They are interested in the following bicycle aspects for safety and maneuvrability: small wheels, extreme handlebars.

18 boys apporixmatelly 12 years old were tested on 3 different styles of bicycle (standard bicycle, small wheeled (collapsable bicycle) and rodeo style) with 2 different handlebars - standard and extreme (extra high and extra wide) (total of 6 bicycles). each test was carried out twice.

tests include:

looking backwards while riding : subject looks over shoulder to observe a combination of nuerals held up by the experimenter. once facing forward again the subject calls out the number he observed. measures were: correct number ballance assessed by an expert, deviation across sidelines, toutching the ground with feet.

slow riding between two lines: ride as slowly as possible between the two lines: time was measured, balance assessed. errors: toutching the ground with a foot, toutching sidelines with wheels. two seconds were counted for each error.

riding between wooden block pairs: time measured, balance assessed. errors: toutching a foot to the ground, or omitting a block pair.

Block slalom: time measured, balance recorded, errors: touching the ground or ommitting a block

riding balance: subject rides the length of a curved course with only the left hand on the handlebars then turn around while shifting hands and ride back in opposite direction with only the right hand on handlebars. time measured balance recorded. errors: touching ground, toutching sideline.

relay riding: subject required to move balls (similar to tennis balls) from one cone to another with one hand. the ball on the first cone was moved to the second, ball on the third cone was moved to the fourth, ball on the fith cone was transfered to th opposite hand before being placed on the last cone. time measured. balance recorded. errors: touching the ground, dropping the ball

riding in a circle: ride into a circle (diameter 3m) ride round 3 times and then back out again through the same point. time measured. balance recorded. errors: foot to ground, touching circle edge

stationary balance: subject rides to a small circle stops there such that front wheel remains in the circle and keeps balance for as long as possible. measure: time, balance.

Riding through narrow gates: subject had to ride through 2 narrow gates spaced 3 m appart. time was measured, balance recorded, errors: touching the ground or the gates.

acceleration test: accelerate as quickly as possible while riding between two lines. rider had to remain seated during the test. measure: time, balance. Errors: rising from seat, touching sidelines, foot to ground.

interviews held with subjects on rating the 6 bicycle based on safety

each error = extra time (deducted) to total time for experiment.

A pre-test was held with 12 year old boys from another school to work out aproximate measures for tests and to train experimenters.

all subjects allowed one practice run.

Results were analysed for variance on total time.

Conclusions: bicycles with extreme handlebars have poorer maneuvrability performance than those with standard handlebars. Rodeo style bicycles has the worst maneuvrability performance out of the three tested models.

21 refferences ongst which, Herfkens 1949 and Stassen and van Lunteren 1970 and Godthelp 1974.

Paper written in a nice style, Developes a human controler with a neural network, with only 2 neurons. A proportional controller on the heading and a pd controller on the lean (steer into the fall). The heading has some preview.

They define Handling as: easy to drive

They define Manoeuvrability as: able to perform complex manoeuvres fast.

Handling and manoevrability are measured by a penalty function which is the integral of time of the individual scaled penalties. they define an optimal control problem. The trajectory inequality constraints are added as weighted penalties (the tracking problem).

They do not solve the minimum time problem but fix the time and optimise the distance travelled.

As an example they study 2 slightly different motorcycles together with experimental data. the experiment fits the prediction but with a constant offset. the socalled improvement between the 2 motorcycles is open for debate.

Chapter one:

literature study

Chapter 2

HE first tries to identify a suitable tire model for the complete forward speed range. Clearly different tire models give different weave and capsize speeds.

Chapter 3

open loop no handsexperiments. he compares results with analog computer simulations. The disturbance is a steering impulse. good agreement is found between the simulations and results at 40 mph and moderate at speeds below this.

Chapter 4.

the man machine system is analysed to determine what constraints mus be placed upon the riders transfer function in order to have good system characteristics. For now we don't understand the analysis.....

Chapter 5.

The experiments are described (For chapter 4). No external excitation is applied only the riders remnant torques are applied. The task to be carried out was to stay upright and on the road (no real tracking). Measures steertorque with a single handed steertorque bar as an extention of the steering assembly. awkwared single handed steering (right hand steering) and speed control (throttle) on the rear frame.

Chapter 6.

The two transfer functions - motorcycle and rider control - are estimated with cross spectral analysis. which we don't understand yet..... The outcomes are bode diagrams with theoretical gains and phase and experimental gains and phase together with confidence levels... The rider delay is estimated? ( 0.3s )

He estimates the remnant rider torque contribution.

chapter 7

conclusions: main conclusions for the motorcycle model is that the model predicted a stable speed range around 15mph - the experiments did not confirm this. fidelling with tire models gives no significant improvement to the model results.

The main conclusions on the rider model we dont understand at the moment...

simple bicycle model, only a point mass on the rear frame, vertical steering axis, no trail (no tires, no slip) and on skates. Rear wheel is the x/y coordinate. He controles the steering variable ( which parameterises the steering angle (. he wants to let the rear wheel track a trajectory. for the control he uses his dynamic inverter stuff (which we don't understand). he does do path tracking with balance,

The examples he does are : straight path constant speed. Sinusoidal path (speed is controlled). Circle at constant speed. Figure 8 trajectory. All examples start with an offset from the track and all show counter steering effects.

we don't understand most of the work!

To remain on track, the following criteria are important: bicycle charachteristics, rider charachteristics, the course, surrounding hinderence.

They mention Sharp1971 and Roland1974 for the motion of a bicycle

mention van lunteren(1970) weir eaton and Roland for rider control

They note that Many accidents occur as a result of loss of bicycle control - due to weight changes (extra person on the back) or due to traffic rules (sticking hand out to indicate you are going to turn). they not that the rider can anticipate these changes most of the time however side wind cannot be anticipated.

experiments:

tracking a straight road: 0.15m wide straight road had to be cycled allong 3 times -quickly as possible - at normal speed, and as slowly as possible with one or 2 hands on the handlebars. and with and without side wind. (wind machines used)

keeping course in a bend: track exists of a sloping straight road followed directly by a sharp bend to the left and was indicated by 2 white lines 0.15m appart. complete the course at high speed and normal speed with one or 2 hands on the handlebars.

offset slalom: carry out as quickly as possible, with one or 2 hands on the handlebars.

3 types of bikes used: an instrumented bicycle where charachateristics could be changed (such as trail moment of inertia of front wheel, distance between sadle and handlebars (rearwards placement of large mass).

popular bicycles: collapsable bike, conventional male bike, race bike, conventional womens bike

popular mopeds.

The Human controller is modelled by upper body lean relative to the skateboard in order to stabilize the board and do tracking tasks. Instead of of doing torque control the dynamics of the skateboard itself are neglected and it is assumed that the lean angle of the skateboard is the control input.

The control task is stabilising and tracking the heading. It is a full state feedback control where the controller is designed by means of an LQR. Analytical results are compared to some experiments and show qualitiative agreement in the time series.

They report ongoing research in system identification to determine the exact feedback gains. (where is it?)

12 segment rider is added to the motorcycle model. some educated and uneducated guesses are made for the mass and mass moments of inertias and passive joint stiffness and damping. the only result shown is 1 figure of posture effect on riding in a circle (in line, lean in, shift lean in, lean out) and shows roll angle and yaw rate.

He gives 4 principle sollutions for the connection between the trailor and the motorcycle. they are all based on the specific location where the trailor is connected to the motorcycle (infront or behind the rear wheel, high up or low down).

From the analytical model the expected resulatant of the trailor on the motorcycle dynamics are:

He predicts that the position of the articluated joint of the trailor is very important for the lateral dynamics - in particular the weave instability.

In one case it pushed up the weave speed above the 20 m/s at low weave frequency (1 Hz)

In General it was predicted that the trailor would have an affect similar to adding a load rearward and high on the motorcycle.

Experiments:

instrumentation is identical to JAMES2002.

Typical testrun was running at a constant speed and wiggle the handlebars.

System Identification with black box ARX.

Various trailor loads used (0, 80, 140, 200kg)

Weave mode is clearly identified. There is no good comparison between the experiments and the analytical Wobble mode.

Finally he measures steer torque in steady turning - again with 0, 80, 140, 200kg at various forward speeds. The analytical model shows no dependancy on the trailor payload! and predicts a steertorque inversion at 12m/s whilst the experimental model shows an clear dependancy on the pay load and the torque inversion varies from 4.5 to 6.5 m/s.

He ends with some design guidelines for a single wheeled trailor.

The motorcycle is an off-road Honda XL250. Measured quantities are: forwards speed, steertorque, yaw and lean rate, lateral acceleration, steer angle.

The first set of tests looks at the hands free lateral dynamics of the motorcycle at constant speed. He applies a latteral pertubation to the handlebars, and the speeds are: 2, 4, 8, 12, 16 and 20 m/s and are carried out on both bitumin and dirt road surfaces.

He uses Ljung's black box ARX model structure for the motorcycle. Matlab SI toolbox was used.

He predicts the weave eigenvalue with a fair agreement to the experimental results. He also visiualizes the weave mode in a phasor (polar plot).

The identified wobble frequency does not comply with the analitical one for all forward speeds. The trends look the same but there is both a frequency shift and a damping shift.

He also measured steer torque in steady turn, compared with ZellnerWeir1978 and found his motorcycle the torque inversion around 9m/s. He does not make the connection with the capsize speed.

He makes bode plots for steer torque input to yaw rate output at various forward speeds. He compares these with the annalitical model. At high speeds the wobble frequency (mode) is identified but the weave cannot be clearly identified from the meauserments.

Unfortunatelly the discussion is very short..

The literature servey was carried out to determine what jurisdiction should be enforced by government on bicycle aspects.

Chapter 2:

the report starts with a historical review of the bicycle. Then the report looks at the following components (with respect to safety)

:brakes,

lights

stability

frame and front fork

drive train

tires.

Chapter 3:

traffic accident numbers are recalled

Chapter 4:

literature review, including Rice&Roland(1970), Roland et al, 1979, Whit and Wilson (1982 ) (bicycle science)

Chapter 5:

brakes: report insists there should be a minimum required average braking deceleration by law. The german 4m/s^2 law is not achieved by the majority of bicycle! a deceleration of 6 - 8 m/s^2 is enough to flip over!

lighting: conclusion is that the total falure of the lighting requirements by almost the complete bicycle population is catastrophic for the safety!

stability: Reports GodhelpWouters1978 and wouters(1980), and gives design parameters for stable bicycles. - according to deLong1974 - and mentions NORM ISO 4210 and its implications. Gyroscopic effect is recalled as the stabilizing effect for bicycles.

braking and handling qualities of childerens and elderly people bicycles: extra attention is payed to these 2 groups due to their large amount of trafic accidents with injuries: primarily childeren have been looked at till now - elderly has always been a small group - but is growing rapidly now.. They report the Abraham bike (low sadel, low tubing etc) and a tricycle as a solution for this group. for the childeren they report studies by wierda&roos and Wierda&wolf,

frame and voorvork: ISO norm required.

Chapter 6:

catagorisation of bicycles: 6 catagories: a: full size bicycles, B: childerens bicycles. 1:normal use, 2. heavy use (transport bicycles) 3. all terain use. however it is noted that there are sirious consequences to giving different catagories different safety requirements.

chapter 8: conclusions/discussion

it is noted that it is important that elderly remain mobile and keep cycling. It is said that elderly should use bicycles that compensate for their problems and that such bicycles already exist. They end the discussion by asking if it is pointless to impose seperate saftey requirements for this group of biycles.

The control output is a steering torque, upper body lean torque and lower body torque. the input is the lean angle (motorcycle lean angle) and the average heading error. In this case case the average heading error is taken as the lateral seperation of the desired path and the straight line predicted motorycle path (not curve!!!!), weighted around a preview point.

all experiments are carrie ont on a single lane change. The single lane change maneuver is a 3.6 lateral over 30m longitudenal opperation. All tests are carried out at 60kph. 1 medium sized motorcycle and 12 experienced riders are used. The controler is a simple P controler with constant gains.

With theoretical studies on this lane change maneuver he looks at various control sinarios, and concludes that steering torque is dominant, lower body torque assists and that upper body torque is such that the upper body is kept vertical and does not really contribute to the control but is used for comfort of the rider.

They build a very nice device for measuring the rider body movements. They are able to measure the lean angle, yaw angle and pitching angle of the upper body and the lateral motion of the lower body of the rider.

They are able to find gains, preview points, passive rider stiffness and damping, such that they can fit the simulation time series on the experimental time series for every individual rider. They clame that the time series on the rider lean angles confirm the hypothesis.

First some history on bicycles and motorcycles is given. Then a history on modeling and control is given. Box item on Whipple. Discussion on Getz and Marsden point mass model and control. Discusses Whipple model and show the bench mark bicycle with slightly different parameters. Discusses special cases like locked steering, no head angle, no trail, gyroscopic effects on the dynamic behaviour of an uncontrolled bicycle. Then they take the whipple model and recast it into a feedback system (there is no control) in order to explain self stability and steering behaviour (steer to lean coupling) with the help of transfer functions. Next they add tires and frame flexibility and discuss the effect on the dynamic behaviour (wobble). Box item on tire modeling. Box item on wobble. Next they step from bicycle to the motorcycle modeling and explain the simularities and differences.Box item on Tommy Smith wobble (high speed on utah salt flats). Discusses the following aspects on the dynamics of straight running motorcycles: tire modeling, aerodynamic forces, structural flexibilty and rider modeling. Next they look at steady cornering of a motorcycle with suspension. They draw some conclusions.

skidding, and turning over = the stability problem for cars!

he applies concepts of system control to make statements on manoeuvrability and stability: controlability, observability and stability are investigated.

Unfortunately there are no physical numbers to go with the described work.

Only stabilizing task on a straight course is considered. at a moderate speed - 10 to 20 kph. They hypothesize that human control is carried out by a PID controller with a delay.

The simulator ties to mimmic a bicycle running straight at a given speed (15kph). The simulator is stationary, and pivots about a motor controlled horizontal-longitudenal axis. A second motor is used to mimmic front wheel gyroscopic effects.

This is an approximation of a bicycle as the contact points dont move.

They measure the bicycle and rider lean and the steer angle and carryout all the experiments at 15kph.

They conclude that the human bicycle controller can be described by a pd controler with a time delay. Where the input is the frame lean angle and the outputs are the steer angle and upper body lean angle. Average time delay on the handlebar control is 150ms and on the upper body control is 100ms. NO RIDER LEAN OR STEER TORQUE!

They use the polarity coincidence correlation method (PCC) (from chapter II) to identify the human controller. To identify the coefficients of the Delayed PID controller. the characteristics of the riders inputs and outputs.

They finish with future work: they wish to add a visual display unit to carryout cours followng (tracking) control.

Full non linear tire model is incorporated into the model.

the rider lean motion is a seperate control system with inputs lean angle and desired lean angle and output rider lean. The rider lean may or may not stabilize the bicycle.

The steertorque controller has input the lean angle and desired lean angle and output the steer torque. The transfer functions are PID controlers with lead filters and a velocity dependant gain. The coeficients are determined with the concept of H infinite performance index. Two examples are shown. The steer torque disturbance and a cornering manauevre. A pre-filter is used to get realistic steer angles.

3 handlebars: racing, standard and highrise

6 tasks:

Circle : time

figure eight : time

lane change: time

straight lane tracking fixed low speed: number of boundary crossovers (errors)

cornering fixed speed: radius

slalom fixed speed: number of boundary crossovers and cones nocked over

Slalom: maximum speed

two groups of riders used: standard and race.

rider rating of the bicycle and a rating of the task - on a 5 point scale.

conclusions: The race handlebars make the bicycle least maneuverable and a high rise is ok.

tests carried out with 2 bicycles : high riser and light weight conventional bicycle.

brake test include:

Stopping distance versus speed.

coaster versus caliper brakes.

affect of rider weight on stopping distance.

lateral stability and control tests include:

minimum speed for handsfree straight path following

timed slalom test

The report ends with a large appendix on the equations of motion for a computer model of the dynamics an uncontrolled bicycle. The model includes tire models.

model is not used due to limmited time and funding.

The main objective of these reports is to study the effect of design parameters on bicycle stability and control.

This report has 4 distinct parts:

chapter 2 : a rider control model for stabilising and tracking.

chapter 3 : bicycle tire testing

chapter 4 : experimental testing of various bicycle configurations.

chapter 5 : computer graphic animation of a bicycle simulation.

Rider control model:

The rider lean torque and steer torque are outputs and they control the lean angle with a delayed PID controler on the lean angle for both torques. The delayed PID controler is a simplification of a delayed lead-lag controler from literature:

Elkind, J.I. 1956 "Characteristics of simple manual control systems" MIT, Lincoln laboratory Techinical Report No. 111. They are aware of the work of Stassen and van lunteren and clearly identify the difference in control strategy : steer agle versus steer torque.

The tracking control is calculates from the states and the desired path a command rol angle. They tune the coeficients of the stabilising controler by looking at systems response driving straight ahead or applying a 20 degree comand roll angle like driving straight and getting into a curve. Even for the best controler we see an offset between the desired and obtained lean angle. The tracking controler makes a prediction of the path based on the state and compares this with the desired path and generates in adition to the desired lean angle a lean angle.

To our knowledge this controller is never used.

Bicycle Tire testing:

11 types of tires are tested. results are shown in graphs and shown in tables in the appendix. The general idea is that bicycle tires should have a camber thrust factor of about 1 or in other words: the tire force should always be aproximately in the plane of the wheel. The presented results are suspicious on this.

Experimental testing of Bicycle:

They use the instrumented bicycle of Rolland & Massing 1971. They have 9 configurations by playing around with load on rear, rider and front, increasing the mass moment of inertia of the front wheel and underinflating the tires.

the tests carried out are:

lows peed stability: riding in a 3 feet wide lane at minimum speed

obstical avoidance: at the end of the 3 feet lane a dustbin is placed (4 feet from the end), at maximum speed

narrow slalom: inline 10 feet apart, maximum speed.

wide slalom: 2 feet lateral separation, 10 feet apart, maximum speed.

They conclude that the standard bicycle the best! the other general conclusions are that load in the rear basket (not on rear rack) is good for maneurverability and rider load is bad for maneurverability.

Computer Graphics Animation of the Bicycle Simulation:

A large part of the report is devoted to generating computer graphical animation of the biycle simulations. It was very state of the art! the climax is a camparison of an experimental and simulated bicycle slalom maneuver based on a strip chart of 6 images 0.2 seconds apart! the resemblance is striking! the simulated results were obtained by applying the following "bang - bang" control: the sign of the desired lean angle is opposite the sign of the current steer angle. lucky strike??

Impressive results. unfortunately they do not elaborate on the rider control and validation but this is probably due to a lack of time and funding.

Report starts with a mathematical model of a bicycle with a fixed rider and tire models. They present the matrix equations of motion condensed on 1 page!

Then the input / output of a computer program is described which implemented the equations. The program is able to do time series simulations.

Next the physical characteristics of a bicycle and rider are measured. The bicycle was a 22inch single speed Schwinn Suburban.

Mass moments of inertia are measured with torsional springs. The CG's of the rear and front frame are clearly identiefied on the bicycle by markers.

A bicycle tire tester was built. It is a monowheel trailer behind a Ford Mustang. Normal load, slip angle and inclination angle could be adjusted. The trailor is loaded with weights and the tire side forces are meauserd by strain gauges on the structure.

Two types of tires are measured: Puff High Pressure Road Racer, and Breeze Sports Touring.

The results are nicely shown in 6 graphs of side force versus slip angle for a range of parameters.

Physical characteristics of the rigid rider are calculated from a table of individual parts (head, torso, arms, legs and feet). (in slugs)

Bicycle is instrumented, forward speed is measured with a generator like DC tacho, Roll angle Gyro and Geared steer angle potentiometer (1:8). lateral accelerometer on the rear frame. Data is recorded on a two channel strip recorder mounted in the chase car. Sensors are connected by wires held up by a cable boom from the car.

A number of experiments are preformed.

10 mph no pertubation.

10mph lateral rocket impulse on the rear frame

9mph steer torque impulse by rocket

suspended bicycle with spinning wheels had steer torque impulse by rocket applied, to observe gyroscopic coupling between steer and roll. Visual recording only. Conclusion: rotating front wheel has little effect on the resulting steer and lean motion

They calibrate the rocket.

A comparison is made between a computer simulation and full scale experimental tests. They find a fair agreement in the time series solutions.

Finally a bicycle design parameter computer study is made where the effect of the change of 9 bicycle parameters was studdied on the motion for three different speeds: 7, 9 and 15 mph

parameters are: Wheelbase, Total weight, Total c of G height, Steer mass moment of inertia, Caster angle, Fork offset, Undeflected rolling radius, wheel moment of inertia, Puff tires and 102 pound rider. The results are presented in a nice table, and discussed.

It is a nice and comprehensive paper! there is a large appendix with the equations of motion for the bicycle model (including a 1 page matrix equations of motion), tire model, and rider control model. all very nicely layed out!

The effect of road irregularities on the stability of a motorcycle is investigated. 2 types of irregulararities are looked at: Cats eyes, and Skew Drainage grooves. Tire forces and moments are measured for traversing a cats eye and skew drainage grooves.

The motorcycle is equiped with a rider robot for safety reasons!

Motorcycle is a BMWR60 / 6, Remote controlled outriggers are added (and fueltank, seat and buddy seat removed)

Control scheme is a velocity dependant feedback of the lean angle, lean rate, and steer rate to the steer torque.

Steer Rate: constant gain.

Lean Rate: high at low speed, crosses zero around the capsize speed, and linearly decreases with increasing forward speed.

Lean Angle: High at low speed, zero around the weave speed an linearly increasing with increasing forward speed.

Motorcycle is stable from about 5m/s till 60 m/s.

Measured the roll angle in two manners: 1 with trailing arms, and seccondly with an accellerometer. Both were succesful.

Tests proved that motorcycle is stable from 10 to 110 kph.

(still looking for Dutch report by Ruijs 1981 (Ref 9) - check local library)

He uses the benchmark bicycle for which he again presents the linearised equations of motion but yet with slightly addapted bicycle design parameters (he lumps some parts together). he applies optimal linear preview control to the bicycle, which is infact a LQR controller with preview (Full state feedback). Studies different weight factors for tracking errors against control power. The feedback gains are clearly speed dependant and become unrealistically high. tracking is carried out on the basis of the lateral separation of the front wheel contact point and the ideal path.

For the path following simulations he looks at a random road, straight line into a circular path (90 degrees) and then straight agian,

conclusions:

Tight control requires about 2.5s preview. the necessary preview time depends little on speed, so that preview distance are roughly proportional to speed.

With his autosim motorcycle dynamics model (tires, suspension, flex arm, and lateral lower body and pivoting rider upperbody ) he calculates at various speeds eigen frequencies. He shows a number of eigenmodes (phasors) at 1 speed (20m/s). next he shows a bode plot of yaw and lateral displacement to steer torque input at 30 m/s.He touches cornering behaviour by looking for 1 angle and at various speeds at the equilibrium positions and shows the root locus and some eigen modes at 1 speed (phasors).

This is a literature survey on bicycle and motorcycle dynamics. some attention is given to handling and control . Most is on uncontrolled dynamics.

two test riders are used: an expert and a novice.

Theoretical results (for comparison) are from an open loop, hands free model.

4 configurations are tested:

rider alone

rider + windscreen

rider + rear load

rider+ passanger

They focus on steady turning at various speeds and radii where they try to generate a limit cycle which is most visible in the yaw rate or a slightly divergant cycle.

For upright tests, osscillations are induced by steer torque pulse for the various speeds and configurations.

Some conclusions for the various configurations on opperations are drawn. They try to explain the cornering weave phenomena from the weave eigenfrequency an the pitch eigenfrequency (in the upright configuration), doubtfull.

Chapter 1

Review of rider control is given.

Chapter 2

Devides the opperator in three parts: compensate, persuit, Precognitive (nothing is done with the precognitive).

identifies delay in the human controler and includes it in the transferfunctions (approximate).

A dynamic model for the motorcycle including tires is derived. He also draws block diagrams!

First he discusses the control statagies steer angle versus steer torque. Next he considers a number of input - output feedback stratagies where he presents the results in the form of transfer functions and bode plots:

yaw rate to steer torque

yaw rate to rider lean

lean angle to steer torque

lean angle to rider lean

steer angle to steer torque

steer angle to rider lean

lateral velocity to steer torque or rider lean

lateral acceleration to steer torque

lateral position to steer torque (lateral position of the rear frame)

A review of single loop control is given.

lean anlge to steer torque is the best followed by lean angle to rider lean and yaw rate to rider lean the rest sucks...

multiple-loop control systems: multiple input with multiple outputs are discussed. Inputs based on the above states. outputs are now steer torque and rider lean. he draws some conclusions which we don't understand yet.....

Chapter 3

For preferred handling he argues that it is sufficient at open loop characteristics. then he looks at the effect of changing the forward speed and some design parameters on the capsize, weave and wobble eigenfrequency in terms of small - large, increase - decrease.

Based on the open loop bode plots at various forward speeds he discusses the implications on the rider control gain.

Next he compares a standard motorcycle to a chopper and its implications on handling and control.

For good handling properties he wants a small positive capsize eigenvalue. he tunes this analitically by making changes to the trail.

Chapter 4

ends with a discussion and conclusions

There are 41 references and the derivation of the equations of motion for a motorcycle are in the appendix.

The complete Motorcycle model based on Weir Ph.D thesis 1972 are shown in the appendix. linear tire models are used.

Rider control: the control loop is a single loop feedback of the lean angle to a steer torque following Weir 1973 (paper SAE 73018) and confirmed by the experiments carried out by Eaton Ph.D 1973.

Lateral control and tracking is acchieved by multiple loop feedback control. Feedback of rearframe lean angle, yaw angle and the rear frame lateral displacement to a steer torque and rider lean. No experimental validation given.

Implications for design and handling are discussed in terms of eigenvalues of the open and closed loop system.

Finally he theoretically optimises the trail with respect to the eigenfrequency of the capsize mode.

bicycle safety experiments based on trafic situations and carried out on at the time currently available childerens bicycles. different bicycles for boys and girls are used.

children between the ages of 8 and 12 were tested on different bicycles at normal cycling speeds. tests included different types of straight ahead cycling and braking with and without extra added mass. all bicycles were fitted with a reverse pedel brake.

First three experiments were carried out inside in a heated hall. riders had to try to stay in the 40cm wide 17m long straight track. the experiments were carried out 3 times tests were:

A: riding in a straight line with right hand on steeringwheel and left hand pointing outwards (to indicate going left)

B: same as A, but also looking rearwards over left shoulder. where the child had to count the number of fingures that the examiner held up and shout this out loud.

C. Braking in a straight line. a horn sounded when the cyclist had his/her feet in the horizontal position. from that moment on they had to stop as quickly as possible, but with 1 or 2 feet touching the ground when they had no speed anymore, and on top of this the feet had to be in the 40cm wide track. (simulate stopping in traffic between cars).

There are substantial differences between the different bicycle styles and also between the boys and girls bicycles. some of the boys bicycles are much less "safe" than the girls.

D: brake test outside: ride 13kph, stop as quickly as possible, no instructions on the placement of feet. sometimes the bicycle has been fitted with extra mass to represent school usage. each child rode 3 different types of bicycle, and carried out the experiment 3 times with each bicycle. for the bicycles with added weight the another 9 tests were carried out. all bicycles give similar results for no extra added weight. There is one bicycle that stops much better than all the rest: this is a boys bike that is fitted with front and rear hand brakes (no rear pedel brake).

For the experiments carried out with extra weight (35kg for rider plus weight for bicycle sizes upto 46cm and 45kg for rider plus weight for bicycle sizes larger than 46cm). no real conclusions could be made as a the added weight was to much for the children and they became far more cautious!

Chapter 1

introduction

Chapter 2.

discussion of the main charachteristics of the 6 different bicycles (3 for boys, 3 for girls) and a discussion on the chosen test group. (group ages range from 9 to 13.

chapter 3.

emperical research to types of bicycles

first the ergonomics of bicycles is investigated and it is concluded that anthrapometric tables are of no use for kids in this age group as there is no correlation between body lengths and wieghts etc. therefore kids should try a number of bicycles and sizes before buying a bicycle.

16 boys and 16 girls are used as test persons.

Experiments: experiments were carried out 4 times by each test guinnypig.

a: riding in a straight line, right hand on handlebars, left hand extended to the left (to indicate going left) and looking over left shoulder and upon request recount the number of fingers shown by the experimenter. it is concluded that kids of this age are not very profficient in safely opperating in traffic in this manner. (they do not keep to the lane enough)

b: evasive maneuver and emergency stop. the riders ride at normal cruizing speed in a central lane (40cm wide). when the cyclist has his feet horzontal, a horn is sounded and a traffic light on one of either side of the lane is lit. the rider then makes an evasive maneuver to this lane (also 40cm wide) and stops as quickly as possible - and places at least 1 foot on the ground (preferebly in the lane). no significant differences are noted between the different types of bicycle.

c: outside brake test. riding 13kph the child has to stop in a 50cm wide lane as quickly as possible. and put at least 1 foot on the ground (in the lane) main difference between girls and boys appear to be a much slower reaction time for girls using a revers pedel brake. double brake bikes stop quicker than singel (rear) brake bikes.

4 different mopeds and 1 motorcycle are considered. he looks at the difference in geometrical properties and mass distribution and the characteristic roots at various speeds. he looks at bode plots of the lean angle to steer torque and lean angle to rider lean for a moped and a motorcycle at 30mph. we don't understand these diagrams yet. his conclusion is however that: the moped is more sensitive than the motorcyle for steer torque control and the rider lean input is for motorcyle and moped the same suprisingly since the motorcycle is so much heavier.

He draws some general implications on control and handling for mopeds. That they are more sensetive to steer torque control. the moped has a less damped wobbel compared to the motorcycle, however the rider can probably damp this mode of the front assembly as it has much less inertia.

He gives suggestions in terms of "increase" and "decrease" of design parameters for the three modes (weave, capsize and wobble), all based on analytical results.

2 different tests: single lane change, and steadyturning. 4 different motorcycles: light, medium, medium, heavy, heavy. Instrumented motorcycle with the usual sensors including steer torque. rider lean and rider pitch by potentiometer, rider lateral position by movie camera.

Steady turn tests: at various forwards speeds and various radii are carried out. the measured steer torque devided by roll angle is plotted for different speeds and compared to the results from the analytical model (for the different bicycles) torque reversal at capsize speed is well predicted. Data is a bit scatterd. Yaw rate over steering angle is presented for the various vehicles showing totally different behaviour - for both the experimental and analitical results. Steer torque over steer angle at various forward speeds is shows two critical speeds the capsize speed and a steer angle imput crittical speed. which is probably due to tyre mechanics as this is not present in the bicycle model.

Single lane change maneuver: criteria for this experiment are: various performance measures are discussed. they use the deviation from a ramp plus sine ideal pathline which we find pretty awkward. The RMS of this performance index is scatterd all over the place.

future work is predicted.

He finds most influential for stability are: trail and mass moment of inertia of the front wheel. less significiant is rider position.

Feedback gains were determined for the rider lean torque in order to stabolise the bicycle. making the bicycle no-hands stable.

Experimental setup is built. where the leaned rider is an inverted pendulum. The bicycle is driven by a gasoline engine to maintain constant speed and a geared electric motor and control systems are added to control the inverted pendulum. The feedback gains from the study are implamented.

He couldn't get the system to opperate (no stable motion).

control is a linear feedback of the frame and body lean and their rates to the body lean torque where the gains are determined by pole placement .

time responses to intitial conditions are calculated.

many hardware problems are cited as possible causes for the bad opperation. Most important conclusion was that he neglected the geared inertia of the electric motor in his calculation of the required gains. He also hypothesises that the front fork axis friction was to large.