CAPSTONE Final Report, Winter

8/8/2019 CAPSTONE Final Report, Winter

1/111

FACULTY OF ENGINEERING AND APPLIED SCIENCE

Active Trailer Steering for Multi-Trailer Articulated Vehicles

Prepared for CSME - Student Design Competition

Prepared ByPhemelo Bogosi, Nicholas Buttery, Phil Fracz,

James MacMillan, Jack Wentzell

Faculty Advisor: Dr. Yuping He, PEng

Submitted:April 30, 2010


2/111

Executive Summary The design team has taken eight months of research, development and testing andcreated an effective active steering controller to improve the low speedmaneuverability of multi-trailer articulated vehicles. We were presented with theproblem of improving the low speed maneuverability of multi-trailer vehicles asoutlined in the Ontario LCV (Long Combination Vehicle) Pilot Program and requiredby our industry sponsor, Genist Systems. Our active steering system is intended tocontrol the systems used to steer the wheels of the trailers. We are employing twosoftware packages to accomplish our goals: TRUCKSIM and LabView. The controllerwill be constructed in LabView using a six degree of freedom vehicle model tocontrol the rear steering angles of the trailers relative to the steering angle of thetruck. TRUCKSIM will be used as a simulator to test every iteration of our controller.

We are able to change and update any variable associated with the operation of amulti-trailer articulated vehicle. We have been able to create custom sensors within TRUCKSIM which are needed to gather the information necessary for our controllerto operate properly. We have taken our project one step further than initiallyintended by creating a 1/14 th scale remote control prototype of the multi-trailerarticulated vehicle which is used to validate our controller. We have taken the basicoperation and control of our final LabView controller and accurately representedusing C to employ in an on board microcontroller. The on board controller controlsthe rear trailer steering angle. The trailer steering is achieved through the carefulconstruction on purpose built hubs and couplers which combine at a central servovia precision cut connecting rods. We are able to control the servos on each wheelindividually in order to achieve the optimum turning radius for our articulatedvehicle. We have shown through the careful construction of our prototype and theextensive virtual testing of our controller that we have significantly improved thelow speed maneuverability of multi-trailer articulated vehicles. It is expected thatthe developed controller can be further validated in a real physical prototype, andthen it may be applied for improving the performance of multi-trailer vehicles.


3/111

Table of Contents65535

Table of FiguresFigure 1: LCV A-Train Double Description, Dimensions and Weights, 1 [16]............8

Figure 2: LCV A-Train Double Description, Dimensions and Weights, 2 [16]............9

Figure 3: LCV B-Train Double Description, Dimensions and Weights, 1 [16]..........11

Figure 4: LCV B-Train Double Description, Dimensions and Weights, 2 [16]..........12

Figure 5: US 3,894,773.............................................................................................20

Figure 6: Rollover Test, Double Trailer Combination [6]...........................................22

Figure 7: Self-Steering Trailer Axles [7]....................................................................23

Figure 8: Steering Method for Trailing Section [10]..................................................24

Figure 9: Path Following Strategy Diagram..............................................................32

Figure 10: Vehicles Required Path [25]....................................................................33

Figure 11: Kinematic Controller Model......................................................................35

Figure 12: Vehicle Paths - Simple MPC Control.........................................................37

Figure 13: Advanced Dynamic Controller Flow Chart...............................................39Figure 14: Trailer Steering Geometry.......................................................................40

Figure 15: LabView Initialation Diagram...................................................................41

Figure 16: Kinematic Controller Diagram..................................................................42

Figure 17: Dynamic Controller Diagram...................................................................43

Figure 18: A-Dolly vs. B-Dolly 1................................................................................44

Figure 19: A-Dolly vs. B-Dolly 2................................................................................45

Figure 20: A-Dolly vs. B-Dolly Lane Change Graph...................................................46

Figure 21: A-Dolly vs. B-Dolly Lateral Acceleration Graph........................................46Figure 22: A-Dolly vs. B-Dolly RWA Graph................................................................47

Figure 23: Lateral Acceleration and RWA, B-Dolly....................................................47

Figure 24: Lateral Acceleration and RWA, A-Dolly....................................................47

Figure 25: A-Dolly vs. B-Dolly Test Data...................................................................49

Figure 26: Lateral Acceleration vs. Speed (Loaded).................................................50


4/111

Figure 27: Lateral Acceleration vs. Speed (Unloaded)..............................................50

Figure 28: RWA.........................................................................................................51

Figure 29: Full 90 degree turn, No Steering Figure30: Full 90 degree turn, Steering..............................................................................52

Figure 29: Full 90 degree turn, No Steering Figure30: Full 90 degree turn, Steering..............................................................................52

Figure 31: Full 180 degree turn, No Steering Figure 32: Full 180degree turn, Steering...............................................................................................52



Figure 33: Full 360 degree turn, No Steering Figure 34: Full 360

degree turn, Steering...............................................................................................53Figure 35: Offset Values for Turns............................................................................53

Figure 36: 90 Degree Turn, No Steering, Plot...........................................................54

Figure 37: 90 Degree Turn, Steering, Plot................................................................54

Figure 38: 180 Degree Turn, No Steering, Plot.........................................................55

Figure 39: 180 Degree Turn, Steering, Plot..............................................................55

Figure 40: 360 Degree Turn, No Steering, Plot.........................................................56

Figure 41: 360 Degree Turn, Steering, Plot..............................................................56

Figure 42: B-Dolly Production...................................................................................58

Figure 43: Fifth Wheel Attachment...........................................................................59

Figure 44: Fifth Wheel Attachment Top and Side.....................................................59

Figure 45: Side View of Trailer Steering Assembly...................................................60

Figure 46: Top View of Trailer Steering Assembly....................................................61

Figure 47: Trailer Steering Hub Assembly with Tire..................................................61

Figure 48: Control Rod Mounting Plate.....................................................................62

Figure 49: Y Yoke Steering Concept......................................................................63

Figure 50X Yoke Steering Concept........................................................................64

Figure 51: Steering Control Linkage Assembly.........................................................65

Figure 52: Steering Hub Assembly............................................................................66

Figure 53: Full Left Articulation.................................................................................66

Figure 54: Full Right Articulation..............................................................................67


5/111

Figure 55: Assembly Drawing...................................................................................68

Figure 56: Exploded View + BOM.............................................................................68

Figure 57: Hub Mount Drawing.................................................................................69

Figure 58: Wheel Hub Drawing.................................................................................69

Figure 59: Control Rod Plate Drawing.......................................................................70

Figure 60: Wheel Shaft Drawing...............................................................................70

Figure 61: B-Dolly.....................................................................................................71

Figure 62: Trailer Steering Assembly Drawing..........................................................73

Figure 63: Assembly Isometric View.........................................................................73

Figure 64: Assembly Bottom View, Turned...............................................................73

Figure 65: Assembly Top View, Turned.....................................................................74

Figure 66: Assembly Side, Turned............................................................................74

Figure 67: Assembly Side, No Turn...........................................................................75

Figure 68: Assembly Top, No Dolly...........................................................................75

Figure 69: Assembly Isometric, No Dolly..................................................................76

Figure 70: Assembly Side, No Dolly..........................................................................76

Figure 71: Assembly Front, No Dolly.........................................................................77

Figure 72: Table of Physical Prototype Components.................................................78

Figure 73: Trucks Servos for Steering and Gear Shifting..........................................79

Figure 74: Trucks Three Speed Transmission...........................................................80Figure 75: HSC12DT256B Microcontroller [26].........................................................81

Figure 76: Rotary Encoder [27] Figure 77: Rotation Sensor[28]...........................................................................................................................81

Figure 76: Rotary Encoder [27] Figure 77: Rotation Sensor[28]...........................................................................................................................81

Figure 78: Multi-Axis Accelerometer [29] Figure 79: RC ServoMotor [30].................................................................................................................82

Figure 78: Multi-Axis Accelerometer [29] Figure 79: RC Servo

Motor [30].................................................................................................................82Figure 80: Rotation Sensor Mounting Configuration for Truck..................................82

Figure 81: Wheel Speed Encoder Mounting Configuration for Truck.........................83

Figure 82: Switch Mounting Locations......................................................................84

Figure 83: DB9 Wiring Connections..........................................................................84

Figure 84: Current Circuit Schematic Diagram.........................................................85


6/111

Figure 85: Controller Flow Chart...............................................................................88

Figure 86: Small Scale Prototype Offset Testing Data..............................................89

Figure 87: Active Trailer Steering Enabled Front View..............................................90

Figure 88: Active Trailer Steering Enabled Side View...............................................90

Figure 89: Active Trailer Steering Disabled Side View..............................................91

Figure 90: Active Trailer Steering Disabled Leading and Following Trailer...............91


7/111

1. Introduction and Statement of the Problem

Introduction

Methods of transporting goods from one place to another have been around for avery long time however it wasnt until the 1890s when the first gasoline poweredtrucks were produced. The first trucks were produced in the United States of America and slowly grew in other countries. People started realizing how usefultrucks were during World War I where they were used to transport food and suppliesto everyone both at home and overseas [11]. At this point in time there were veryfew manufacturers of trucks and things stayed this way until the 1920s whensystems of paved roads began to emerge. More and more companies started toproduce different types of trucks until the great depression in the 1930s [12]. Justbefore the great depression, there were approximately 300 different brands of trucks on the road. However most did not last long and eventually went out of business. By the 1990s there were only 9 manufacturers of heavy duty trucks leftin the United States of America and they produce about 200 000 trucks per year[12].

Heavy duty trucks consist of a separate tractor or cab and a trailer whereby thetractor is specifically designed the pull the trailer. The trailer connects to the tractorusing a mechanism called a fifth wheel. The fifth wheel allows the tractor tosecurely move the trailer in both the forward and backward directions and alsoarticulates or pivots to allow the trailer to turn at the mounting point. This is wherewe start to encounter some of the maneuverability issues which this project is going

to address. Due to the long lengths of the trailers and the single articulation point, alarge offset is introduced when the vehicle turns a corner. The offset occursbecause the tractor is hauling such a long trailer which is only articulated at onepoint which is at the very beginning of the trailer. Also the trailers have fixed wheelsand axles which do not allow the trailer to steer like a regular vehicle. This meansthe driver of the tractor must take wide turns in order to compensate for the traileroffset during cornering.

Due to the economic crisis which is currently happening and also the rising costs of fuel due to fossil fuel depletion, there is a need for a more efficient way of transporting goods from one area to another. Companies would like to have each

trailer carry more weight to reduce the number of trips to deliver the same numberof goods. Weight limits on roads and bridges do not allow this. However, if theextra weight is taken by extra trailers the vehicles are still able to drive on the sameroadways. This is why manufacturers are beginning to develop multi-trailer systems.Multi-trailer systems involve a single tractor pulling a series of interconnectedtrailers. The trailers are connected to each other by a device called a dolly. Thereare two main kinds of dollies used to connect the trailers: A-Dollies and B-Dollies. A-Dollies use a pintle hitch to connect the dolly to the leading trailer and then use a

1


8/111

fifth wheel to connect to the rearward trailer. A B-Dolly is essentially built-in andfixed to the leading trailer and then utilizes a fifth wheel to connect to the rearwardtrailer. There are advantages and disadvantages to each kind of dolly which shouldbe taken into account when deciding which kind should be used. The A-Dollyessentially creates double the number of articulation points which means the offset

during low speed maneuvering will be reduced. The disadvantage of the A-dolly isthat the high number of articulation points causes the system to become unstableat high speeds which creates a high risk of unstable motion modes such as roll over,trailer swing, jack knifing and snaking. The B-dolly is basically the exact opposite of the A dolly because it has fewer articulation points. This means better high speedstability but poor low speed maneuverability [13].

The high speed stability is based on one very important aspect in vehiclesdynamics, lateral acceleration. As a tractor/trailer system performs a high speedmaneuver, such as a lane change the entire vehicle undergoes a lateralacceleration which is in the opposite direction to that of the lane change. The driverwill only experience the lateral acceleration equal to that of the lateral accelerationfrom the tractor. The trailer will actually experience a greater lateral accelerationbecause of its length, articulation point and the fixed rigid axles. If there is morethan one trailer in the case of multi-trailer systems like the ones being investigatedin this project, each successive trailer will experience a greater lateral accelerationthan the one before it. This phenomenon is titled the Rearward Amplification. TheRearward Amplification Ratio is defined as the ratio of the lateral acceleration of therearmost trailer divided by the lateral acceleration of the tractor. This value isdesired to be as low as possible for optimal high speed stability [13].

Current regulations on the number of trailers are different for every province, stateand country. Places where longer tractor/trailer systems are permitted normallyrestrict them to specific roads or paths which have low traffic flow and these roadshave large radii of curvature. This is because these vehicles cannot maneuver onnormal roadways due to the off-tracking and stability issues described in the aboveparagraph. In Ontario, there are only a few numbers of major highways androadways that allow multi-trailer systems which are restricted to two trailers and anoverall length of 40 meters including the tractor. For these reasons, it is necessaryto research and develop some kind of system to allow the trailers to exactly followthe path of the lead tractor. This will theoretically improve or even entirely get rid of trailer offset during corners which result from the improvement of low speedmaneuverability. If this can be done, then the traffic regulations can be changed toallow longer tractor/trailer trains which will result in less fossil fuel usage and alsofewer number of tractor/trailers on the roads [13].

Problem Statement The active steering system we are designing will be an essential element in theemerging multi-trailer articulated vehicle systems. Tractor/trailers currently only

2


9/111

pull single trailers in full size applications. New government legislation will make itpossible for full size, multi-trailer vehicles to operate on Ontario roads. Our activetrailer steering system will improve low speed maneuverability of these new multi-trailer vehicles by using rear trailer steering systems controlled by a controllerwhich we will develop. The high speed stability improvements needed to improve

safe operation will come from the implementation of the proper hitching method forthe rear trailers. Systems to improve these aspects of multi-trailer vehicles will berequired for all trailer manufacturers and trucking companies who want to takeadvantage of the new government legislation to allow multi-trailer vehicles to runon the designated highway systems.

Trailer manufactures may apply our technology to incorporate intelligent control totheir trailer steering and differential braking systems. Without our controller,manufacturers will rely on applying a passive trailer steering system to their trailersteering. This would not create an optimum turning radius for the trailers and itwould not take into effect the current conditions of the truck and other trailers.

Trailer manufactures who want to have the most advanced, efficient and easilymaneuverable trailers may use our controller incorporated into their trailers as theircontrol systems. Trucking companies may introduce our controller to control anyretrofits they apply to their existing trailers to make them capable of being used ina multi-trailer application.

We know that multi-trailer vehicles will be a new addition to Ontario roads becausethe government legislation required to allow such vehicles is in the process of approval. The Ontario Long Combination Vehicle Pilot Program Conditions will makemulti-trailer vehicles a new site on Ontario roads. Similar legislation is already ineffect in other provinces and countries. We know that multi-trailer vehicles can bemore efficient than single trailers because double the load can be drawn with onetruck by adding a second trailer. Through our literature review and patentsearches, we have found that a control system does not currently exist to control asystem which includes trailer steering. We hope to design this controller to beeasily incorporated into any new trailer or existing one.

The need for this controller is imperative. Mechanical systems cannot adapt tochanging conditions or account for the additions of different trailers. Our controllerwill take inputs from the truck and all trailers to make intelligent decisions about theappropriate steering angles required to navigate corners and remain stable at high

speeds. Our controller will improve the low speed maneuverability of these newmulti-trailer vehicles.

For trailer manufacturers, the application of active trailer steering systems willstrengthen their competitive ability in the trailer market due to the distinguishedperformance of our system. Trucking companies may also benefit from the systemthrough increasing transportation efficiency and safe operation of multi-trailerarticulated vehicles.

3


10/111

Currently multi-trailer vehicles experience many hazards on the road because of their large size, high center of gravity and variable loading. Roll-over and jackknifing are common unstable motion modes for multi-trailer vehicles with noassistance systems for steering. We have been presented with the task of designing a controller to improve the low speed maneuverability of a multi-trailer

vehicle. Our controller is intended to control the systems used to steer the wheelsof the trailers. We are designing a controller to control existing systems of steering.

4


11/111


12/111

maneuverability of LCVs a controller will be employed to steer the rear wheels of the trailers keeping the trailers in line with the front axle of the truck. With thecorrect implementation, the improvements made to the operation of LCVs willmake them the obvious alternative to single trailer trucking.

6


13/111

The LCVs have two configurations: A-Train Double and B-Train Double. These twoconfigurations will be briefly described in the following sections.

LCV A-Train Double The A-Train Double utilizes an A dolly to join the two trailers together. Theoperation and linkages involved with A-dollies make them more suitable for lowspeed applications. Theoretically, the increase in articulated joints makes themmuch more effective in performing low speed maneuvers. The increase inarticulated joints decreases the high speed stability of these vehicles and creates aneed for many more corrections to be made to keep the trailers in line with thetruck when operating at high speeds. Our baseline tests of A-dollies have shownthat they do provide an improvement to the low speed maneuverability of thevehicles but they also greatly decrease the high speed stability.

The overall dimensions of the A-Train Double configuration are shown in Figure 1. The dimensions listed have a set range in which the values can vary. There areseveral configurations of trailers and trucks that can be used to make up an A-TrainDouble. The different combinations are also outlined in Figure 1. Furtherdocumentation of the weight limits can be found in Figure 2. The weights aredetermined by the width of the axles and tires.

7


14/111

Figure 1: LCV A-Train Double Description, Dimensions and Weights, 1 [16]

8


15/111

Figure 2: LCV A-Train Double Description, Dimensions and Weights, 2 [16]

9


16/111

LCV B-Train Double The operation and linkage of a B dolly is such that the high speed stability willincrease where the low speed maneuverability will decrease. The reduction in

articulated joints on a B dolly give it the opposite operational characteristics of an Adolly. The increase in high speed stability associated with a B dolly make ittheoretically more suitable for operation with our controller. The increase in highspeed stability, when coupled with our low speed controller should yield a vehiclecombination which is safe for travel on our highways.

The overall dimensions of the B-Train Double configuration are shown in Figure 3. The dimensions listed have a set range in which the values can vary. There areseveral configurations of trailers and trucks that can be used to make up a B-TrainDouble. The different combinations are also outlined in Figure 3. Furtherdocumentation of the weight limits can be found in Figure 4. The weights aredetermined by the width of the axles and tires.

10


17/111

Figure 3: LCV B-Train Double Description, Dimensions and Weights, 1 [16]

11


18/111

Figure 4: LCV B-Train Double Description, Dimensions and Weights, 2 [16]

12


19/111

Design Criteria

Functional RequirementsIn general the electronic trailer controller has to steer the wheels of all trailers in themulti-trailer articulated vehicle system. The controller will need an algorithm to takein various inputs and compute the optimal steering angle for each successivetrailer. The controller needs to reduce the rearward amplification ratio and bring itas close to 1.0 as possible. The controller must also make the trailers follow thepath of the leading tractor in order to be deemed a successful design. This meansthe off-tracking of the trailers needs to be greatly reduced or in a perfect situationeliminated all together. Furthermore, the controller needs to reduce or eliminate theoccurrence of jack-knifing, snaking, trailer swing and rollover. Due to the numerous

variations in lengths and weights, the controller will need a universal algorithmcapable of calculating steering angles based on the user inputs. Furthermore thealgorithm which computes the steering angles needs to be as simple as possibleand have absolutely no bugs in the logic. Lastly, the controller will need to outputdata to a steering mechanism which will physically turn the wheels.

Constraints The constraints used when developing our low speed stability controller were thedimensions outlined in the Ontario LCV Pilot Program. We followed the overalldimensions outlined in the LCV program. The maximum and minimum combinedlengths were used and tested using TRUCKSIM to ensure that they would each

perform the necessary tasks correctly and accurately. TRUCKSIM is a commercialsoftware package for modeling and simulating various articulated vehicles underdifferent scenarios. We used the completion of these tests to verify that ourcontroller will be capable of handling any combination of dimensions outlined in theLCV program.

When creating the 1/14 th scale prototype, the constraints which governed itscreation were the Ontario LCV Pilot Program and the controller we have designed.

The scaled dimensions of the prototype fall well within the accepted values of theLCV program. The prototype is closer to the maximum length of the LCV program.When creating the controller for the prototype, the calculation method of finding therear trailer steer angle was kept as close as possible to that of the controller toensure an accurate representation.

OutcomeBy following the guidelines found in the Ontario LCV Pilot Program we havedeveloped a fully functioning controller as well as a working 1/14 th scale RC (RadioControlled) prototype. We have shown in multiple tests that the controller greatlyimproves the low speed maneuverability of LCVs in a variety of scenarios. Our

13


20/111

prototype is able to quickly show the operational superiority of our controlled reartrailers compared to a non-steerable trailer. We can show the prototype operatingwith and without the steering controller implemented. We have shown throughextensive testing and numerous visuals that we have greatly improved the lowspeed stability of LCVs.

14


21/111

Physical Requirements There are several physical requirements that have guided the path of our controllerand prototype development. Our controller is required to improve the low speed

maneuverability of LCVs. We were to keep the overall vehicle dimensions within theguidelines specified in the LCV pilot program. The trailers are required to remainwithin the lane markers when navigating corners. Our prototype is required tovisually show the operational capabilities of what our controller will do whenimplemented in a full scale LCV. We have used sensors which will make creatingour C representation of our LabView controller as accurate as possible. LabView is asoftware package which is used to create complex block diagrams which takevarious inputs and use user generated algorithms to calculate the desired outputs.

Constraints The physical constraints which governed the completion of our controller were thelimitations associated with testing different configurations in TRUCKSIM. Thedimensional constraints of the LCV pilot program were implements to the best of our ability to assure we produced an accurate controller. We have worked aroundthe limitations within TRUCKSIM and used the values outlined in the LCV program tocreate the test platform for our controller. The physical constraints which werepresent in the prototypes development are the limitations of C programmingcompared to LabView and the limited number of sensors which could be placed onkey vehicle points. Programming in C required the controller to only include themost essential operational statements and formulas. We were unable to perform allof the feedback calculations which are included in our controller. The sensors which

were available for our prototype were chosen to fit in the confined space availablearound the vehicle axles and drive train.

OutcomeOur controller has satisfied all of the physical requirements which were set for thecompletion of our controller. We are able to keep the rear trailers within the roadsline markings while navigating corners. We have been able to run simulations withour controller under several different dimensional configurations which lie within theLCV constraints. We know that our controller can complete any scenario it is put inwhile remaining safe in its low speed operation. Our prototype has beensuccessfully created to accurately represent the path produced by our controllersimulations. We can show the operation of the scale model in each test scenariowhich is performed on the virtual controller in TRUCKSIM.

15


22/111

Literature ReviewDuring the review of the literature, it was found that the University of Cambridgeand the University of Michigan are doing work in the field of trailer steering. TheUniversity of Cambridge developed an electronic controller for the steering of asingle trailer system. The controller developed by the researchers at Cambridgemakes the center of the rear of the trailer follow the same path as the fifth wheelhitch. This allows the trailer to become more maneuverable.

Many documents were reviewed, including tests performed on current steeringtechnologies as well as research papers on developing technologies. Most of theliterature reviewed focused on one area of trailer performance, such asmaneuverability or stability. The information obtained from the research papers willaid in the design of an electronic trailer steering controller.

Review of University of Cambridge Technical PaperA recent research paper published by the University of Cambridge was reviewed.

The paper titled "Improving roll stability of articulated heavy vehicles using activesemi-trailer steering" [15] focuses on the methodology used in the design andimplementation of the controller. In this paper, the authors design a virtual driverfor a single semi trailer. The electronic controller is the virtual driver for the rearwheels of the trailer. It receives inputs from sensors and calculates a path in orderfor the rear of the trailer to follow the front of the trailer. This path following keepsthe rear of the trailer from deviating from the path of the front of the trailer.

The controller is designed by analyzing the dynamic model of the truck and trailercombination. Using this information, a set of dynamic equations can be formulated.

These formulas are then used as the basis for the controller design. Themethodology used by the research team at Cambridge will serve as a basis forthe multi-trailer controller.

The testing of the controller showed the performance of the trailer in comparison totrailers with no steering, self steering and command steering. The electronicallycontrolled model displayed an improvement over the other types of steering. Thisinformation could serve as a baseline for performance standards and testing.

Limitations of University of Cambridge Paper One downside found in the testing of the controller was that the tests focus

on low speed maneuvering. The testing did not account for high speedstability.

The test did not analyze forces and torques involved in vehicle stability

16


23/111

The test did not consider the effects of road surfaces and LCV geometry andconfiguration

Review of Ontario LCV Pilot Program Conditions The main focus of this paper is the conditions for licensing and driving LCV in

Ontario. This paper provided rules and regulation on the size of the trailer thatwould need to be tested. The information obtained can be seen in the appendix. Italso states that the trailer must incorporate an ABS breaking system as well asbeing able to steer +/- 40 degrees and incorporate electronic stability control. Thisinformation will be useful in determining which features to include in the controller,as well as designing a worst case scenario truck model [16].

Limitations of the Ontario LCV Pilot Program The pilot program did not take into account the situations where trucks and

trailers came as separate bodies from different companies, as they do inmany countries

The program thoroughly limits the total length of the LCVs to be seen onhighways

Due to this Pilot program, importing LCVs from other countries is of greatconcern since many international LCVs are beyond these limits

Addition materials reviewedIn addition to the previously mentioned articles, many other articles and researchpapers were found. Some of the information contained in these additional paperswill be useful over the course of this project. Information includes prior testing of trailer systems, background information and statistical information. These articleswill provide excellent information in choosing and testing a final product.Information on testing conditions can be found in many of these articles, which willaid in the development of testing conditions for the electronic controller. They arealso extremely useful in providing general information about the transportationindustry, allowing the team members to increase their knowledge on the subject.

The additional papers reviewed are listed in the References section.

Limitations of the Additional Material Most the articles are focused on self steering concepts, so they have the

same idea Some papers ignored dynamic s of LCVs and therefore the performance of

their controls might be questionable until they proof that it does work

Many of the articles did not include other factors affecting constitutingrollovers such as road surfaces vehicle configurations

17


24/111

Patent ReviewA review of the current trailer steering patents showed that there are few electronicsteering controllers developed. Any controllers that have been patented areprimarily focused on a single trailer model. There are numerous patents developedfor mechanically controlling the steering, basing the trailer axle steering radius onthe angle the trailer makes at the fifth wheel hitch. They use long steering rods tocontrol the geometry based trailer steering. The review of the patents has shown usthat mechanical geometry based steering is a mature technology. Using thisinformation, the focus of the project can be focus on replacing the mechanicalsteering mechanisms with an electronic controller.

Robotic vehicle that tracks the path of a lead vehicle

United States Patent number US 6,494,476 B2, which is dated December 17, 2002contained valuable information and design criteria from which we can use and learnto help us with the design of our controller. In general, the patent advances theconcept of a robotic vehicle capable of tracking the path of a lead vehicle. Theinventions objectives are to have a system fully capable of steering itself inresponse to an input, design a mathematical model to allow both path tracking andnon-path tracking steering algorithms to be combined, provide numerousalgorithms based on physical principals of the vehicle configurations which will beused for steering of the robotic vehicle, present an electronic control system whichincludes all hardware and present a mechanical system which is capable of beingcontrolled by actuators to track the path of the lead vehicle. The invention usesthree different steering algorithms to control a single trailer based on several inputsfrom various sensors. The three different modes include relative angle steeringmode, rate of orientation steering mode and variable ratio with over steer whichwas taken from US Patent number 60, 204, 513. The invention uses sensors togather information such as angle between the tractor and trailer, distance travelledby the tractor, distance travelled by the trailer and angle between the tractorcenterline and centerline of the tractors front wheels. The invention steers itself byturning the entire axle assembly for the modified dolly about a single fixed axis onthe robotic trailer. The invention only controls the rearmost trailer as to correct thetrailers ahead of it. Also the system incorporates two pneumatic motors which can

provide power to the wheels of the dolly for additional steering force. Furthermorethe invention uses a long rigid tongue to connect the trailer in order to allow roomfor the mechanical steering components for the dolly.

Although this invention is not completely related to our focus, we would like toexpand off of this patent because does have some key features and similarities toour basic design. We will however have some major improvements to this system.

The first improvement will be in the controller algorithm. We hope to design a single

18


25/111

algorithm which can be used for any combination of tractor and trailer. Thecontroller will take the different lengths and weights of the units and use them in ageneral algorithm to control the trailer. Furthermore we will incorporate speedsensors which will be used to influence the outputted steering angle. This willeliminate the need for different steering modes which are used in different driving

situations like the reviewed invention has. Also we will steer each wheel of eachtrailers individually as opposed to steering the entire axel assembly of only therearward trailer. This will eliminate the need for a long tongue section on the trailersdolly which will allow more space for cargo storage due to the current overall lengthregulations. [17]

Limitations of Patent: US 6,494,476 B2 Invention does not take into account forces applied to LCV bodies

The control proposed by the inventions only controls the rear most trailer tocontrol other trailers ahead of it

Control algorithm is only limited to few trailers behind the tractors

Multi steering controls might not be effective as it is likely to add more timewhile changing from one steering method to another

The regulations on the lengths of LCVs will limit the lengths of the trailersproposed by the inventions

Differential Braking Systems for Tractor Trailer TrucksIn mechanical engineering, differential breaking is a system in which a braking

operation depends on a difference between two motions. These two motions are thespeeds of two wheels being used to drive a vehicle. The two wheels could refer toeither front or rear wheels on the same axle. It is well known that when the vehicleis cornering on a curved track, its wheels rotate at different speeds. In differentialsystems, the idea is to design a mechanism that will drive a pair of wheels withequal force, while allowing them to rotate at different speeds. The concept of differential breaking comes in to play when the team started looking at possiblesolutions to our problem of controlling vehicle (multi trailer systems) dynamicbehavior on curved tracks and lane changing periods. As the team continues tosearch for control designs which may lead to the solution of the problem, weinvestigate the use of a patented vehicle dynamic controller which was encounteredin our research. The following is a block diagram of a differential breaking controland paragraph that follows the diagram describes how it works. [18]

19


26/111

[18]

Figure 5: US 3,894,773

From the above patent, a sensor based potentiometer is used to prevent jackknifingand rollover to improve vehicle stability during deceleration and lane changing onhigh ways. The potentiometer measures the angle between the centre lines of tractor and trailer. This potentiometer is mounted at the back of the tractor and atthe same attached to the hooking mechanism of the tractor and trailer. Themethod uses computed angular rate information as the input to the system. There

is a point called the null point; this where the tractor trailer angle is equal zero. Toavoid the need for periodic calibration and drift correction in the control logic, thisnull point should nominally be kept at zero output from the potentiometer toprevent saturation of the control circuit. As seen from the above diagram, theinput into the control is the tractor trailer angle. The differentiator control module isthen used differentiate the angle into angular rates of the front wheels of the tractorand angular rates of the tractor trailer angle. Other speed sensors are used tomeasure the speed of each tractor wheel in the front. The two angular rates of speeds are summed up together and the result is added to the differentiatoroutcomes, which will then alert the driver to apply differential break forces to thetractor front wheels based on the computed sensor values. [18]

The control looks somehow simpler and easy to implement. It does address theconcepts of jackknifing and rollover prevention. However, there are some limitationsto this control. Also the use of centre of mass of bodies is also not addressed in thispatent and hence, ideas from this patent could only form part of conceptdevelopment as the team decided that there is a possibility of using differentialbreaking as an addition to the current solution which has already been chosen. [18]

20


27/111

Limitations of Patent: US 3,894,773 The differential breaking system does not address problems of trailer swing

and stability at high speeds

The control does not use centre of mass of composite bodies, no torques

and forces involved No trailer and or truck geometries involved but this could greatly affect the

performance of such a control

The differential breaking is only suitable for LCVs with pulling few trailers

Does not fully address problems encountered at very high speed such assystem in stability

LCV will still drive at limited speed in highways

21


28/111

Rollover Test and Double Trailer CombinationIn this article, two types of trailers have been compared by testing how each truck-semitrailer performs when exposed to rollover conditions. Full-scale tests were

conducted of the rollover stability of several double trailer truck combinations incornering and single and double lane change maneuvers. This comparison involvedthe A- train and the B-train types configurations with more than two articulationpoints. A train had two articulation points while the B-train had three articulationpoints. Trailer suspensions springs and converter dolly hitch were also investigated.

Figure 6: Rollover Test, Double Trailer Combination [6]

It was found that the B-train configuration has the speed at incipient rollover of about 10% higher than that of the A-train in dynamic lane change maneuver. It wasalso concluded that the fewer the number of articulated points for the B-train andthe roll coupling between its trailers, the less the rearward amplification of rollresponse. The rear most trailer could have an acceleration of 20% higher than theA-train without rolling over. The only disadvantage with the A-dolly truck trailerconfigurations is that they are all subject to improvement when it comes low speed

maneuverability, that is to say they exhibit poor cornering maneuvers. Also the A-trains become unstable in high speed highway lane changes.

Limitations of the Rollover Test for Double Trailer Combinations The test could have been extended to more than two trailers

The test was limited to high speed rollover testing only but performanceagainst each other under low speed is not considered

22


29/111

The tests did not account for different road surfaces therefore its validitymight be questioned by some appropriate organizations

Could have also considered systems of different steering mechanisms ratherthan using the conventional truck-trailer systems

Test of Self-Steering Trailer Axles The figure below depicts the sketch of a design of a self- steered trailer axles. Thistest was done in Ontario and because of its success, many manufacturers havebrought these systems in to use in Ontario.

Figure 7: Self-Steering Trailer Axles [7]

It has been noticed that self-steered vehicles exhibits reduced tire wear whencornering with wide axle spreads developed to take advantage of load regulations.Self-steered axles also improve low speed maneuverability at reduced off-tracking.

The purpose of these tests was to ensure there is some trade-offs between theabove advantages and the cost of reduced trailer stability and safety in steeringand breaking maneuvers at normal speeds. The test was done in normal trucktrailer system vs. self steered trailer axles, either a trailer with one or two self-steered axles. They chose a tri-axle van equipped with two self-steering axlesystems (one with single axle and one with a bogie), tested for both low and highspeed maneuvers. The results of the test were compared to those of standardconventional semitrailers without self-steered axles. It was found that both the

single axle and the bogie exhibit reduced tire scrub. However, its been identifiedthat only the self-steering bogie is capable of reducing trailer off-tracking while thesingle axle showed no results of improved off tracking. For the high speed breakingand maneuverability test, the bogie system failed to reduce the instability of thetrailers and this was not acceptable by the designing companies. This became thearea of improvement where the manufacturing companies are currently working on,and its a similar project assigned to my team. The team is currently attempting to

23


30/111

improve these trailer self-steered axles by embedding feedback controllers in to thesystems. The proposed system is expected to be more integral than any LCVsystem currently in practice.

Limitations of the Self-Steering Trailer Axles

The test did not consider effects of road surfaces No system geometries were considered in this test

No performance under load was tested

There test could have been repeated with vans using many self steeredaxles- more than three

Steering Method for Trailing Section of an Articulated Vehicle

Figure 8: Steering Method for Trailing Section [10]

The invention introduces a control system that could be used to steer the trailers of articulated joints in a multi-trailer systems, or the so called long combinationalvehicles. The inventor focused on using a self steering dolly that could possibly besteered manually or automatically switched among many possible modes. Basicallythis dolly receives yaw rates information from the tractor/truck front steering axle orfrom the rear of the front trailer. In this invention, there are two ways to combinethe steering mechanism of the LCV trailer systems. In the first case, the trailer axlesare steered in the opposite direction to that of the towing vehicle to improve theslow speed maneuverability of the LCV. On the other case, trailer axles could besteered in the same direction s that of the towing vehicle, depending on theintended direction of the systems, such as in making a reverse. This will results in amore stable system in a high speed maneuver. However, the fact that the inventoruses an A-dolly articulation point will make the LCV to perform poorly under normal

24


31/111

operations. Therefore the system will exhibit off-tracking and unstable at highspeed. The team found that ideas adapted from this invention could be of mainimportance and might as well develop a control that could attempt to self steer thetrailers at both high speed and low speed, in a TruckSim software package.However this control will not be an easy thing to design, especially because the

teams focus will be to integrate other softwares into TruckSim to create suchcontrols.

Limitations this Steering Method for Trailer Section Might not be effective when the LCV has many trailers

At high speeds in LCVs with many trailers, more load will be shifted to theopposite direction hence there is a risk for rollovers and jackknifing inhighways

The mechanism will be suitable for the B-dolly connections only

Geometries of the truck and trailers need to be taken into account in thiscase

Design Solution The teams optimal design was generated using some of the ideas and basicconcepts we found in our research and couple them with the ideas we createdduring our brainstorming sessions and meetings. Our optimal design will utilize asingle controller which will calculate and output the steering angles for all thewheels on each of the separate trailers being towed. We decided to use a single

controller in order to keep the cost of the unit itself down and also because wecould design a unique algorithm will be completely adaptable to any number of trailers. This allows the unit to be used in a very large number of applications.Furthermore we have concluded that we will need various sensors to record datasuch as yaw angle, yaw rate, velocity, distance travelled by both the tractor andtrailer and steering wheel angle. From these variables we will be able to calculateall other variables we need in order to derive the optimal steering angle for each of the wheels on the trailers. After researching both the A-dolly and B-dollyextensively, we have decided to use the B-dolly in our final design and testingprocess. The B-dolly was picked because it offers better high speed stability than

the A-dolly does. Although the B-dolly has poor low speed maneuverabilitycharacteristics, our trailer steering system will get rid of that problem. Byincorporating velocity sensors in our design we have designed a controller that willautomatically compensate for the different steering ratios needed at differentspeeds. This is a major improvement from current systems which utilizes differentmodes that the user has to switch between based on the different driving situations.We feel the less the user needs to input data or change settings with the controllerthe less likely it is a malfunction or miscalculation will occur.

25


32/111

The design of the virtual controller is done using LabView which interfaces nicelywith the simulation software TRUCKSIM. TRRUCKSIM is used to design the actualtractor and trailer combinations as well as simulating our various designs.

TRUCKSIM has given us some basic baseline truck and trailer combination designsthat we will need to modify to meet our requirements. From there are be able to run

tests based on the SAE standards and gather various data to be further analyzedand improve the controller code. Repeating this process over and over has allowedus to arrive at the best possible solution which satisfies all project requirements.

Final SolutionOur final solution utilizes one LabView controller to take various sensor readingsfrom both the truck and trailers. The readings are sent through our algorithm andsteer angles are produced for each of the rear trailer wheels. The steering anglesproduced are able to successfully guide the trailers around a corner in the samemanner as the truck. We have been able to show in multiple tests that ourcontroller is capable of navigating any type of turn it will encounter on our roads.

We have taken special consideration to develop a 1/14 th scale prototype that willfunction in the same way a full scale, controlled LCV would using our controller. Wehave performed extensive modification to make the rear trailers not only steerablebut controllable. We have included a B dolly to link the trailers and created aprogram similar to our LabView in C to control the steering. We can demonstratevery easily the effectiveness of our controller design by driving our prototype inboth controlled and un-controlled modes. It is easy to see how much more controlis delivered to the system when the rear trailer steer is initiated.

Constraints The one major constraint of our controller is the lack of an on board database whichhouses all of the truck and trailer combinations that would be possible under theLCV program. With further time, development and availability of detaileddimensions of all types of trucks and trailers this database could be produced andimplemented alongside our controller. A database would allow for instantreconfiguration of an LCV without the need for reprogramming. The only constraintthat exists with our 1/14 th scale prototype is that of weight. The axle weight of ourscale model is not the same as a full LCV. This is due to the delicateness in this1/14 th RC model. With further development it would be possible to designcomponents which would support such loads but for our purposes the visualrepresentation of our prototype is more than enough.

FeasibilityOur controller has proven to be extremely feasible in todays market. Thecontroller, when produced will have no other costs associated with it other than thecost of trailer modification. Cooperation between both truck and trailermanufacturers will be paramount in the final implementation of this vital tool.

26


33/111

When trailer manufacturers have the robust systems of steering in place LCVs willbe more than able to perform the same low speed maneuvers as single trailertrucks.

Engineering AnalysisBasic Math Model

The basic math model is represented in the figure below. For this model, rear axlepath following was implemented. This means that the centre of the axle of eachtrailer follows the centre of the real axle of the trailer. This basic model built doesnot consider, time or velocity. It also assumes small angles of turning. The greaterthe steer angle, the higher the degree of error in the outputted trailer steer angles.

In formulating the math model for this situation, like angles were used in order tokeep the turn radius of the rear tires of the truck constant with each trailers steer

axle. The following equations formulated the basic math model.

Assuming, for path following of the rear axles, the ideal situation will give:

And using the trigonometric relationship:

Where:

Since the radiuses of the turns are assumed to be equal, the angle at the hitch willequal the steer angle created by the controller, or

Using a relationship and assuming small angles:

A simple relationship between the trucks front steer angle and the trailer steerangles can be made.

27


34/111

Therefore the basic constant radius a simple controller can be constructed.

Note that at high steer angles, ( F1>10 deg) the error associated with this methodbecomes quite large. It should also be noted that this model is only accurate forconstant radius turns, since it does not consider the velocity of the vehicle.

Advanced Math ModelAn advanced model was then created using 5 DOF. Due to the limitations of thesolvers in the TruckSim, calculations will be based on a two axle truck and singleaxle trailers. The data that is gathered can be extrapolated from these findings aswell as serve for a base for future models with multiple axles.

Assuming small angles where:

This leads to a constraint between the tractor and the trailer each trailer where:

Using a linear model for the tires and the above equations, a state space model canbe constructed. The state space equation has not yet been constructed for thismodel at this time. Further calculations are required before this model can beimplemented into a controller. Consideration for a more advanced tire model is

being made, prior to advancing this model into matrix form. This model will be usedas a preliminary model in the next semester. Variables in this model will beoptimized using matlab.

A limiting factor in using this 5 DOF model is that it does not consider the roll of theaxle in relation to the suspension and neglects dynamic forces such as windresistance. Also, it assumes a linear tire coefficient. A more comprehensive nonlinear model could be developed pending the success of testing of this model. Byadding more degrees of freedom as well as non linear properties, the accuracy of the controller can be greatly increased, but will also increase the complexity of thecontroller. The additional complexity will possibly require more input data to be sentto the controller, which could increase chance of failure.

Variables to Consider in the Controller

Rearward Amplification; Load Transfer Ratio; Trailer Overshoot (TransientHigh-Speed Off tracking)

o The standard test SAE J2179 is used to evaluate RWA, load transferratio and trailer overshoot. TruckSim, a computer based modeling

28


35/111

system will virtually show the performance results of this test andcollect values of these three criteria. RWA will be verified if the value islower than 1.5. The load transfer ratio should be 0.6 for the lanechange test and 0.75 for the low speed maneuverability test. Trailerovershoot should be under 0.8m [24].

Static Rollover Threshold (Static Roll Stability)o TruckSim will be used to reach and verify the rollover goal of minimum

0.35g to 0.4g [24]. High-Speed Steady State Off tracking

o TruckSim will be used to reach the recommended value of 0.5m at alarge constant radius turn of 393m at 100km/h [2].

Low-Speed Off trackingo The TruckSim test will validate the performance when an offset of

7.4m is reached during an 11.25m 90 steer path [24]. Frontal Swing

o TruckSim and first principles can validate the recommended value of 1.5m from 11.25m 90 steer path [24].

Tail Swingo TruckSim will be used to verify that the recommended distance of

0.5m overhang during an 11.25m 90 steer path [24]. Tracking Ability on a Straight path;

o The recommended value of tracking in a straight line is the trailersmust stay within 100mm on both sides of the tractor unit path [24].

Optimization of Variables

Optimization will be performed using MATLAB. The values and parameters will be

optimized to improve the overall performance of the double trailer B dollyarticulated vehicle.

Rearward Amplificationo Improving RWA involves decreasing the amount of articulation joints.

This requires no optimization due to the limitation of the two trailer Bdolly setup. This arrangement must use two articulation points, onewith the tractor towing unit and the other on the B dolly. Shorteningthe distance between the hitch point and CG of the tractor towing unitis a variable and improves the RWA [24]. This is a variable that will beoptimized. Decreasing coupling rear overhang can also be optimized toimprove the RWA [24]. Increasing the wheelbase on trailers alsoimproves RWA and will be optimized [24]. There is an inverserelationship with RWA and tire cornering stiffness. A higher corneringstiffness lowers the RWA [24].

Load Transfer Ratio

29


36/111

o The load transfer ratio is significantly correlated to static roll stabilityand RWA [24]. Optimizing these factors will decrease the load transferimproving performance and safety.

Trailer Overshoot (Transient High-Speed Off tracking)o Overshoot will be monitored and optimized through virtual testing and

observing the program computed data. Static Rollover Threshold (Static Roll Stability)

o From first principles (appendices) acceleration in gs = velocity^2/(turn radius*g) = track/ (2* C.G height). Rollover stability will improve if track width is increased or CG height is decreased. This will beoptimized for the best combination.

High-Speed Steady State Off trackingo Since the lateral force is proportional to the square of the velocity, and

inversely to turn radius, at high speeds the lateral force increases. Thiscauses tracking outside that of the tractor steer axle [24]. This offsetwill be optimized through virtual testing.

Low-Speed Off trackingo Opposite to RWA, increasing the distance from the kingpin to thecenter of the rear axle will not be beneficial, it will increase the off

tracking [24]. An optimized compromise will be met. Smaller turnsincrease offset while increased speeds decrease it [24].

Frontal Swingo Is directly related to the amount of overhang forward of the steer axle

and the forward of the kingpin for semi-trailers [24]. Tractordimensions are set per tractor make and model where as the trailerswing can be optimized by varying the distance in front of the kingpin[24].

Tail Swingo Tail swing is directly related to the amount of rear overhang [24].Large overhang creates large tail swing. Optimization of this value willbe reached to create the best result while minimizing the effectstowards the other variables.

Tracking Ability on a Straight path

o This depends on many factors. The number of trailers and couplinginvolved, axle alignment, tire cornering stiffness, vehicle length, speed,wheel base of prime movers and trailers, tow coupling overhand andthe location of the fifth wheel on the tractor [24]. Optimization of thesevariables will be computed to get the best result.

30


37/111

Controller Design Using LabView The virtual controller was designed and developed using the LabView platform. Testing was performed through TruckSim simulations. Two controllers weredesigned. One is based on kinematic constraints, while the other is based on thedynamics of the system.

Design ConsiderationsIn the design of this controller, many things were considered. The final design needsto be functional and realistic. It also needs to be viable. The code written also needsto be user friendly and easy to understand, allowing future users or developers tomodify and enhance the code with minimal effort.

Design Resources and Tools Used The design and implementation of the controller was assisted with the use of computer software. LabView was used to design the controller. TruckSim was usedto simulate the effects of the controller. With the use of these software packages,the design could be tested and simulated.

LabViewLabView was used in this project for developing the controller code. LabViewprovided a robust building block for development with various integrated scripts. Italso provides a great graphical user interface making it easy to use. As a simulatedcontroller, the LabView program can be seamlessly integrated into TruckSim,

allowing for virtual testing.LabView script can also be enabled on microcontrollers based on various ARMmicroprocessors. This would allow for ease of developing the virtual model into aphysical model.

TruckSim TruckSim provided the virtual plant for the controller simulations. Various modelscould be constructed and tested. Also various roads and conditions could besimulated. The use of TruckSim allowed the controller to be tested numerous timesand in different situations. Future testing can easily be implemented and tested on

TruckSim based vehicle simulators.Methodology and Critical Thinking

The final controller will be designed to increase the maneuverability at lowerspeeds. This is accomplished by implementing a path following strategy. For highspeed stability, the controller needs to reduce the rearward amplification, or atminimum, not increase it.

31


38/111

Path Following StrategyIn order to increase the maneuverability of the vehicle, a path following strategyneeded to be implemented. As the lead unit of a conventional steered, 3 unitvehicle travels around a curve of radius R, the dynamics of the system will causethe trailing units to travel around a path with the same centre of rotation, but the

radii will be smaller. The diagram below shows the paths followed by the individualunits centre of gravity as the combination travels around a constant radius curve.

To improve the maneuverability of the system, the goal is to eliminate thedifference in the radii. This is accomplished by implementation of a path followingstrategy, where a point on each trailing unit follows the one in front of it. Since thegeometric properties of each unit are different, only one point on each vehicle canbe selected. In research performed at the University of Cambridge, [8] a pathfollower strategy was implemented by having the rear centre point of the trailerfollow the hitch point on the truck. This could be extended to multi trailer units,using the hitch points on all trailers other than the rear trailer. In order to increasethe modularity of the controller and allow for the possibility of different trailercombinations, such as single, double, triple, etc, the strategy of this controller usesthe centre of the rear axle in the truck as the path and the centre of the trailer axleas the follower. The diagram below shows a representation of this strategy.

Figure 9: Path Following Strategy Diagram

The decision of this approach is based on kinematic steering principles. As a vehicletravels around a bend, the centre of gravity rotates around the centre of rotation.Assuming the vehicle is designed to adhere to the Ackermann steering principles,

each wheel will be tangential to a line pointing through the same centre of rotation. The vehicle will also require a certain amount of space based on this principle. Thediagram below shows the principles behind the Ackermann steering as well as theswept path of a single vehicle.

32


39/111

Figure 10: Vehicles Required Path [25]

As the vehicle gets larger, the path required to navigate the turn also increases.Even with trailer steering, the swept path of the trailer will be greater than the pathof the truck. For the purpose of the controller, to improve the maneuverability, thevalue for Rmin should be maximized. Also, the trailer, including any overhang,

should not extend past the Rmax value for the truck. This would allow a driver tomaximize the turn without having to worry about the positioning of the outerportion of the trailer.

Rearward Amplification Reduction Strategy The controller is designed to be implemented on vehicles with a B dolly. Since the Bdolly is more stable at high speeds than the A dolly, initial controllers will bedesigned to limit the controllers influence at higher speeds. As the design of thecontroller progresses, the rearward amplification will be controlled dynamically byreducing the load transfer ratio. Also, implementing a peak and hold system couldbe implemented. By using the peak value obtained by the lead unit, counter

measures can be implemented in order to limit the rear trailers peak to be less thanthis value, resulting in a ratio of 1:1.

Kinematic Controller

Overview The first controller designed was based on kinematic properties of the system.Using a kinematic model is accurate at lower speeds, but the dynamics of the

33


40/111

system will cause adverse effects at high speeds. The main purpose of developingthis kinematic control was to enhance our knowledge of the system, while gainingfamiliarity of the software packages being used.

Design

The idea for the controller is to have the centre of the trailer axles follow the centreof the truck axle. To achieve this, wheel speeds are sampled from each wheel.

Using basic trigonometry, the radius of the curve being traveled is calculated. In thisimplementation of the controller, the angle of the trailers steering is determinedusing two calculations. First, the controller calculates the change in the steering of the lead unit. This number is used in the first part of the trailer steering anglesoutput by calculating an anti steer angle. Basically, if the truck is turning right, theanti steer calculation turns the trailer wheels left. This is needed to allow the trailerto travel the same path as the lead vehicle. The number is also stored in a queueand used again in the second part of the calculation. After the vehicle has traveled

a distance equal to the distance between the path point and the follow point, thecontroller will release the stored value to be used in the formula as a correctivesteer number. At any given time, the output to the steering of the trailer is acombination of a real time anti steer angle and a delayed corrective steer value.

The basic flow chart is show below.

34


41/111

InputsIndividual Wheel SpeedsTrailer Articulation Angle

Calculation Of Counter Steer

Angles For

Truck Steer InputCalculations

Figure 11: Kinematic Controller Model

ResultsOther than the errors caused by the limitations discussed below, this controllerperformed well at low speeds. The paths of the trailers followed closely to the pathof the truck. Results can be seen in the TruckSim section of this report. Since thiscontroller is designed for pure low speed maneuverability, the high speedperformance was extremely poor, which was expected since generally anything thatimproves the low speed maneuverability will cause high speed stability to getworse.

Limitations The major issue with this controller design is the lack of feedback control. It is anopen loop design that allows the trailers to follow the path of the truck. Afternumerous turns, any error that occurs in the calculation can be compounded andcause unstable conditions for the system. These errors can be caused by thedynamics of the system, a queue size that is too small or auxiliary inputs, such ashitting a curb. Another limitation is that it does not consider the vehicle at highspeeds. The LabView code is also a bit confusing and hard to follow.

35


42/111

Although this controller can be refined by adding feedback, improving the anglecalculations and limiting its operation to lower speeds, a more sophisticated modelshould be developed for real world applications. Also, the next design should also bemore user friendly.

The developmental path of this controller in LabView has ceased in order to focuson a more advanced model. Although development has stopped for this model,subsequent designs are based on the principles learned during its development.

Future Developments The development of this controller has ended in order to focus on the dynamicsystem controller. This controller could be advanced with additional focus onfeedback and error correcting, but since it does not consider the systemdynamically, will not perform well at high speeds.

Due to the differences in the dynamical properties of the scale model and a full

sized vehicle, this controller could be used as the basis of the scale modelcontroller. The code could be refined and enhanced to suit the models properties,then, either be converted to C Code, or ran directly on a controller powered by anARM processor.

Dynamic ControllerBy using a state space model of the vehicle system, the overall effectiveness of thesystem can be enhanced. Since the vehicle is a dynamic system, a more accuratemodel is build to provide better results.

Design

Using the knowledge gained with the kinematic controller, work on a new controllerwith error correction. Various controller topographies were considered. Aproportional integral derivative controller (PID) could be used. PID controllers arecommonly used in various applications. The feedback system allows the controllerto limit errors. Another consideration was using an optimized state space model.After reviewing controller setups that are integrated into the LabView program,utilizing a Model Predictive Control was chosen.

A Model Predictive Controller or MPC is a type of control that predicts future outputsand errors and combines this with past data to optimize the parameters of thecontrol action.

Using LabView, a basic MPC was implemented using knowledge gained from theprevious controller. This controller was implemented to assess the properties of theMPC and to decide if this type of controller should be considered for futureimplementations of the controller. A basic controller was developed by calculatingthe radius of curve for the truck and using it as a set point for the controller. Thecontroller then adjusts the curve radii of the trailer steering to follow the set point.

The control action of the MPC is then converted to steering angles and fed back to

36


43/111

TruckSim. Since this model was designed to assess the functionality of an MPC, noqueuing was used. For low speed cornering, the controller allowed the trailers totrack the trailers to a reasonable degree. At the end of the curve, the truck didcontain an output error, but this could be attributed to the basic functionality of thecontrol. The output can be seen in the figure below.

Figure 12: Vehicle Paths - Simple MPC Control

Since this controller was designed to assess the function of the MPC, the resultsappeared reasonable. This controller will not be refined to achieve better results,but a new MPC based controller will be implemented based on the knowledgegained during this implementation.

The design of the final controller had 2 main focuses. One focus was to obtainaccurate results, while the other focus was to make the controller as user friendly aspossible to allow for ease of future development and refinements. For the purposeof this report, only the controllers function will be looked at. More information on

setting up the controller can be found in the supplement, using the CAPSTONELABVIEW.

An MPC controller was designed in order to achieve optimal results. To implementthe MPC controller, inputs, outputs and controls used need to be defined. Theproposed defined parameters are outlined as follows:

Term Component Notes

37


44/111

Plant Long CombinationVehicle

2 Trailer B dolly system

Plant output Trailer Off tracking ErrorRearward Amplification

Values read into the system

Plant Setpoint

0 for off tracking error0 for RWA

Ideal goals for the outputs.

ControllerAction

Trailer Steer Angles The action implemented by the controllerto achieve the goals

Disturbances External Road Inputs Not included in this implementation

To obtain more accurate results, a five degree of freedom (5dof) bicycle model wasconstructed. The free body diagrams, system of equations and additional equationscan be found in Appendix 1.

Combining all the above equations, the following a state space model is derived andconverted to the discrete state space model to be used in the controller:

Input: Xdot = AX + BSOutput: Ydot = CX + DS

Converting to discrete time, the equations become;

Input: X(k+1) = A(k)X(k) + B(k)S(k)Output: Y(k+1) = C(k)X(k) + D(k)S(k)

Where, A is the state matrix, B is the input matrix, C is the output matrix, D is thedisturbance matrix, X is a matrix containing the state variables and S is the controlvariable matrix. Matrix values can be found in the Appendix.

The basic flow of the MPC Controller is shown below. Values are retrieved from TruckSim. The values are then used to calculate the off tracking error. These valuesare fed into the MPC controller. The state of each vehicle is recorded and stored in aqueue. As the vehicle travels forward, the data of the queue is updated. Dataretrieved from the queue is used to compare the path point with the follow point.Using the supplied values, the controller will calculate the optimum steering angles

38


45/111

for the trailers.

Trucksim

Set Point Setup

Storage of Trailer

Retreival of required states

TruckSim Solver Calculations

Stored Data

Figure 13: Advanced Dynamic Controller Flow Chart

LimitationsOne of the limiting factors of the State model is that it is linear and the system in

non linear. Angles are assumed to be small, resulting in greater errors whenarticulation angles are high. The velocities of the vehicles are dependent variables.Due to this limitation, either a single model that best suits the system at all speedsneeds to be used, or a lookup table with different models based on speed needs tobe implemented. Another limitation is that it may require too much processingpower in a physical controller.

Other limitations will be included in the next version of this document.

Future Development This controller could be easily enhanced for future development. The controller

could be enhanced by increasing the models degrees of freedom. Also, the linearmodel could be replaced with a nonlinear model.

The model designed in a good platform for future developments. Also, since thecontroller was designed with future development in mind, the user interface allowsusers to easily adapt or modify the model.

39


46/111

Additional Controller DevelopmentsWhen connected to TruckSim, LabView needs to send data to each wheel for thesteer input. Initial models just used equal angles for all wheels in each set. Since

real trailer steering will not be set up this way, the final controller used a moreoptimal design. The final controller will have one steering axle per trailer, set tooutput pure Ackermann geometry, while the other axles will self steer. The anglesused for the self steer axles will reduce the scrub. Also, the final setup also limitsthe maximum angles that are obtainable by steering. This final model is designedas a separate VI and will be able to be ported to other models in the future. Below isa diagram of the steering principles.

Figure 14: Trailer Steering Geometry

40


47/111

LabView Block Diagrams The following block diagrams represent the final controller developed in LabView. This controller was used to test in TRUCKSIM. We have verified that this controller

significantly improves the low speed maneuverability of LCVs.

Figure 15: LabView Initialation Diagram

41


48/111

Figure 16: Kinematic Controller Diagram

42


49/111

Figure 17: Dynamic Controller Diagram

43


50/111


51/111

Figure 19: A-Dolly vs. B-Dolly 2

The above figures are TRUCKSIM screen shots of the SAE standard lanechange test. These figures have overlapped runs. The blue LCV vehicle is astandard B train while the white LCV vehicle is a standard A-dolly. It can beseen that the added articulation joint of the A dolly makes the LCV moreunstable at high speeds. This shows difference between A and B dollyconfigurations. At higher speeds the B train is more stable compared to the Adolly. At low speeds the B train is less maneuverable compared to the A

dolly. Due to this our design incorporates the B train configuration. Ourcontroller improves the low speed maneuverability while using the benefit of the B train at high speeds.

45


52/111

Figure 20: A-Dolly vs. B-Dolly Lane Change Graph

This graph was produced from the lane change. It shows the roll angle of both LCVs. The B train LCV has less roll compared to the A dolly LCV. The Adolly LCV becomes unstable and rolls over which is shown by the black line.

Figure 21: A-Dolly vs. B-Dolly Lateral Acceleration Graph

46


53/111

Figure 22: A-Dolly vs. B-Dolly RWA Graph

The above figures are RWA graphs of the B train and A dolly LCVs. Thesegraphs were produced from the lane change test ran at 100km/h comparedto the standard 88km/h. The 100km/h gave the B train a desired RWA whileincreasing the A dolly RWA to an undesired amount.

Figure 23: Lateral Acceleration and RWA, B-Dolly

Figure 24: Lateral Acceleration and RWA, A-Dolly

The data from the above tables was extracted from the RWA graphs at 100km/h. The negative values imply lateral acceleration in the opposite

47


54/111

direction. Unit 1 stands for the tractor unit while unit 2 stands for the lasttrailer. Although visually shown from trucksim already, the data from the Adolly table shows higher undesired RWA values. In the B train table all valuesare well within the acceptable ratio level around one once again showing thebenefit of the B train at higher speeds.

B Train loaded unloaded Speed Truck t1 t2

RWA

truck unloaded t1 t2

RWA

50.0004

470.0003

660.0002

230.5

0 0.0004360.0002

720.0002

030.4

7

10 0.00180.0014

7 0.00090.5

0 0.00175 0.00110.0008

350.4

8

150.0040

60.0033

40.0020

60.5

1 0.003940.0024

9 0.00190.4

8

20 0.00726 0.00595 0.00368 0.51 0.00703 0.00446 0.00341 0.49

25 0.01110.0092

40.0058

30.5

3 0.01080.0069

60.0053

60.5

0

30 0.0161 0.01340.0085

70.5

3 0.0155 0.010.0077

70.5

0

35 0.0219 0.0184 0.01190.5

4 0.0206 0.0136 0.01070.5

2

40 0.0279 0.0237 0.01580.5

7 0.0264 0.0176 0.01410.5

3

45 0.0345 0.0292 0.020.5

8 0.033 0.0224 0.01810.5

5

50 0.0411 0.0356 0.02490.6

1 0.0399 0.0276 0.02260.5

7

55 0.048 0.0422 0.03040.6

3 0.0471 0.0333 0.02770.5

9

60 0.0554 0.0494 0.03710.6

7 0.0545 0.0393 0.03320.6

1

65 0.0625 0.0563 0.04370.7

0 0.0619 0.0455 0.03920.6

3

70 0.0697 0.0629 0.05150.7

4 0.0694 0.052 0.04560.6

6

75 0.0756 0.07 0.05810.7

7 0.0767 0.0585 0.05220.6

8

80 0.0819 0.077 0.06580.8

0 0.0838 0.0651 0.05910.7

1

85 0.088 0.0836 0.07320.8

3 0.0906 0.0716 0.06610.7

3

90 0.0935 0.0899 0.08090.8

7 0.0972 0.0779 0.07310.7

5

95 0.0984 0.0956 0.08850.9

0 0.104 0.0842 0.08030.7

7

48


55/111

100 0.104 0.101 0.09770.9

4 0.109 0.0903 0.08730.8

0

105 0.109 0.107 0.1080.9

9 0.115 0.0961 0.09420.8

2

110 0.112 0.112 0.121.0

7 0.12 0.102 0.1010.8

4

115 0.116 0.115 0.1271.0

9 0.125 0.107 0.1080.8

6

120 0.12 0.12 0.1321.1

0 0.13 0.112 0.1150.8

8

125 0.126 0.121 0.1451.1

5 0.134 0.117 0.1210.9

0

130 0.129 0.132 0.1421.1

0 0.137 0.121 0.1250.9

1

135 0.136 0.136 0.1571.1

5 0.141 0.126 0.1310.9

3

140 0.139 0.144 0.16

1.1

5 0.144 0.13 0.137

0.9

5145 0.143 0.142 0.168

1.17 0.147 0.135 0.143

0.97

150 0.149 0.151 0.171.1

4 0.151 0.139 0.1490.9

9

155 0.149 0.152 0.181.2

1 0.154 0.143 0.1541.0

0

160 0.154 0.155 0.1841.1

9 0.156 0.147 0.161.0

3Figure 25: A-Dolly vs. B-Dolly Test Data

The above lateral accelerati

CAPSTONE Final Report, Winter

Documents