Study and design of motion base for driving simulators

MotivationMany motion base simulators have been developed in the last thirty yearsfor dierent types of vehicles. Mainly two main types of simulators can bedistinguished: xed and motion base. The former are based solely on visualand instrument cues while the latter also provide the pilot with realisticmotion cues. It has been often acknowledged that a good motion base cansignicantly enhance simulation realism.Flight simulators have been the reference point in the eld of vehicle sim-ulation for the last 30 years. This has been due to the high costs of aircrafts,if compared to other vehicles as cars or motorcycles. Flightsimulators havealways been less expensive than the actual aircraft they were trying to repro-duce, thus allowing pilots and crews to be trained at lower costs and lowerrisks. The same cannot be said for car and motorcycles and here lays one ofthe basic dierences between such types of simulators. Land vehicle simu-lators have been developed with dierent purposes, most often as a tool fordesigners to test new prototypes before actually building them or to studyhuman behavior in specic situations.Car simulators have become more and more common in the last ten years.Now, most automotive companies haveavailable some kind of simulators eventhough not all of them are motion base. An example of this is the NADS carsimulator, which is the most technologically evolved car simulator to date.Suchsimulator is an internationally funded facility for conducting controlledexperiments for drivers, vehicles, highways and o-road situation in a highlyre-congurable synthetic environment.The present work would be a contribution in the eld of simulators re-search with particular emphasys in the design of the motion base of such atype of devices.It has been organised as follows. After this short introduction, in Chapter1 a general analysis of the State-of-the-art on Simulators and some generalconsiderations derived by the analysed driving simulators will be reported.In Chapter 2 the Inertial Feedback of a simulator, the part that allows tomove the Rider in the 3D space in order to simulate the real movements ofa vehicle will be introduced and described and it will be proposed a designprocedure that allows to determine the requirements of a possible solution ofa simulator actuation subsystem starting from a set of nite typical vehiclexxi
Figure 1: NADS Simulator Overviewmaneuvers. In Chapter 3 it will be reported the application of the design toolin the MORIS Project. The MORIS project was a collaboration betweenEuropean universities and industries with the scope to realise a completemotorcycle simulator. Finally, in Chapter 4 some conclusions will be reportedregarding the proposed design procedure and the realised simulator and thepossible enhancements.
xxii
Chapter 1INTRODUCTION
In this chapter the denition of simulator, the utility of suchatype of deviceand the analysis of the State-of-the-Art of several driving simulators areintroduced.1.1 SimulatorsWhat is a simulator? The rst step to nd an answer around this questionis to open a dictionary [P+00] and take a look at the denition of this word.Simulator: One that simulates, especially an apparatus thatgenerates test conditions approximating actual or operationalconditions.This denition means that each activity and each device in which onetries to replicate the reality without use the normal instruments to performthe task, can be called \simulator".As example, one can simulate a battle with a miniature landscape ofthe battleeld and some models of the friendly and enemy troops. Anotherpossibility is to simulate the urban traÆc in a city with a computer usingtraÆc nodes, the matrix origin{destination with the scope to evaluate theresults of a lamp cycle changing in a cross or the modication of a waydirection.As one can immediately recognise, here it is present a great advantage insimulate something: one can "stop" the simulation every time, change someparameters and restart. For instance, if the general of the simulated battlediscover that the troops in a corner cannot advance because they have in fronta deep river, he/she can stop the simulated battle, place the troops in a newinitial position and restart the battle. If this unpleasantevent occurs duringa real battle, this general may lost the war due the unhappy placement ofthe troops. If the civil engineer that is running the traÆc simulation discoverthat some traÆc nodes are not well connected, he/she can stop the traÆc1
1.1. Simulators 1. INTRODUCTIONsimulation, solve the problem bychanging, for instance, a way direction andrestart the simulation. If this event happens during a real traÆc day, theurban drivers may remain entrapped in a queue.Second point is that one desires to obtain realistic results from virtualactions. As a consequence the simulator shall contains models that are agood representation of the reality. If the general of the simulated battledoes not know the consistency of the enemy troops and does not possess agood model to replicate the enemy ght plane, surely the results obtainedfrom the simulation are not realistic. For instance, if the general supposesthat the enemies forces are composed only by n soldiers, he/she can simulatesuccessfully a front attack with his n2 soldiers. But if the enemy possessesn=2 soldiers but n cannons, one can easily think that the front attack strategywill become a suicide attack. If the civil engineer possesses a linear model ofthe behavior of the drivers in the urban traÆc, he/she may obtain a solutionthat, when applied, has a completely dierent result as the real case due thehighly non{linear (and maybe casual) modelisation of a human behavior.As conclusion, a simulator has the advantage to reduce a physical probleminto a "virtual" timeless dimension in which it is possible to prove moreand more the event until some encouraging results are obtained. But italso possesses the drawback that it is necessary to operate with instrumentsand/or models capable to correctly represent the real outgoing of the thingsin the "virtual" dimension. In the case of the general, he/she can send somespies at the enemy eld in order to gain more knowledge around the enemyforces, he/she can know the personality of the opposite general in order toplan the eective enemy strategy. In the case of the civil engineer, he/shecan use powerful computer with good models of the city streets and of thehuman drivers in order to obtain realistic behavior of the cars in the urbantraÆc.The development of a simulator depends on the capacity of the humanbeing to study, realise and implement models that reproduce the reality. Asa consequence, simulating a war is a task that has been performed fromcenturies, while simulation of urban traÆc, that requires technologies anddevices available only in the last years, has been performed only recently.The possibility to use standard, cheap, reliable computers, mechanicalcomponents, software and so on, has speed up the research activities in theeld of simulators in the last fourteenth years. One can simulate, beforethe eective development, an electric circuit, a chemical plant activity, arobotised assembly station, a surgery operation and so on.Concentrating the attention only to driving simulators, i.e. simulatorsin which the interaction of a person with a vehicle is performed, one canrecognise that the problem of simulation is bounded with the vesensesof aperson: touch, smell, taste, sight and hearing. The perfect simulation is theone in which ahuman cannot distinguish the same real phase because all of2
1. INTRODUCTION 1.2. State of the Art Analysishis senses are tricked in some manner. At this stage of the progress this isnot possible. But, fortunately, the importance of the senses depends on theanalysed task. Simulating a driving simulator, the smell and taste senses canbe neglected while sight, touch and hearing are the most important senses.This and other typical simulators capabilities can be evinced studying thetechnical literature around this argument. Several industries and researchlaboratories around the world are investigating on driving simulators. In thefollowing section a resume of dierent kinds of applications will be reported.1.2 State of the Art AnalysisIn this paragraph it will be reported the analysis of some simulators realisedor under realisation around the world. This analysis is helpful in order tostudy about the general guidelines adopted and to evidentiate the basic so-lutions selected. The main characterisation of a driving simulator can be thefollowing: xed-base,nomotion feedback is provided on the driver that is solidalwith the ground. The dynamic eects of a real life conditions (acceler-ations and velocities) are replicated only with the sight and the hearingof the driver; motion-based, in which between the driver and the ground it is in-troduced a mechanical devices that reproduce the dynamic eects, inaddition to the sight and hearing feedbacks.The aircraft simulation literature reports of benets in using motion-based simulators, among which [GB+94] : a) reduced training time, b) re-duced subject variability, c) enhanced operator performance, d) decreasedsimulator discomfort sensitivity to visual system delays. For those reasonsthe conceptual design of several simulators is based on a motion-based type.As a consequence, in this state-of-the-art analysis the xed-base simulatorsare not reported. The characteristics of the motion-based simulators havebeen itemized as follows:1. KinematicsThe adopted kinematics of the mechanical system moving the vehiclemock up have been described in terms of number and distribution ofthe degrees of freedom (DOFs); where data are available, references toallowable range of motion, velocity and acceleration for each DOF aregiven.2. Dynamics Mathematical ModelInformation on the mathematical model that simulates the vehicle dy-namics. 3
1.2. State of the Art Analysis 1. INTRODUCTION3. Control DevicesThe presence of control devices such as steering handlebar, brakelevers,gas throttle, clutch lever, gear shift lever, are indicated together withtheir force feedback (if any).4. ActuatorsThe adopted devices for the actuation of the DOFs of both the movingsystem and the control devices are described in terms of type and maincharacteristics.5. SensorsThe adopted devices for the position/force sensorisation of the DOFs ofboth the moving system and the control devices are described in termsof type and main characteristics.6. Real-time ComputingThe main characteristics of the real-time system included in the simu-lator.7. Feedbacks ProvideThe adopted interface devices for the main feedbacks have been de-scribed in terms of type, characteristics and performances where dataare available.8. Virtual RealityThe main characteristics of the virtual scenario (i.e. environment, stilland moving objects) generated on the screen by the simulator.9. Simulated ManeuversThe maneuvers that can be performed by the rider while using thesimulator.1.2.1 The HONDA Motorcycle Riding SimulatorThe HONDA Motor Company begun in the 1970 the TraÆc Safety PromotionActivity. Since 1996 this Promotion Activity started with the TraÆc SafetySimulation with Riding Simulator Program in which a motorcycle simulatorwas developed in order to experience as many user as possible about thedangerous situation during a motorcycle riding. The last version of thisseries of simulators is the devices described in the following [Com99].1. KinematicsTwo DOFs: roll angle of 15Æ and pitch angle of 10Æ combined in aserial structure. 4
1. INTRODUCTION 1.2. State of the Art Analysis
Figure 1.1: HONDA Simulator2. Dynamics Mathematical ModelSimplied Dynamic Model in which only the pitch angle is bound withthe simulator DOF.3. Control DevicesSteering handlebar, gas throttle, gearshift lever, brake levers; only therst has an active force feedback; steering range of motion: 30Æ.4. ActuatorsAll electric DC Motors.5. SensorsForce sensors on the steering handlebar, force sensor on the pedals,rotoidal transducer on the acquisition devices.6. Real-time ComputingNo data available.7. Feedbacks ProvideAll resents. Visual feedback realised via a 52 inches screen with 15 : 9screen ratio. Two lateral fan simulate the wind action on the Rider.Two lateral speakers simulate the acoustical feedback.5
1.2. State of the Art Analysis 1. INTRODUCTION8. Virtual RealityEvans & Sutherland VE engine. Urban and Highway traÆc can besimulated. Test circuit can be tested.9. Simulated ManeuversAll the possible motorcycle maneuvers can be simulated.1.2.2 The DAIMLER-BENZ simulator
Figure 1.2: DAIMLER-BENZ SimulatorIn [Kad95] a car/track simulator developed at DAIMLER-BENZ is de-scribed with the following characteristics:1. Kinematics7 DOFs (wide lateral, lateral, longitudinal, vertical, yaw, roll, pitchmovements). The rst DOF is the wide lateral movement obtainedusing a linear actuated guide; The other 6 DOFs are realised by meansof a Stewart Platform. In table 1.1 are reported the main features ofthe moved mechanical frame.2. Dynamics Mathematical ModelThe dynamic equations are set up using the D'Alembert's principlewith 18 to 37 DOFs. The model is completely 3-D and features non-linear kinematics and elastokinematics for axles and steering system.The achievable frame time for calculating the vehicle dynamics systemincluding input/output is < 6 ms.6
1. INTRODUCTION 1.2. State of the Art AnalysisDOFs Disp. Value Vel. Value Acc. Valuewide lateral 2:30m 4m=s 0:7glateral 1:50m 1m=s 1:2glongitudinal 1:88m;+1:38m 1m=s 1:2gvertical 1:07m 1m=s 1:2gyaw 47:4Æ 25Æ=s 200Æ=s2roll 30Æ 25Æ=s 200Æ=s2pitch 32Æ 25Æ=s 200Æ=s2Table 1.1: Daimler-Benz simulator workspace3. Control DevicesSteering wheel, accelerator, brake, clutch, gear shift lever; active forcefeedback on steering wheel, accelerator, brake, clutch.4. ActuatorsHydraulic servocylinder for the 7 DOFs of the mechanical system mov-ing the vehicle mock-up.Stewart platform: frequency range 3:0 Hz,smoothness  0:2 m=s2; additional lateral movement: frequency range5:0 Hz, smoothness  0:1 m=s2. A hydraulic actuator acts on thesteering assembly, which is set back behind the steering box. No infoavailable for the actuation of the control devices.5. SensorsNo data available.6. Real-time ComputingMotion and driving dynamics handled by two identical systems ofDEC 3000=900S AXP with a performance of 264 SPECfp92 and 189SPECint92. Real Time I/O System: Turbochannel VME Interface.Bus rate: 100 MB/s.7. Feedbacks ProvideVisual: 6 color video projectors (xed with respect to the mock-up);resulting angle of vision 180Æ  35Æ front ,48Æ  20Æ rear.Sound: no data available.8. Virtual RealityImage generator: ESIG 3000, 6 channels (5 for forward view, one forrear view) with resolution of 1024  926 pixels. 252 moving models.70 85 ms transport delays.9. Simulated ManeuversVirtually all maneuvers performed by a car. Possibility of testing truck7
1.2. State of the Art Analysis 1. INTRODUCTIONbehavior.1.2.3 IOWA State University SimulatorThe IOWA State University [Uni95] simulator is an automotive simulatordeveloped at the Iowa State University(provided with 3 interchangeable, in-strumented cabs mountable on a common interface: a Ford Taurus, a GMSaturn, and a U.S. Army HMMWV). It possesses the following characteris-tics:1. Kinematics6 DOFs Hexapod motion base (Stewart platform). Motion envelope:1:02 m horizontal; 0:92 m vertical. Accelerations: 1:1 g to 3:5Hz.2. Dynamics Mathematical ModelBased on the Real-Time Recursive Dynamics (RTRD) kernel, whichis a general purpose multibody dynamics package based on a mini-mal, joint coordinate formulation of the equations of motion for rigidmultibody systems. Used to represent vehicle chassis, suspensions andsteering linkage systems. Other models encapsulated in the RTRD are:powertrain, tires, steering, braking, aerodynamics.3. Control DevicesSteering wheel (with a DC servomotor connected to the original steeringcolumn to provide tactile feedback to the driver), accelerator and brakepedals, transmission gear selection lever, emergency brake. Completevehicles are used for car simulation and complete cabs in the case ofcommercial vehicles.4. ActuatorsNo data available.5. SensorsNo data available.6. Real-time ComputingThree real-time computer systems: an Alliant FX2800 parallel com-puter which features 26 high-speed i860 RISC processors; a HarrisNighthawk 4404, which features four 88000 processors; a Harris Nighthawk5806, which features six 88110 processors.7. Feedbacks ProvideVisual: images projected by 4 projectors xed respect to the plat-form and projecting on the dome's inner surface. Visual eld of view:8
1. INTRODUCTION 1.2. State of the Art Analysis190Æ40Æ front,60Æ40Æ rear. Sound: 6voice digital sampling work-stations. Multiple speakers placed throughout the dome. Typicallythere are 4 speakers for the non-specic environmental noise, plus ad-ditional speakers in the wheel wells. The system supports cross-channelfading and phasing of sounds to provide directionality cues.8. Virtual RealityEvans & Sutherland ESIG-2000 image generator. 4 channels, eachcongured to display 786000 pixels with a 30 Hz refresh rate. Thevirtual scenario subsystem models semi-autonomous vehicles, regulatestraÆc control devices, controls lighting and weather conditions andmodels any other non-static entity in the virtual environment. Eachentity is associated with a state machine. State machines react toeach other and make decisions, for instance regarding their motion.The scenario control subsystem can simulate over 40 vehicles at anexecution frequency of 60 Hz, on a40Mhzi860 processor.9. Simulated ManeuversVirtually all, including limit-handling conditions.1.2.4 The GENERAL MOTOR SimulatorThe GENERAL MOTOR simulator is an automotive simulator developed atGeneral Motors [GB+94]. It possesses the following characteristics:1. Kinematics6 DOFs (lateral, longitudinal, vertical, yaw, roll, pitchmovements) arerealised by means of a Hexapod platform. Range of motion: Lateral,Longitudinal, Vertical motion: 0:18 m;Yaw, Pitch, Roll motion: 7Æ.2. Dynamics Mathematical ModelFour DOFs model: equations written for the yaw, roll, lateral andlongitudinal directions. Models are supplied for: engine, transmission,driveline, brakes, rolling and aerodynamic resistance.3. Control DevicesSteering wheel, accelerator, brake; active force feedback only on steer-ing wheel.4. ActuatorsNo info are available for the actuator of the hexapod platform. DCtorque motor is used for force feedback on the steering wheel; 24 Nmmax. steering torque can be developed.5. SensorsNo data available. 9
1.2. State of the Art Analysis 1. INTRODUCTION6. Real-time ComputingEquation of motions are solved in real time using a two-step Eulermethod on a VAX 4500. Integration time steps of 10 ms.7. Feedbacks ProvideVisual: Images projected on xed screen; resulting angle of vision 140Æwide, 40Æ high with a resolution of 8:5minutes of arc and 60 Hz non in-terlaced display rate. Sound: Sounds are generated by a bank of soundsamples and multichannel synthesizers and passed through lters, dig-ital delay units, mixers, ampliers, and speakers.8. Virtual RealityAverage scene consists of a two- or four-lane highway with sparse signsor vegetation and up to three other moving vehicles.9. Simulated ManeuversSpecial maneuvers that can be simulated: varying radius curves. Single-and double-lane change. Test truck evasive maneuvers.1.2.5 The JARI SimulatorThe Japan Automotive Research Institute (JARI) simulator is an automotivesimulator developed at Tsukuba, Japan [HS+96]. It possesses the followingcharacteristics:1. Kinematics6 DOFs (lateral, longitudinal, vertical, yaw, roll, pitchmovements) arerealised by means of a Stewart Platform. The Range of motion arereported in table 1.2.DOFs Disp. Value Vel. Value Acc. Valuelateral 0:4 m 1:0 m=s 5:0 m=s2longitudinal 0:4 m 1:0 m=s 5:0 m=s2vertical 0:4 m 1:0 m=s 5:0 m=s2yaw 0:4 rad 0:8 rad=s 0:7 rad=s2roll 0:4 rad 0:8 rad=s 0:7 rad=s2pitch 0:4 rad 0:8 rad=s 0:7 rad=s2Table 1.2: JARI simulator workspace
2. Dynamics Mathematical ModelNo data available. 10
1. INTRODUCTION 1.2. State of the Art Analysis3. Control DevicesSteering wheel, accelerator, brake; active force feedback only on steer-ing wheel.4. ActuatorsThe Stewart Platform is moved by means of 6 Hydraulic Actuators. DCtorque motor is used for force feedback on the steering wheel: 10Nmtorque.5. SensorsNo data available.6. Real-time ComputingEquation of motions are solved in real time using a Unix System with2xR3000@33MHz and 2xR3010@33MHz processors. Integration timesteps of 10 ms.7. Feedbacks ProvideVisual: Images projected on moved screen: horizontal 50Æ and vertical35Æ front of view; 2000 polygons and resolution of 1M pixels and 60 Hzrefresh rate. Sound: Sounds are generated by a bank of digital sounds(sampler S1000HD), managed byaPC9801DA computer with midiextension. 2 front, 2 rear speakers and a Super Woofer.8. Virtual RealityLane changing environment, StraightHighway environment.9. Simulated ManeuversVirtually can be simulated all typical car maneuvers.1.2.6 The NASA VMS SimulatorThe Vertical Motion Simulator (VMS) is a 
ight simulator developed atNASA [Cen]. It possesses the following characteristics:1. Kinematics6 DOFs (lateral, longitudinal, vertical, yaw, roll, pitchmovements) arerealised with mixed serial/parallel mechanism. The lateral, longitudi-nal, vertical and yawmovement are replicated using a serial mechanismwhile the roll and pitch are replicated using a parallel one. The Rangeof motion are reported in table 1.3.2. Dynamics Mathematical ModelNo data available. 11
1.2. State of the Art Analysis 1. INTRODUCTION
Figure 1.3: VMS Simulator3. Control DevicesComplete aircraft interchangeable cockpit with force feedbackjoystick.4. ActuatorsElectrical DC motors equipped with gearshift. The motor of the ver-tical displacement possesses 150 Hp @ 1350 RPM, the lateral motorpossesses 40 Hp @ 3500 RPM. No informations available for the otherDOFs.5. SensorsTachometers solidal with the electric motors.6. Real-time ComputingEquation of motions are solved in real time using a VAX 9000. Inte-gration time steps of 10 ms.7. Feedbacks ProvideVisual: Images projected on moved screen; Evans and SutherlandCT5A visual scene generator. Sound: no data available.12