw rate and sideslip controllers for passenger vehicles

Published by:

http://www.sagepublications.com

On behalf of:

The Institute of Measurement and Control

can be found at:Transactions of the Institute of Measurement and ControlAdditional services and information for

http://tim.sagepub.com/cgi/alertsEmail Alerts:

http://tim.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://tim.sagepub.com/content/29/2/117.refs.htmlCitations:

What is This?

- Jun 25, 2007Version of Record >>

at UNIV OF CALIFORNIA SANTA CRUZ on October 11, 2014tim.sagepub.comDownloaded from at UNIV OF CALIFORNIA SANTA CRUZ on October 11, 2014tim.sagepub.comDownloaded from

A review of yaw rate and sideslipcontrollers for passenger vehiclesW.J. Manning1 and D.A. Crolla2

1Faculty of Science and Engineering, Department of Engineering and Technology,John Dalton Building, Manchester Metropolitan University, Manchester M1 5GD, UK2University of Sunderland, School of Computing and Technology, Sunderland SR6ODD, UK

The development of chassis control schemes has been a major area of study for automotivecontrol engineers over the past 30 years. The volume of published literature is large, exceeding1000 papers. Of this literature, there are 250 examining yaw and sideslip control. Here is acomprehensive review of this field of study to identify the current state of the art and researchin yaw rate and sideslip control. The survey shows that there is still a significant research effortneeded to address the subjective performance of handling systems, and more research isneeded to develop schemes that integrate systems to achieve high-level performance objectives.

Key words: braking; driveline; integration; sideslip; steering; suspension; yaw rate.

Nomenclature

0s Total steer angles Driver input steer anglem Additional active steer angleKc Proportional gainrd Desired yaw rater Actual yaw rateNm Newton-metreg Gravitational constant 9.81m/s2

V Vehicle forward velocityf Front steer angler Rear steer angle

Address for correspondence: W.J. Manning, Faculty of Science and Engineering, Department ofEngineering and Technology, John Dalton Building, Manchester Metropolitan University,Manchester M1 5GD, UK. E-mail: [email protected] 2 appears in colour online: http://tim.sagepub.com

Transactions of the Institute of Measurement and Control 29, 2 (2007) pp. 117135

2007 The Institute of Measurement and Control 10.1177/0142331207072989

at UNIV OF CALIFORNIA SANTA CRUZ on October 11, 2014tim.sagepub.comDownloaded from

K1,K2,K3 Feedforward and feedback gainsFn Normal force at the tyrem Tyre-road coefficient of frictionFB Longitudinal tyre forceFn Lateral tyre force

1. Introduction

This study reviews a selection of 68 papers from around 250 in the field of lateralhandling control. The 68 papers have been selected because they have validated theirapproach, offered opportunity for further work and, in some cases, gave a practicaledge in a largely theoretical field.

Taxonomy of the field of study is challenging. Mapping the active vehiclesystems to the objective control criteria is challenging, but more difficult is mappingthese control objectives to the more subjective criteria. Figure 1 shows the flow fromhigh-level criteria to the systems for controlling vehicle motion. While criteria suchas stable and controllable can be expressed in very objective terms, whether its yawrate or sideslip, other criteria such as predictable and affective are much moresubjective.

The lack of published work that maps the high-level criteria to vehicle motionobjectives simplifies taxonomy to either that of systems or control objectives.By focusing on the control objectives, this review maintains a top-down analysis ofthe control approaches. The review focuses on work done with three control objectives:

yaw rate control sideslip control combined yaw and sideslip control.

Predictable

Conventional

Consistent

Controllable

Stable

Affective

Yaw rate

Sideslip

Forward speed

Roll and pitch

Ride

Steering

Braking

Driveline

Suspension

Ride

SystemsCriteria

High levelcriteria

Vehiclemotionobjectives

Figure 1 Criteria and systems for vehicle handling

118 Yaw rate and sideslip controllers


There are other control lateral objectives but these three are chosen as they exemplifyvehicle lateral handling. The yaw rate control objective is primarily concerned withimproving steering feel. Most studies employ a yaw rate tracking approach, where thetarget yaw rate is usually generated from a first-order lag on the steering input.Sideslip control relates more to the vehicle stability and is important near the limit ofvehicle handling. The combined approach is usually employed with two systems, eg,braking and steering, and should offer the benefits of improved handling feel as wellas increased stability near the limit.

2. Yaw rate control

Yaw rate following studies are dominated by the application of active steeringsystems, with some work on active driveline, steering and suspension. Overall, thereare two quite separate regions in which active systems may offer benefits:

(1) limiting conditions where vehicle stability may be lost because the tyres are close totheir limit of adhesion;

(2) normal driving implying modest lateral accelerations and good surfaces.

2.1 Limit handling studies

The retention of stability under severe manoeuvres has been studied extensively.Kramer and Hackl (1996) from Bosch gives a good overview of yaw control by ActiveFront Steering (AFS). The underlying principle of this study is that the resultant steerangle may be expressed as

0s s m 1where 0s is the total steer angle, s is the driver input steer angle and m is theadditional active steer angle.

In practice, the additional steer angle m is actuated through a motor that drives thering gear of a planetary gear connected in line with the steering column (Figure 2). Thisis the basis of recent commercial systems fromBosch andContinental. Feedforward andfeedback control approaches are suggested for yaw control under steeringmanoeuvres,and an additional control approach is presented for lateral stability under braking onsplit-m surfaces. The feedforward control scheme (Figure 3) has the effect of transientlyincreasing the resultant steer angle, 0s, and tends to reduce the time lag between steerangle input and vehicle reaction, eg, yaw rate or lateral acceleration. The yaw ratefollowing scheme (Figure 4) uses a simple two-degree-of-freedom (2dof) model togenerate an ideal yaw rate, which is then fed through a first-order lag to generatethe demand yaw rate. The controller is then a simple proportional gain, ie,

m Kc rd r 2

Manning and Crolla 119


Under a severe lane change, both systems are an improvement on the uncontrolled

vehicle, but the feedback control scheme requires less driver control effort andmaintains good adequate sideslip behaviour on icy surfaces. Yaw stability compensa-

tion (Figure 5) involves intervention under limiting conditions and thus it can be

viewed as additional to controllers (i) and (ii). The idea is that the control system

PlanetaryGearbox

Motor

Figure 2 Active front steering

dtd TL

Vehicleds

dm

d s Yaw ratelateralacceleration

+ +

Figure 3 Feedforward yaw control by AFS

Referencemodel

Vehicleds

dm

d syaw rate, w

Controller

wddemandyaw rate

+

+

+

Figure 4 AFS yaw rate feedback control



automatically generates an additional steer angle, m, to assist the braking-basedstability control system on the vehicle. This can, in principle, be advantageous insplit-m braking conditions, where the conventional stability control system has aninevitable compromise between maximizing retardation and maintaining yaw control.While these experimental tests show the controller has significant benefits, it is not

clear how the hardware compares with that used in the current Bosch systems.In particular, the power of the steering actuator is key to performance and has clearimplications on fuel consumption.

Ackermann et al. (Ackermann, 1992, 1997; Ackermann and Bunte, 1997; Ackermannand Sienel, 1993; Ackermann et al., 1996) carried out separate studies on both ActiveFront (AFS) and Active Rear Steering (ARS) to decouple the vehicle yaw motion from

the lateral acceleration. The aim of this decoupling law is to allow the driverto complete path-following tasks while the control will reject disturbances due tocrosswinds or split-m road surfaces. The more recent work on AFS has been evaluatedthrough both simulation and road tests on a BMW 735i. The controller concept issimilar to Figure 4 with an additional positive feedback element from the yaw rate tothe steer angle. It is this feedback element that removes the yaw dynamics from the

drivers control and gives direct lateral acceleration response to steer angle inputs.While this system is successful at dealing with unexpected yaw disturbances such as

crosswinds and split-m surfaces, its performance in severe lateral handlingmanoeuvres is questionable. In these studies, Ackermann (1997) highlights thesteering actuator bandwidth and saturation as the key limiting factor in achieving the

desired yaw dynamics. In contrast to Kramer and Hackl (1996), this work used ahydraulic cylinder that actuated the steering rack. As little detail is given on eitheractuator, it is difficult to infer which will give the best overall performance in yaw rate

response for given constraints on weight and power demand.Segawa et al. (2000) extended same concepts as in Figures 3 and 4 to include

feedforward and feedback control of yaw rate (Figure 6). The approaches were

experimentally tested on real vehicles under severe manoeuvres, including high-speed slalom tests and lane changes on packed snow, and split-m braking.

Braking-based stabilitycontrol programme

e.g. DYC, ESP

Vehicleds

dm

d s+

+

Controller

Figure 5 AFS addition of yaw stability compensation term



The feedforward part is a Variable Gain Steering (VGS) ratio, scheduled with speed.Two strategies are proposed for feedback control: the conventional approach in

Figure 4 and a D* strategy where the lateral acceleration and yaw-speed component of

lateral acceleration are feedback under separate gains. Experimental results show that

the yaw rate feedback attenuates the yaw response best; however, the D* feedback

does reduce the lag in the system and does improve the path following on

low-m surfaces.The feedforwardfeedback tracking control in Figure 6 is more common in other

yaw rate stability studies that use braking, driveline and suspension systems.

Matsumoto et al. (1992) describes Nissans early work on Brake Force Distribution(BFD) control and validates their approach with some experimental results. Therehave been various other papers that corroborate this work. There is significant detail

on how the systems are implemented on the BMW direct stability control systems

(Donges, 1996; Leffler, 1996; Leffler et al., 1998a,b; Straub, 1996). While there are clearadvantages of using braking for yaw control, it should be made clear that longitudinal

performance is compromised due to the braking action. If the aim of the controller is

to expand the envelope in which the vehicle handles with good feel and stability, then

braking systems will cause problems. If safety and stability are the only objectivesthen braking systems are best; however, they are usually used in this capacity to

bound sideslip and not for following yaw rate.Other papers have used the feedforwardfeedback control principle but with

different actuators. Naito et al. (1992) present a front/rear four-wheel drive (4WD)torque split mechanism for improving cornering performance. This Electronic Torque

Split (E-TS) system developed by Nissan features in the companys Skyline vehicle.

When a vehicle accelerates, load is transferred from the front wheel to the rear wheels.

This decrease in load and increase in tractive force reduces cornering forces, hence

promoting understeer. By increasing torque at the rear wheels (promoting oversteer),

Referencemodel

Feedforward

controller

Feedbackcontroller Vehicle

ds

uff

r ufbrdes

Figure 6 Feedforwardfeedback control of yaw rate



a balance can be reached by which the vehicle cornering performance is improved.Experimental results show that the system offers improvements over other 4WDsystems and is equally effective on low friction surfaces. When combined withthe system above, it is noted that cornering and tractive force is improved on bothlow- and high-m surfaces during subtle and rapid acceleration/deceleration. A similarsystem is presented by Matsuno et al. (2000). The aim is to improve handling on dryroads and stability on slippery roads. The Variable Torque Distribution (VTD) systemvaries the torque transfer between front and rear axles depending on a number ofconditions, such as the estimated road surface coefficient of friction. Results showgood yaw rate control, but are rather limited. The control methods of the VTD are notpresented in this paper, although a method of m estimation is presented that useslateral acceleration of the vehicle (the most common method being through thelongitudinal acceleration of the vehicle). Results are accurate despite the need for steerinputs to be present before estimation occurs.

A paper presented by Motoyama et al. (1993) shows the effect of left/right torquesplit on both axles for a 4WD vehicle. Yaw moment control is implemented throughidentical left/right torque distribution in both axles. Multi-plate clutches are installedat all wheels to control the torque transfer. Results show improved yaw ratetracking behaviour over conventional 4WD systems. The paper also compares therelative effectiveness of both left/right and front/rear torque split, showing thatleft/right shows the greater potential for improving vehicle turning characteristics.With both simulation and actual test results showing good yaw rate tracking andincreased turning limit, conclusions as to the effectiveness of such a system arejustified.

Smakman (2000) investigates inverse model control of wheel load response for yawrate tracking. The systemworkswell until actuator limits are imposed and it is clear thatwhile wheel load control can offer some benefits, it does not offer the power of brake-based systems. Key to choosing the correct actuator is identifying the available yawmoment by that actuator. Smakman (2000), Hac and Bodie (2002) andHe (2005) have allproduced results fromARS, AFS, BFD, Roll Moment Distribution (RMD) andMagneto-Rheological (MR) suspensions, and judged themwith amixture of criteria that includedmaximum yaw moment available and invasiveness to driving performance. These areall excellent papers for analysing actuators based on operating regime.

2.2 Linear handling studies

Work on VGS has been proposed as the gain of the yaw rate response variessignificantly with speed. Millsap and Law (1996) describe an early system developedby Delphi. They concentrate on on-centre handling performance and use twoparticular measures, which previous work (based on passive steering systems) hasshown to be important: steering sensitivity at 0.1 g lateral acceleration and steeringwheel torque gradient at 0 g. Steering sensitivity is a measure of how much the vehicle



responds to a steer angle input. The authors claim that a desirable characteristic ofsteering sensitivity versus speed is that shown in Figure 7. Typical values of torque

gradient should be in the range 1825Nm/g. They studied the performance of their

Variable Ratio Electric Steering system (VRES) using a detailed model of the steering

system with a 3dof model. The outcome was that the VRES offered the opportunity of

overcoming the compromise in passive systems between steering sensitivity and

torque gradient throughout the speed range. The authors recognized that theperformance measures on which they based their work are quasi-steady-state

measurements, whereas steering feel is also strongly influenced by the transient

phase involving a phase lag between steering input and vehicle response.A later study by Abe et al. (2000) measured the performance of a prototype VGS

system against optimum response parameters (Figure 8), collated in an earlier study

by Weir and Di Marco (1978). The results showed that too much change in steering

gain pushes the yaw response outside the optimum bounds. However, there is no

account for the subjective nature of drivers in this study. Akita et al. (2000) describe theAisin Seiki system, which they installed and tested on a prototype vehicle. From

measured handling results with a range of drivers both professional and

inexperienced they claimed to confirm user benefits of improved vehicle stabilityand reduced workload. They also carried out some interesting failure mode

experiments by setting the system to fail by reverting to a fixed ratio suddenly

during a manoeuvre. However, it is not clear that these results, relating to measured

deviations from some prescribed path, generalize to other manoeuvres. Kojo et al.(2002) describe the practical implementation of a Toyota VGS system. Driving

simulator and field tests show there are different optimal steering ratios fordisturbance rejection and course tracking. The authors use the VGS to add some

derivative steer (Figure 3) and show that less corrective steer is needed by the driver to

reject disturbances.

g/100 deg dsw

100 180Vehicle speed (km/h)

1.4

1

2

0

Figure 7 Desirable characteristic of steering sensitivity withspeed according to Millsap and Law (1996)



There are a number of other studies that examine how active steering wheel torquecontrol affects the vehicle handling. McCann (2000) presents some simulations of aDelphi torque feedback concept, where the steering wheel torque is a function of thedriver input torque and some lead compensated yaw rate feedback. Takehara andSakamoto (2002) have studied a Mazda prototype that tracks yaw rate by Hinf controlof an Electronic Power Assisted Steering (EPAS) system. These studies do not addressthe human factor issues of intervening directly and noticeably in the drivers demand.

2.3 Critical view of yaw rate stability studies

Yaw rate tracking algorithms have been used to improve stability near the limit andhandling feel during normal driving. However, a recurrent problem throughout theliterature is that this distinction is commonly either muddled or blurred. It is fair toobserve that whereas the retention of stability is a clear performance objective, it ismuch less clear what constitutes a performance improvement under normal drivingconditions. Inevitably, this latter goal involves a strong element of subjectiveassessment and interpretation.

The studies of yaw rate stability are dominated by the use of model referencefeedback control of yaw rate. Here the controller attempts to push the non-linearbehaviour towards a linear bicycle model response. As the vehicle behaviourapproaches limiting conditions, the actuation power needed to reduce the trackingerror becomes larger and beyond the bounds of active steering, active wheel loadcontrol and active driveline. Active braking systems offer the most power for

0.0 0.1 0.2 0.3 0.4 0.5

0.1

0.3

0.4

0.5

0.2

0.0

Optimum yaw rate gain atV=80km/h for expert driver

for typical driver

Yaw rategain(1/s)

Time constant (s)

Figure 8 Optimum yaw response parameters



generating corrective yaw moments, but these systems are not without their ownproblems. Braking achieves the required yaw response at the expense of thelongitudinal acceleration demand. For steerability, this is counter-productive.

It is interesting that feedback control only is the norm in active steering studieswhile a combination of feedforward and feedback is used in most braking, drivelineand wheel-load studies. One would expect the phase advance characteristics of afeedforwardfeedback strategy to be beneficial regardless of the system. Indeed, theone piece of work that stands out is by Segawa et al. (2000), who clearly demonstratesthe benefit of a feedforwardfeedback strategy. This work also includes a variable gainsteer algorithm that would be useful for linear handling feel studies. The drawback ofthe study is the requirement of a steer-by-wire capability. However, this idealizedwork is an excellent benchmark on which to develop steering systems that use activeplanetary gear sets.

The linear handling studies rely less on actuation power and more on understand-ing the drivers subjective view of handling feel. While desirable characteristics forlateral acceleration response and yaw rate response to steer angle have been found inseparate studies, the complete dynamic response, including transient phases of yawrate and lateral acceleration, at different speeds are not clear from the research.Further experiments like those carried out by Crolla et al. (1998) are required tocorrelate the subjective handling feel with objective measures of vehicle handling.

3. Sideslip stability

Early studies on sideslip minimization were dominated by linear feedforward controlof ARS systems (Senger and Schwartz, 1987; Whitehead, 1998, 1990; Xia and Law,1990). Figure 9 shows the basic concept. The control algorithm is calculated by solvingthe 2dof bicycle model equations for zero sideslip at steady state.

Some practical applications of this work are given in Sato et al. (1991, 1992),Kawakami et al. (1992) and Tanaka et al. (1992) describing the production system usedon Toyotas 91 Soarer. The control principle follows that in Figure 9, however, thealgorithm to represent the control is modified to Equation (3):

r K1V f K2 _f,V K3V r 3

df = K1 ds + K2r Vehicledf rds

Figure 9 Steady-state zero-sideslip control



The steer gain is scheduled with velocity to overcome the actuator lag. Many of these

early studies examined the performance of the vehicle in the linear range of handling.

However, sideslip stability is more critical near the limit. Shibahata et al. (1992)presented a non-linear analysis of the steady-state performance of Direct Yaw-moment

Control (DYC) systems, which they termed the beta-method. The beta-method gives

an indication of stabilizing yaw moment and stability margin for different sideslipangles and front steer angles, Figure 10. This method allowed the evaluation of a

feedforward DYC control (Shimada and Shibihata, 1994). The DYC algorithm

decouples the roll motion from the lateral dynamics. It is actuated by three different

methods: differential braking left to right, RMD control and active rear wheel steering.

The conclusions of this study summarize where these three systems have an effect on

vehicle handling:

(1) Leftright differential braking can control sideslip over a full range of vehiclemotions;

(2) RMD is only effective for when the lateral acceleration is44 m/s2;(3) ARS is effective at only small sideslip angles, and near the limit reduces available

yaw moment to control the vehicle.

The Honda Active Torque Transfer System (ATTS), described by Kuriki and Shibihata

(1998), is used to implement DYC control on the front wheel drive Honda Prelude TypeSH. Results show improved handling performance during combined cornering and

acceleration/deceleration and improved stability. Torque steer in front wheel drive

vehicles is also reduced by ATTS leading to a reduction in steering effort. An additional

benefit is a reduction in tyre side force, which will reduce tyre wear during cornering.These conclusions are true for steady-state behaviour only. The beta-method

assumes there are no dynamics in the jump from one vehicle state to another.

To increase the fidelity of the analysis, researchers suggested the beta-phase plane

analysis (Inagaki et al., 1994), where the rate of sideslip angle is plotted against the

b ()0

10

6

4

d = 3d = 4

d = +2

d = +4

Yawmoment(KNm)

Figure 10 The beta-method to analyse sideslip stability



sideslip angle (Figure 11). Analysis of these plots gives much more information on thedynamic response, such as sideslip damping and natural frequency. In addition,

the beta-phase plane is further used for feedback control that ensures the sideslip

behaviour stays within defined bounds.The phase plane tool can be used to determine the necessary control action for a

given sideslip and sideslip rate. This is done by measuring how far the state is from

the stable region and scaling the control action accordingly; the control action is then

mapped to an appropriate brake action. Yasui et al. (1996a,b) shows someexperimental results of this approach on a prototype Aisin Seiki system that uses

active braking.Several researchers are now studying the effect of integrating several vehicle

controller subsystems for sideslip control. For example, Smakman (2000) compares the

performance of a braking intervention system and an integrated braking and wheel

load control system for lateral vehicle motion control. This work shows that braking

intervention has the most significant effect on the lateral dynamics of the vehicle,though it is invasive to driving demands. In comparison, wheel load control has little

effect on the longitudinal dynamics but cannot generate the high moments required.

The integrated approach splits the yaw moment demand from the DYC into separate

demands for the brake controller and the wheel load controller. The co-ordination

block applies wheel load control until it reaches saturation and then applies brake

intervention. The work in the paper also examines the effect of different loads atdifferent lateral accelerations, and shows that until the lateral acceleration is greater

than 0.3 g, the effect of wheel load variation is minimal. Konik et al. (2000) reports thatthe frequency their drivers reach 0.3 g and beyond is no more than 67% of all lateral

manoeuvres.A previous study by Alleyne (1997) shows that a performance enhancement of

59% can be seen in longitudinal deceleration with an integrated Anti-lock Braking

System (ABS) and active suspension system. However, it is questionable as to whether

Controlboundaries

b (/s)

b ()Stable

Unstable

Unstable

Figure 11 The -phase plane control



both these works go far enough in exploiting the interaction between braking andsuspension systems. Selby et al. (2002) and He et al. (2004) examined the integration ofactive braking/driveline systems with an AFS system to delay the onset of braking.In both cases, it was clear that the integrated approach delayed intervention in thelongitudinal dynamics significantly.

There is a body ofworkmotivated towards robust control to allow for non-linearities,mismodelling and parameter variations. Abe et al. (1999, 2000a,b, 2001) andFurukawa and Abe (1996) used sliding mode control to track a desired sideslipresponse. As the desire for increased robustness increases, the controller requirementsincrease also. In particular, there becomes a strong need to estimate vehicle sideslipangle. These studies estimate the vehicle sideslip angle by estimating the rate ofchange of lateral velocity and from a tyre model. By integrating this and monitoringthe vehicle speed through wheel speed sensors, the estimation is more robust.

3.1 Critical view of sideslip control studies

The large number of studies on sideslip control using ARS resulted in little uptake inproduction vehicles. The benefits of ARS on stability and safety are just not significantwhen compared with braking. For stability and safety, the invasive braking is not anissue. The large amount of production vehicles with active safety via DYC withbraking is testament to its ability. There are arguments that brake-based stabilitycontrollers intervene too early in the longitudinal dynamics of the vehicle. Workcarried out by Selby (2003) and He (2005) have shown that an integrated approach candelay the onset of braking.

A number of non-linear control concepts for sideslip stability have been examined.To be commercially viable, these systems require an estimate of sideslip to operate.There is little in the literature explaining how sideslip is estimated in commercialsystems. In the theoretical literature, there are three approaches:

integration of the lateral acceleration measurement on-board tyre model a combination of integration and tyre model.Integrating the lateral acceleration measurement is prone to drift and the on-boardtyre model would rely on an accurate measure of the road surface coefficient offriction. The work in Abe et al. (2001) and Furukawa and Abe (1996) on combiningintegration and the on-board tyre model gives the most likely approach to obtaining areliable and accurate measurement.

4. Integrated yaw rate and sideslip stability control

There are two main approaches to this problem: systems that use only one actuatorand systems that use two or more. The principle of controlling two values with only



one actuator is dealt with using Hinf control. The first process is to shape the vehicleyawsideslip response to first-order dynamics. Following this, the sideslip iscontrolled using the convention zero-sideslip law. Mammar and Baghdassarian

(2000) and Mammar and Koenig (2002) present a comprehensive theoretical study on

this work on AFS. Some experimental studies are presented by Hirano and Fukutami

(1996) on a Toyota vehicle. Yoshioka et al. (1998, 1999) have adopted two slidingsurfaces, where the target yaw rate is calculated from the output of a sideslip sliding

surface.Integrated control of two or more actuators for combined yaw and sideslip control

has been applied with both linear and non-linear control strategies. Nagai et al. (1996,1997, 1998, 2002) present the control concept in Figure 12. The feedforward control

calculates the actuator demands from an inverse model; the feedback control is a statefeedback gain that can be calculated using Linear Quadratic Regulator (LQR), Linear

Quadratic Gaussian (LQG) or Hinf schemes. The aim is similar to earlier works ofAckermann (1997), to decouple the disturbance rejection and steering tasks. Here,

though, there is active control of the braking and front steering, Kleine and Van

Niekerk (1998) follows a similar approach using 4-wheel steering (4WS) with both

AFS and rear wheel steering.Smakman (2000) adopts an integrated approach that uses internal model control

for yaw rate tracking and the phase-plane approach for sideslip stability. The

integration strategy is heuristic and involves a gradual shift of authority between

wheel-load control and brake control. Selby (2003), He (2005) and Cooper et al.(2004) adopt a similar approach for brakingsteering control, drivelinesteering and

drivelinewheel load control. In all cases, the integrated strategy performs better

than the combined stand-alone controllers. These studies were carried out in

simulation, with idealized actuators and assumptions on the availability and

accuracy of measurements such as sideslip and longitudinal speed. It is not clear if

REF

FF

FB VEH

uff

ufb

bdesds rdes

br

Figure 12 Feedforward and feedback control of both sideslip andyaw rate



the model-based control algorithms used in these studies will perform satisfactorilyon a real vehicle.

The best practical application of the studies on sideslip and yaw rate control is byVan Zanten et al. (1995, 1996, 1998) and Van Zanten (2000). They adopt the strategy inFigure 12 and realize it through ABS and Traction Control Systems (TCS). However,the key importance is in the observation of sideslip angle and estimation of the roadsurface coefficient of friction. The sideslip angle is observed through solving thelateral dynamics and applying it to a Kalman Filter. The coefficient of friction is foundby calculating the longitudinal and lateral forces from slip estimates, and by using therelationship below.

Fnm F 2B F 2S

q4

This estimation of m is used in the reference model to ensure the desired yaw rate isnot too large.

4.1 Critical view of combined yaw rate following and sideslip stability studies

The work on integrated control can be summarized as

(1) good theoretical work with no indication of the practical implications;(2) excellent practical work with little indication of control algorithms used.

It is evident from the theoretical approaches that a switching style of integration ispreferred where control authority is gradually moved from one actuator to the next.While this methodology clearly works for the systems investigated, it is not clear howit will be extended to incorporate:

(1) additional actuators;(2) additional control objectives vehicle ride, roll control, speed control, path tracking.

There is a concern that as the number of control objectives increases, and the number ofsystems increases, then the bottom-up integration approach will be more difficult todesign and predict. Manning et al. (2002) and Gordon et al. (2003) have both proposedmethodologies for integrating any number of vehicle systems. It is of the view thatapproaches like these offer better solutions in the long-term for system integration.

The real value from the experimental work is the detail on the practicalimplementation of the Bosch Electronic Stability Program (ESP) (Van Zanten, 2000;Van Zanten et al., 1995, 1996, 1998), particularly in understanding the control anddetails on sideslip and m-estimation. However, there is no indication as to (1) theaccuracy of the estimates and (2) the controller sensitivity to errors in the estimatedvalues.



5. Conclusions

A survey of 68 key papers from the 250 available on yaw and sideslip control has beencompleted. From this review the following conclusions can be drawn:

There is obvious potential of using active steering and VGS to improve the yaw rateresponse in the linear handling regimes. Subjective evaluations are needed to justify theapproaches and further work is needed to embed the subjective nature of drivers in theoverall design methodology. Overall, there is little research that examines how the high-level performance criteria in Figure 1 map down to the vehicle motion objectives.Studies in affective design may allow criteria such as sporty and classic to be engineeredinto the handling design.

AFS, driveline and wheel-load control can push the yaw rate performance envelopefurther towards limit handling but the bandwidth and saturation levels of theseactuators make them unsuitable for more severe stability manoeuvres. Studies by Heet al. (2004), Hac and Bodie (2002) and Smakman (2000) give the best analysis of whichsystems are most appropriate for different regimes of handling.

The selection and comparison of actuators for lateral motion control systems have notbeen clearly addressed in terms of the wider issues such as power consumption, costand weight. This is clearly more important for systems that operate frequently in thelinear handling regime.

Brake-based systems offer the best solution for pure safety and stability, yaw rate orsideslip but do interfere in the drivers longitudinal speed demand.

More advanced sideslip stability controllers require an estimate of sideslip. A combinedapproach using the vehicle model and integration of the lateral acceleration sensor givesthe best results. An estimate of the road surface coefficient of friction is necessary toadjust the reference model demands for different surfaces. This is more straightforwardwhere a sideslip estimate is already available. There is little work examining thesensitivity of the controllers to errors in the estimates of sideslip angle and road surfacecoefficient of friction.

Integrated approaches that use different systems in different areas of the vehiclehandling regime offer the best solution. These studies have shown how an integratedapproach can be used to delay the onset of braking for stability control. A bottom-upapproach is preferred in the published literature. However, it is not clear how thismethodology can be extended to include additional control objectives and/or actuators.

References

Abe, M., Kano, Y., Shibihata, Y. andFurukawa, Y. 1999: Improvement of vehiclehandling safety with vehicle sideslip controlby direct yaw moment. Proceedings of the 16thIAVSD.

Abe, M., Kano, Y., Suzuki, K., Shibihata, Y.and Furukawa, Y. 2000a: An experimentalvalidation of side-slip control to compensate

for a loss of stability due to nonlinear tyrecharacteristics. Proceedings of Proceedings ofAVEC 2000, August, Ann Arbor, MI.

Abe, M., Kano, Y., Suzuki, K., Shibihata, Y.and Furukawa, Y. 2001: Side-slipcontrol to stabilize vehicle lateral motionto direct yaw moment. JSAE Review 22,41319.



Abe, M., Shibihata, Y. and Shimuzu, Y. 2000b:Analysis of steering gain and vehicle hand-ling performance with variable gear-ratiosteering system (VGS). Proceedings of FISITA2000, 1215 June, Seoul.

Ackermann, J. 1990: Robust car steering byyaw rate control. Proceedings of the 29thConference on Decision and Control, 203334.

Ackermann, J. 1992: Robust yaw damping ofcars with front and rear steering. Proceedingsof 31st Conference on Decision and Control,258690.

Ackermann, J. 1997: Robust control preventscar skidding. IEEE Control Systems Magazine17, 2331.

Ackermann, J. and Bunte, T. 1997: Actuatorrate limits in robust car steering control.Proceedings of the 36th Conference on Decisionand Control, San Diego, CA, 472630.

Ackermann, J. and Sienel, W. 1993: Robustyaw damping of cars with front and rearsteering. IEEE Transactions on Control SystemsTechnology 1, 1520.

Ackermann, J., Bunte, T., Sienel, W., Jeebe, H.and Naab, K., 1996: Driving safety by robuststeering control. Proceedings of AVEC96, June,Aachen, 37794.

Akita, T., Satoh, K., Kurimoto, M. andYoshida, T. 2000: Development of the activefront steering control system. Proceedings ofFISITA 2000, 1215 June, Seoul.

Alleyne, A. 1997: Improved vehicle perfor-mance using combined suspension andbraking forces. Vehicle System Dynamics 27,23565.

Cooper, N.J., Manning, W.J., Crolla D.A. andLevesley, M.C. 2004: Study of the integrationof roll control and torque distribution.Proceedings of AVEC 2004, 2327 August,Arnhem, 51318.

Crolla, D.A., Chen, D.C., Whitehead, J.P. andAlstead, C.J. 1998: Vehicle handling assess-ment using a combined subjectiveobjectiveapproach. Proceedings of the Society ofAutomotive Engineers Congress 1998, Detroit,MI.

Donges, E. 1996: Supporting drivers by chassiscontrol systems. In Pauwelussen, J.P. &Pacejka, H.B., editors. Smart vehicles. Swetsand Zeitlinger, pp. 27696.

Furukawa, Y. and Abe, M. 1996:On-board-tire-model reference control forcooperation of 4WS and direct yaw momentcontrol for improving active safety of vehicle

handling. Proceedings of AVEC 1996,September, Aachen, 50726.

Gordon, T., Howell, M. and Brandao, F. 2003:Integrated control methodologies for roadvehicles. Vehicle System Dynamics 40, 15790.

Hac, A. and Bodie, M.O. 2002: Improvementsin vehicle handling through integrated con-trol of chassis systems. International Journal ofVehicle Design 29, 2350.

He, J. 2005: PhD thesis, School of MechanicalEngineering, University of Leeds.

He, J., Crolla D.A., Levesley, M.C. andManning, W.J. 2004: Integrated chassis con-trol through coordination of active frontsteering and intelligent torque distribution.Proceedings of AVEC 2004, 2327 August,Arnhem, 33341.

Hirano, Y. and Fukatani, K. 1996:Development of Robust active rear steeringcontrol. Proceedings of AVEC96, September,Aachen, 35976.

Inagaki, S., Kshiro, I. and Yamamoto, M. 1994:Analysis on vehicle stability in criticalcornering using the phase-plane method.Proceedings of AVEC94, September, Tsukuba,28792.

Kawakami, H., Sata, H., Tabata, M., Inoue, H.and Itimaru, H. 1992: Development ofintegrated system between active controlsuspension, active 4WS, TRC and ABS.Proceedings of SAE 1992, Detroit, MI.

Kleine, S. and Van Niekerk, J.L. 1998:Modelling and control of a steer-by-wirevehicle. Vehicle System Dynamics Supplement28, 11442.

Kojo, T., Suzumara, M., Fukui, K., Sugawara,T., Matsuda, M. and Kawamuro, J. 2002:Development of front steering controlsystem. Proceedings of AVEC 2002,September, Hiroshima.

Konik, D., Bartz, R., Barnthol, F., Bruns, H.and Wimmer, M. 2000: Dynamic drive thenew active roll stabilization system fromBMW system description and functionalimprovements. Proceedings of AVEC 2000,August, Ann Arbor, MI.

Kramer, W. and Hackl, M. 1996: Potentialfunctions and benefits of electronic steeringassistance. Proceedings of FISITA96 WorldCongress, 1721 June, Prague.

Kuriki, N. and Shibihata, Y. 1998:Development of active torque transfer sys-tems. Proceedings of FISITA 1998, September,Paris.



Leffler, H. 1996: Electronic brake managementEBM a possible approach for brake andcontrol system integration. Proceedings ofFISITA96 World Congress, 1721 June, Prague.

Leffler, H., Auffhammer, R., Heyken, R. andRoth, H. 1998a: New driving stability controlsystem with reduced technical effort andmedium class passenger cars. Proceedings ofSAE 1998, Detroit, MI.

Leffler, H., Krusche, H., Auffhammer, R.,Roth, H. and Vieler, H. 1998b: Accidentprevention by driving stability control forcompact passenger cars. Proceedings of FISITA1998, September, Paris.

Mammar, S. and Baghdassarian, V.B. 2000:Two-degree-of-freedom formulation of vehi-cle handling improvement by active steering.Proceedings of the American Control Conference,105109.

Mammar, S. and Koenig, D. 2002: Vehiclehandling improvement by active steering.Vehicle System Dynamics 38, 21142.

Manning, W.J., Brown, M.D., Crolla, D.A. andSelby, M.A. 2002: IVMC: Intelligent vehiclemotion ccontrol. SAE Journal of PassengerCars: Electronic and Electrical Systems 111,40816.

Matsumoto, S., Yamaguchi, H., Inoue, H. andYasuno, Y. 1992: Braking force distributioncontrol for improved vehicle dynamics.Proceedings of AVEC 92, September,Yokohoma, 44146.

Matsuno, K., Nitta, R., Inoue, K., Ichikawa, K.and Hiwatashi, Y. 2000: Development of anew all-wheel-drive control system.Proceedings of FISITA 2000, 1221 June, Seoul.

McCann, R. 2000: Variable effort steering forvehicle stability enhancement using anelectric power steering system. Proceedingsof SAE 2000, Detroit, MI.

Millsap, S.A. and Law, E.H. 1996: Handlingenhancement due to an automotive variableratio electric power steering system usingmodel reference robust tracking control.Proceedings of SAE 1996, Detroit, MI.

Motoyama, S., Uki, H., Isoda, K. and Yuasa,H. 1993: Effect of traction force distributioncontrol on vehicle dynamics. Vehicle SystemDynamics 22, 45564.

Nagai, M., Hirano, Y. and Yamanaka, S. 1997:Integrated control of active rear wheel steer-ing and direct yaw moment control. VehicleSystems Dynamics 27, 35770.

Nagai, M., Hirano, Y. and Yamanaka, S. 1998:Integrated robust control law of active rear

wheel steering and direct yaw momentcontrol. Vehicle Systems Dynamics Supplement28, 41621.

Nagai, M., Shino, M. and Gao, F. 2002: Studyon integrated control of active front steerangle and direct yaw moment. JSAE Review23, 30915.

Nagai, M., Yamanaka, S. and Hirano, Y. 1996:Integrated control law of active rear wheelsteering and direct yaw moment control.Proceedings of AVEC96, September, Aachen,45170.

Naito, G., Yaguchi, E. and Ozaki, K. 1992:Improving vehicle dynamics by torque splitcontrol system. Proceedings of AVEC 92,September, Yokohoma, 458463.

Sato, H., Inoue, H., Tabata, M. and Inagaki, S.1992: Integrated chassis control system forimproved vehicle dynamics. Proceedings ofAVEC 92, September, Yokohoma, 41318.

Sato, H., Kawai, H., Isikawa, M., Iwata, H.and Koike, S. 1991: Development of fourwheel steering system using yaw rate feed-back control. Proceedings of SAE 1991,Detroit, MI.

Segawa, M., Nishizaki, K. and Nakano, S.2000: A study of vehicle stability control bysteer by wire system. Proceedings of AVEC2000, Ann Arbor, MI.

Selby, M.A. 2003: PhD thesis, School ofMechanical Engineering, University ofLeeds.

Selby, M.A., Manning, W.J., Crolla, D.A. andBrown, M.D. 2002: A comparison of therelative benefits of active front steering andactive rear steering when co-ordinated withdirect yaw moment control. ASMEInternational Mechanical Engineering Congressand Exposition. New York, 2001.

Senger, K.-H. and Schwartz, W. 1987: Theinfluence of a four wheel steering system onthe stability behaviour of a vehicledriver system. Vehicle System DynamicsSupplement 17.

Shibihata, Y., Shimida, K. and Tomari, T. 1992:Improvement of vehicle maneuverabilityby direct yaw moment control. Proceedingsof AVEC 92, September, Yokohoma.

Shimada, K. and Shibihata, Y. 1994:Comparison of three active chassis controlmethods for stabilising yaw moments.Proceedings of the SAE 1994, Detroit, MI.

Smakman, H. 2000: Functional integration ofactive suspension with slip control forimproved lateral vehicle dynamics.



Proceedings of AVEC 2000, August, AnnArbor, MI.

Straub, A. 1996: DSC (dynamic stability con-trol) in BMW 7 series cars. Proceedings ofAVEC96, June, Aachen, 54757.

Takehara, S. and Sakamoto, K. 2002: Controllogic design and development of electricpower-assisted steering (EPAS). Proceedingsof AVEC 2002, September, Hiroshima.

Tanaka, H., Inoue, H. and Iwata, H. 1992:Development of a vehicle integratedcontrol system. Proceedings of the IMECHE1992, 6374.

Van Zanten, A.T. 2000: Bosch ESP systems: 5years of experience. Proceedings of the SAE2000, Detroit, MI.

Van Zanten, A.T., Erhardt, R., Landesfeind, K.and Pfaff, G. 1998: VDC the Vehicledynamics control system of Bosch.Proceedings of the SAE 1998, Detroit, MI.

Van Zanten, A.T., Erhardt, R., Pfaff, G., Kost,F., Hartmann, U. and Ehret, T. 1996: Controlaspects of the Bosch-VDC. Proceedings ofAVEC96, September, Aachen.

Van Zanten, A.T., Erhardt, R. and Pfaff, G.;VDC. 1995: The vehicle dynamics controlsystem of Bosch. Proceedings of the SAE 1995,Detroit, MI.

Weir, D.H. and Di Marco, R.J. 1978,Correlation and evaluation of driver-vehicledirectional handling data. Proceedings of theSAE 1978, Detroit, MI.

Whitehead, J.C. 1988: Four wheelsteering: maneuverability and high speedstabilisation. Proceedings of the SAE 1988,Detroit, MI.

Whitehead, J.C. 1990: Rear wheel steeringdynamics compared to front steering.Transactions of the ASME 112, 8893.

Xia, X. and Law, E.H. 1990: Nonlineardynamic response of four wheel steeringautomobiles to combined braking andsteering commands in collision avoidancemaneuvers. Proceedings of SAE 1990,Detroit, MI.

Yasui, Y., Tozu, K., Hattori, N. and Nishii, M.1996a: Vehicle stability enhancement usingbrake control. Proceedings of FISITA 1996,June, Prague.

Yasui, Y., Tozu, K., Hattori, N. andSugisawa, M. 1996b: Improvement of vehicledirectional stability for transient maneuversusing active brake control. Proceedings of SAE1996, Detroit, MI.

Yoshioka, T., Adachi, T., Butsuen, T.,Okazaki, H. and Mochizuki, H. 1998:Application of sliding-mode control to con-trol vehicle stability. Proceedings of AVEC 98,September, Nagoya, 45560.

Yoshioka, T., Adachi, T., Butsuen, T.,Okazaki, H. and Mochizuki, H. 1999:Application of sliding-mode theory to directyaw-moment control. JSAE Review 20,52329.



162_A Review of Yaw Rate and Sideslip Controllers for Passenger Vehicles

Documents

Transcript of 162_A Review of Yaw Rate and Sideslip Controllers for Passenger Vehicles