AUTONOMOUS CONTROL SYSTEM WITH DRAG MEASUREMENTCAPABILITY FOR A WIND TUNNEL MODEL VEHICLEShadeed MahmudBachelor of EngineeringElectronic Engineering MajorDepartment of Electronic EngineeringMacquarie UniversityApril 4, 2016Supervisor: Dr. Sammy Diasinos ACKNOWLEDGMENTSI would like to acknowledge and sincerely thank Dr. Sammy Diasinos for hissupport, advice and enthusiasm during my thesis at Macquarie University. STATEMENT OF CANDIDATEI, Shadeed Mahmud, declare that this report, submitted as part of the requirement for the award of Bachelor of Engineering in the Department of ElectronicEngineering, Macquarie University, is entirely my own work unless otherwise referenced or acknowledged. This document has not been submitted for qualificationor assessment an any academic institution.Student’s Name: Shadeed MahmudStudent’s Signature: ShadeedDate: 1 April 2016 ABSTRACTAerodynamics in vehicle design are a crucial aspect of the design as the designwill lead to performance. Often these aerodynamic testing is done in scale modelsin a wind tunnel over a rolling road to simulate moving ground. As the groundunder the model car will be moving and the wind speed will be varying themodel vehicle will tend to not remain in its initial position. Therefore structuressuch as stints and struts are used to hold the model vehicle in place and a forcetransducer attached to the struts measures the aerodynamic drag that the vehicleis experiencing. However in reality, in normal mode of operation the vehiclewill not have these structures around it therefore why should we have such anarrangement during testing? To make the testing more realistic and understandthe complete aerodynamics of the model vehicle, an alternative method to holdthe vehicle in place is proposed in this project. This project is about developingan autonomous control system within the model vehicle that will enable the modelvehicle to hold its position with a high degree of accuracy and also measure thedrag that the model vehicle is experiencing. This way we eliminate any additionalstructures around the vehicle that can affect the aerodynamics of the vehicle. Thisdocument is the progress report for the project. ContentsAcknowledgments iiiAbstract viiTable of Contents ixList of Figures xiList of Tables xiii1 Introduction 11.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Project Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Literature Review 52.1 An Autonomous Control System . . . . . . . . . . . . . . . . . . . . . . . 52.2 Self-Driving Car . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 Mathematics 94 Sensors 114.1 Sensor Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114.2 Sensor Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 Implementing Steering Control 176 Conclusions 21Bibliography 23ix List of Figures1.1 Project Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Project Timeline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.1 How self-driving car see the road [2] . . . . . . . . . . . . . . . . . . . . . . 63.1 Radius of Curvature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94.1 Arrangement of LDRs and Laser Dot . . . . . . . . . . . . . . . . . . . . . 124.2 Initial Position of LDR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134.3 Incremented Position of LDR . . . . . . . . . . . . . . . . . . . . . . . . . 134.4 LDR values corresponding to their position increment . . . . . . . . . . . . 144.5 Ultrasonic Sensor Test Arrangement . . . . . . . . . . . . . . . . . . . . . . 144.6 Percentage Error in Measurement Using Ultrasound . . . . . . . . . . . . . 154.7 Measured Distance vs Actual Distance . . . . . . . . . . . . . . . . . . . . 155.1 The Sensor Setup Plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175.2 The model vehicle maintaining its position on the running ground . . . . . 185.3 The Feedback Computer Interface . . . . . . . . . . . . . . . . . . . . . . . 19xi List of Tablesxiii Chapter 1IntroductionWind tunnel testing to understand the aerodynamics of a vehicle has long been a part ofthe automotive industry. If a vehicle is to move in space, it will have to move through airand therefore it will be subjected to air resistance or drag force. Having a body that isaerodynamic, such that the air resistance faced by the vehicle is reduced to a minimumwhere needed and applied where needed, can drastically improve the performance of avehicle. In order to understand the aerodynamics of a vehicle, often model vehicle aresubjected to wind tunnel testing. The challenge has always been there to make thesetestings as realistic as possible. A rolling road under the vehicle simulates a moving groundthus contributing towards making the test more realistic. However having a rolling groundunder the model vehicle will cause the vehicle to lose its position when subjected to thestrong wind forces. Therefore structures such as stints and struts are used to hold thevehicle in position and measure the drag the vehicle is experiencing [3]. The aim of thisproject is to come up with an alternative method using which the vehicle can maintain itsposition independent of stints or struts thus making the test more realistic. Therefore anautonomous control system running a mathematical algorithm that can control the vehicleand maintain position and measure the drag is suggested as a solution. This document isa progress report of the project.1.1 BackgroundThe development of such a control system has been attempted by a student at MacquarieUniversity in NSW Australia. While a lot of preliminary research and work has beendone by the student the mathematics behind the control system needed to be developedfurther. A control system was developed which was not fully autonomous and was notperforming within the acceptable tolerance region. There was room for improvement inthis project.12 Chapter 1. Introduction1.2 Project OverviewThis section discusses the overview of the project including the particular project specifications and a timeline of completion of expected stages of the project.Project SpecificationThe overall aim of this project is to build an autonomous position maintaining ControlSystem capable of measuring drag for a wind tunnel model vehicle. The model vehicleneeds to be at the center of the moving ground inside the wind tunnel. There is atolerance range of +-5mm making a total of 10mm or 1cm length of total variance thatcan be present. At the beginning the vehicle needs to steer itself and maintain positionat a speed of 5m/s. This is illustrated in figure 1.1 below.Figure 1.1: Project SpecificationsThe project is broken down in several steps in order to establish a systematic approachto the design process. The steps are given below.• Develop algorithm to produce a radius of curvature that the vehicle will turn byfrom input parameters of displacement and angle.• List sensors that can be used to measure displacement and angle.1.2 Project Overview 3• Implement steering control such that the vehicle aligns itself when pointing towardsthe central target line.• Implement steering control such that the vehicle aligns itself when pointing awayfrom the central target line.• Implement throttle control.• Implement total system to operate at slow speed (5m/s).• Implement drag calculation for the model vehicle.• Implement total system to operate at high speed.Project TimelineThe project timeline is best illustrated by the aid of a gantt chart. This is given below infigure 1.2.4 Chapter 1. IntroductionFigure 1.2: Project TimelineChapter 2Literature Review2.1 An Autonomous Control SystemThe focus of this project is the autonomous control system that will allow for the modelvehicle to control itself on the running belt inside the wind tunnel and also allow us tomeasure the drag the vehicle is experiencing. The control system will be implemented ina model car, thus making the car self-driven. This does not mean that we are trying tobuild our own self driven car. The focus is the control system, a control system that willbe able to control the vehicle with a high degree of accuracy and measure the drag thevehicle is experiencing. This gives rise to the question that why do we need to have sucha control system? This is the first step in the design process.In the automobile industry, especially in motor sports such as F1, aerodynamics ofvehicles is important down to the point where meticulous precision in aerodynamic designwill result in the car being faster by crucial seconds thus making a difference in the finalresult of a race or so. In 1972, Colin Chapman showed the way ahead for Formula 1. Thelegendary designer and team boss equipped his Lotus 72 with revolutionary aerodynamics. This resulted in Emerson Fittipaldi in winning the World Championship for Lotus.According to Steven de Groote from F1 Technical, aerodynamics are the most importantfactor in the design of a Formula One car [1]. Now, when determining the aerodynamicsof a vehicle, wind tunnel testing for these vehicles is essential.Wind tunnel testing is done to understand the air flow around the vehicle and hencedetermine the aerodynamic of the vehicle. In understating the aerodynamics of a vehiclethe drag that the vehicle experiences due to high speed air flow is a key quantity. Currentlythe drag is measured in the following steps. The vehicle is placed on a moving groundinside a wind tunnel and as the air flows over the vehicle, the drag force causes the vehicleto move back. Here, structures such as stints and struts are used to hold the vehicle inposition and the force that is applied to hold the vehicle in position is measured and usedto calculate the drag.While this is a method that has been widely used, we would like to propose an alternative way of doing it. We believe that having any other body attached to the bodyof the vehicle will alter the air flow around the vehicle and hence the aerodynamics will56 Chapter 2. Literature Reviewbe affected. Therefore we would like to develop a control system that will autonomouslycontrol the vehicles position on the moving ground inside the wind tunnel and in doingso measure the power going to the drive system and calculate the drag the vehicle isexperiencing.This control system will run an algorithm that will take two parameters as its input.The first being a horizontal displacement from the target line and the second being anangle of deviation from the central line. The algorithm will be capable of using thesetwo parameters to calculate a single radius of curvature and the vehicle will follow thatcurvature path and align itself with the central line. We need sensors that will enable usto sense and input the displacement and angle in real time. Since our control system willbe implemented on a model vehicle therefore we looked at self-driving cars and the set ofsensors such cars use to obtain autonomous driving.2.2 Self-Driving CarSelf-driven car has been an interesting phenomenon in the past years and technologicalgiants and automobile giants such as Google and Mercedes respectively and certain othercompanies have managed to establish such cars [2]. Since our control system will essentially result in a self-driven car therefore we look at existing technologies and how thesecan be used to develop our control system.Figure 2.1: How self-driving car see the road [2]2.2 Self-Driving Car 7From figure 2.1 it can be seen that these car use the following list of sensors.• Camera• Radar• LidarA self-driving car uses these sensors along with Global Positioning System (GPS) tosense its surroundings and plan its path. The GPS is used to navigate from an initiallocation to a final destination and the other set of on-board sensors are used to localizethe vehicle with its surroundings. These sensor data are then fed back to a centralprocessing unit where the data is analysed and instructions are outputted to the vehiclefor autonomous driving. Autonomous manoeuvres such as obstacle detection, obstacleavoidance, maintaining position and adjusting speed are amongst the common tasks thata self-driving vehicle would do. Considering the manoeuvres that a self driving vehiclewould do we began to develop a comprehensive list of sensors that we could use to inputdata to our autonomous control system. This is given in the next chapter.8 Chapter 2. Literature ReviewChapter 3MathematicsThe mathematics behind the control system algorithm is briefly discussed here. When thevehicle is out of alignment and needs to get back in alignment it will follow a path similarto an arc of a circle. In order of determine the curvature of that arc it is essential thatwe determine the radius of curvature of an imaginary circle that will result in the vehiclefollowing the arced path and return to its target line. This is illustrated in figure 3.1below.Figure 3.1: Radius of CurvatureThe inputs to the control system algorithm are two parameters1. The horizontal displacement of the vehicle from the central line.910 Chapter 3. Mathematics2. The angle of deviation from the central line.Taking these inputs and applying the rules from circle theorem geometry and simpletrigonometry we can work out the coordinates of the center of the circle and using thatwe can work out the radius of curvature.The x-coordinate of the center of the circle is given by: x = h tan(2)(3.1) 180 – θTherefore the radius is given by:radius = xtan(θ) + h (3.2)Combining these two equations we can have a single equation that takes the displacement and angle as its input and returns the required radius of curvature. This equationis given below. radius =tan(θ) + h(3.3) h tan(1802-θ)where h is the displacement and θistheangle:Chapter 4SensorsSensors are a key component of this project. We are trying to make this control systemsuch that it maintains the position of the model vehicle with an accuracy of +-5mm.Therefore sensors that can provide us with such accuracy are very essential. As discussedabove the inputs to our autonomous control system algorithm that will determine theradius of curvature that will bring the vehicle back to the central position are a horizontaldisplacement and an angle of deviation. So it is necessary that we find sensors thatcan measure these quantities. In order to measure the horizontal displacement of thevehicle, the approach that Nicholas Todesco had taken was to measure the distance ofthe vehicle from a side wall. The sensors that he had used were IR proximity sensors.While measuring proximity is a good way to go about this challenge but the accuracyof the sensors were an issue. Therefore a research was conducted to find the best suitedsensors for our case. The choice of sensors was bounded by certain attributes. The mainfactor was the accuracy of the sensor. In addition to this, the project has a timeline and abudget, these were constraints that were also incorporated in the choice of sensors. Beforewe head into the sensor selection list there are some additional quantities that need to bemeasured. Quantities such as voltage and current drawn from the battery by the motorare required to calculate the drag that the model vehicle is experiencing.4.1 Sensor SelectionThe sensor selection is best described in the form of a table. When researching sensorsthe following attributes were considered.• Range of sensing capability• Accuracy of sensing• Frequency of sensing• Cost of sensor1112 Chapter 4. Sensors• Lead timeIn correspondence to these desired attributes a table was constructed and this tableis given at the end of this document titled Table of Sensors. From that table the desired sensors and other required components were chosen and a purchase order form wascompleted and authorised by my supervisor for purchase. The purchase order has beenlodged and currently awaiting delivery. This purchase order is also included at the end ofthis document titled purchase order form.4.2 Sensor TestOnce we decided that we are going to use a combination of ultrasound sensors and LDRarray, we needed to test the accuracy of these sensors.LDR Array and Laser Dot CombinationIn testing the LDR, I made a linear translator out of the available equipment in thelaboratory. A set of LDRs were attached side by side and a red laser dot was shoneon one of the LDRs. The linear translator was used to translate the LDR arrangementuntil the laser was shining onto the second LDR. The change in output due to changein resistance of the LDRs upon shining the laser light was recorded corresponding to thelinear translation of the LDRs and a graph was plotted to identify the maximum distanceof translation that can occur before the sensors outputs completely change their values.The arrangement of the equipment is shown in figure 4.1, 4.2 and 4.3 below.Figure 4.1: Arrangement of LDRs and Laser DotThe graph that was obtained is shown in figure 4.4 below and it is seen that as weincrement the position of the LDRs, one sensor value decreases and the other increasesand at an increment of 3mm the value of the second sensor becomes greater that the firstsensor. Hence the difference curve changes quadrant on the plot.4.2 Sensor Test 13Figure 4.2: Initial Position of LDRFigure 4.3: Incremented Position of LDRUltrasonic SensorThe ultrasonic sensor was also tested to determine the accuracy of the sensor. Thedistance to an object from the sensor was measured using both the ultrasonic sensor anda ruler to get the real distance. The distance of the object was varied and recorded. Foreach distance value a hundred distance reading samples were taken and the average wascalculated. Then a graph was plotted to understand the difference in measured valuesand actual values. The equipment arrangement is shown in figure 4.5 below.Two graphs were plotted and are shown in figure 4.6 and 4.7 below. It was seenthat the percentage error between the actual values and the measured values fluctuatedbetween 0 and 0.14From both the graphs it can be seen that the distance the sensor has an optimumperformance range when it is measuring distance greater than 200mm. This is actuallysuitable for our purposes.14 Chapter 4. SensorsFigure 4.4: LDR values corresponding to their position incrementFigure 4.5: Ultrasonic Sensor Test Arrangement4.2 Sensor Test 15Figure 4.6: Percentage Error in Measurement Using UltrasoundFigure 4.7: Measured Distance vs Actual Distance16 Chapter 4. SensorsChapter 5Implementing Steering ControlThis chapter talks about the implementation of steering control. The plan was to measure the distance of the vehicle from a side wall and maintain that distance steeringautonomously. Using ultrasonic sensors this can be achieved but this is not good enoughin terms of our tolerance range. Therefore the idea of using an LDR array in combinationa laser dot is introduced. This setup is illustrated with the aid of figure 5.1 below.Figure 5.1: The Sensor Setup PlanWe would have an array of LDRs and have the laser pointing to the center of thearray. Whenever there is a change in the vehicles’ position the laser will not point to thecenter of the array but on the side or corner. Depending on which side or corner of theLDR array the laser is pointing at the vehicle can be steered so that the laser dot alwayspoint to the center of the array. The LDR array and laser combination was tested and itwas found that a translated movement of 3mm can be detected easily. This is good for us1718 Chapter 5. Implementing Steering Controlas we are trying to reach a tolerance range of +-5mm. The size of the LDR array will bedependant on the accuracy of the ultrasonic sensors. The more accurate the ultrasonicsensor smaller the array and vice versa.Keeping this in mind an attempt to implementing the steering control was made justusing the ultrasonic sensor. The outcome of the attempt was very positive. The vehiclemanaged to maintain its position in the center of the running ground. The system was notfully autonomous as the calibration needed to be done manually but once the calibrationwas done and auto pilot mode was enabled the control system was able to maintain thevehicles’ position autonomously. In order to test the autonomous steering of the controlsystem, whilst the vehicle was in the center of the running ground, I intentionally nudgedthe vehicle in both right and left directions to throw it of its alignment and the vehicle wasable to steer back to the center of the running ground. Figures 5.2 and 5.3 below showthe arrangement of the setup and an early version of the computer interface respectively.The computer interface will be modified later.Figure 5.2: The model vehicle maintaining its position on the running groundTo illustrate the auto steering I have included the link to a video showing the vehiclemoving back to its target position once thrown of it.Link: though the vehicle was moving back to the target line, from visual inspection itcan be seen that at times the vehicle is swaying by more than 5mm. So it was not withinour tolerance range but it was very close to it. Therefore I am optimistic that using theLDR array and laser dot combination we will be able to achieve the level of precision andtolerance we are trying to achieve. These part have been ordered and waiting for delivery.19Figure 5.3: The Feedback Computer Interface20 Chapter 5. Implementing Steering ControlChapter 6ConclusionsIn conclusion, this is a progress report that indicates the progress that I have made halfwayinto the allocated time for this project. In summary so far the progress that I have madeincludes the development of an algorithm that takes two parameters, the displacement anddeviation angle as inputs and outputs a single radius of curvature that steers the vehiclein position. Next I researched sensors that can realise the inputs needed for my controlsystem algorithm and conducted tests to determine the accuracy of the sensors. Oncesatisfied with the sensor list a purchase order was issued and currently awaiting delivery.Further I have started to implement the steering control mechanism and program and hada trial. So far the vehicle is steering autonomously and maintaining position as mentionabove. However it is still not within our acceptable tolerance. So further work needs tobe done in terms of achieving the steering tolerance, measuring drag and throttle controlwithin tolerance.21 2324 Chapter 6. ConclusionsBibliography[1] S. D. Groote, The Importance of Aerodynamics,” F1 Technical, 2006.[2] A. Sage, Where’s the lane? Self-driving car confused by shabby U.S. roadways,” LosAngeles, 2016.[3] N. Todesco, Control System For A Self-Driving Wid Tunnel Model,” 2014.25Table of sensorsUltrasonic Sensors Laser Sensors Current SensorProximity Sensors Accelerometers Voltage Sensor NameRangeAccuracyFrequencyCostLead timeSourceParallaxUltrasonicDistanceSensor2cm –3m11 – 12 %5kHzApproximatelyAU$40.53+AU$10.99(shipping)9 days – 26daysebay –400cm3mm(0.075% –15%)40HzAU$1.903 days – 9daysebay$5.90< 4daysebay –80cm20-40%25HzAU$13.20AU$4.80 +AU$1.99(shipping)< 6 days22 days – 35daysebay$29.95+AU$7AU$29.95+AU$92 – 5working days2 – 3working daysJaycar ParallaxLaser Rangefinder15cm –122cm3% avg5% max1HzUSD 99 +(shipping)Refer tofigure 1below10 days – 3monthsRefer tofigure 1belowRobotShop 9-DOFAbsoluteOrientationIMU FusionBNO055BreakoutSensorModule60 mg100HzAU$46.51+AU$34.36(shipping)9 days – 19daysebay 9-DOFAccel/Mag/Gyro+TempBreakoutBoard –LSM9DS0(ADA:2021)±60mg100kHz –400kHzAU$41.95+ AU$7.95< 5 daysebay VoltageDetector &SensorModule ForArduinoADC /Great forBatteryMonitorDC0-25VMaxresolution=0.00489VXAU$5.759 days – 10daysebay VoltageSensorModuleVoltageDetectorDivider forArduino DGSDC0-25V1%XAU$2.423 weeks – 5weeksebay 1 Pc 5ARangeCurrentSensorModuleACS712 forArduino0-5AOutputerror =1.5%XAU$1.88 +AU$0.263 weeks – 5weeksebay RangeCurrentSensorModuleACS712ACS712ELC-30A ChipModule(Arduino)DC0-30A1.5%XAU$7.87< 7 daysebay CurrentSensorConsumeVoltageLoadDetectionModule HotDC3-25VDC0-3AVoltageresolution=0.00489VCurrentreadingaccuracy= 2%XAU$2.253 weeks – 5weeksebay Purchase Order Form ItemNo.ItemDescriptionPrice perunitNumber ofUnitsTotal1Light dependant resistorsAUD 1.0625AUD 26.522N2222 NPN TransistorAUD 0.4025AUD 103Adafruit 9-DOF AbsoluteOrientation IMU Fusion BNO055Breakout Sensor ModuleAUD 46.512AUD 93.02 + 34.36(shipping)430A Range Current SensorModule ACS712 ACS712ELC-30A Chip Module (Arduino)AUD7.872AUD15.745DC Voltage Detector & SensorModule For Arduino ADC / Greatfor Battery MonitorAUD5.752AUD11.506HiTEC HS-422 ServoAUD 201AUD 20 + 8 (shipping)74051 MultiplexerAUD 0.7243AUD7.24 (10 pack – minorder quantity = 10)81050mAh 2 Cell LiPo batteriesUSD 15.952USD 31.90 + 21.05(shipping)Total:AUD226.36 + USD52.95= AUD296.05 (approx) Student Name:________________________Shadeed MahmudStudent Number:________________________42799627Project Title:________________________Autonomous Control System for a Self-driving Wind Tunnel Model VehicleSupervisor Name:________________________Sammy Diasinos Supervisor Signature:Date:________________________________________________24/03/16 When completed please submit this form to the Laboratory Manager together with supporting quotationinformation and supplier details. Without this information the purchase order cannot be processed.Please note that if the total budget exceeds $300, the approval of the Head of Department is required toauthorise further purchases. This document lacks the signature of the supervisor as we have been maintaining digital meetingrecords in google calendar. The digital records of the meetings followed by the calendar itself areattached below for reference.Consultation Meetings Attendance Form (e-version)CommentsDiscuss Project specifications andaimsDiscuss sensors and testing methodsDiscuss sensors and equipmentbudgetSort out room allocation issue due tolaser operation in F9C111 labSHADEED MAHMUD Mar 2016 (Eastern Time – Melbourne, Sydney)2 8 2 9 1 2 3 4 56 7 8 9 1 0 1 1 1 21 3 1 4 1 5 1 6 1 7 1 8 1 92 0 2 1 2 2 2 3 2 4 2 5 2 62 7 2 8 2 9 3 0 3 1 1 21 2 p m – Weeklym e e t i ng forthesis1 1 a m – Weeklym e e t i ng forthesis1 1 a m – Weeklym e e t i ng forthesis1 1 a m – Weeklym e e t i ng forthesisSun M o n T u e W e d Thu Fri S a t


Leave a Reply

Your email address will not be published. Required fields are marked *