Conversion of a Conventional Electric Automobile Into an Unmanned Ground Vehicle (UGV) University, Istanbul, Turkey. University, Istanbul, Turkey

Transcription

1 Proceedings of the 2011 IEEE International Conference on Mechatronics April 13-15, 2011, Istanbul, Turkey Conversion of a Conventional Electric Automobile Into an Unmanned Ground Vehicle (UGV) Volkan Sezer #1, <;agn Dikilita #2, Ziya Erc an #1, Hasan Heceoglu #1, Alper bner #2, Ahmet Ap ak #2 Metin Goka an #1, Ata Mugan #2 #1 Control Engineering Department, Faculty of Electronics Engineering, Istanbul Technical University, Istanbul, Turkey #2 Mechatronics Engineering Faculty of Mechanical Engineering, Istanbul Technical University, Istanbul, Turkey sezervolkan@gmail.com dikilitascagri@itu.edu.tr Abstract- In this study, conversion procedure of a conventional electric automobile into an unmanned ground vehicle (UGV) is illustrated. This conversion process is divided into two main parts as, mechanical and electrical modifications. Interface circuit, interface software, additional power system, selection of the sensors and computer hardware are given in electrical modifications part. Similarly, design of braking and steering system, their computer simulations and strength analysis are given in mechanical modifications part. All these applications are illustrated on a conventional electric vehicle during this study. Keywords-Unmanned Ground Vehicle (UGV), Modification, Sensor, Communication, Interface. I. INTRODUCTION Unmanned ground vehicles (UGV) have been very popular in recent years. Some of the previous practical studies can be seen in [1-3]. Since these vehicles do not have permission to be in traffic, series production of such a vehicle is not possible for now. That's why most of the studies in UGV begin with a conversion of a conventional one into autonomous. Vehicle shown in figure 1 is the base vehicle for our conversion in this study. Its minimal dimensions will help us about our future algorithm tests on road. An internal combustion engine causes vibration to the system because of its working principle. These vibrations may cause extra noise to sensors such as; inertial measurement unit (IMU), laser scanners and etc. Filtering this noise will cause loss of original sensor data. But in a pure electric vehicle vibrations will be lower than combustion engine based vehicle. Since our vehicle does not have a transmission, we do not need to design an additional mechanical system for gear shift. Even though our first aim is to convert the base vehicle into an UGV, we consider some additional features. First of all, human drivability feature is protected. Additional information about this will be given in Mechanical Modifications part. Another feature is drivability by a joystick from outside of the vehicle. So, the vehicle has 3 drive modes: Classic Mode: Drive by human in vehicle. Remote Control Mode: Drive by human outside the vehicle. Autonomous Mode: Drive autonomously to a given desired location. General communication scheme of the vehicle is illustrated in figure 2. II -- RI232 Sluetooth Ik Fig.1 Base Vehicle f._' I I RS232 Base vehicle is a pure electric two-seater vehicle with 6.SkW electric motor and nvi140ah battery pack. More information about the vehicle can be found in [4]. The reasons for choosing it as a base vehicle are illustrated below: c::; "=. ;. ::!!!!IP'_". Fig. 2 Communication Scheme /11/$ IEEE 564

2 Bluetooth protocol is used between PC and joystick. PC sends/receives data to RS232 protocol via radio frequency (RF). There are 2 RF transceivers in system, one is on PC side, and the other is in vehicle. Computers and sensors communicate via CAN and RS232 protocol. Only the 2 main computers are shown in figure 2. RF transceiver electric circuit is illustrated in figure 3. III. ELECTRICAL MODIFICATIONS In this part, electrical modifications for an UGV design are illustrated. A. Communication and Interface Software Our vehicle has an interface software which sends and receives data from vehicle via serial port of the PC, using an RF transceiver module. This program is written using C++ in Visual Studio platform. A screens hot of this interface is illustrated in figure 4. SerialCommunication EJ OesI"edCoorooate IMg Wn * Sc D_-.C' DO.0' VehicieMode RemoleControl Fig.3 RF Transceiver II. MINIMUM DESIGN REQUIREMENTS OF AN AUTONOMOUS VElllCLE There is a minimal set of hardware components that reside on almost all UGV. When designing an UGV test bed, it is important to consider attributes such as cost, size, and weight while ensuring modularity. For example, a designer might feel that integrating GPS onto the processing board would reduce size and weight. This integration, although initially beneficial, stifles upgrades to the GPS and will be useless on test beds operating exclusively in GPS denied areas. The most obvious necessary components of an UGV test bed are: Processing System: The processing system is the central point of device interfacing and is responsible for high level data processing and decision making. This hardware could simply gather and pass data to human operator or act as the brain of a fully autonomous vehicle. Actuator Controller: The actuator controller is responsible for directly or indirectly controlling the platform or component movements. Sensors: Sensors are responsible for measuring some attributes of the world such as position, heading and distance etc. Examples of sensors that provide this type of data include GPS, IMU, range finders and encoders. Communication: Ultimately an UGV will need to communicate with an external device. This may be as simple as sending progress data to a human or as complex as interfacing with a group of UGVs. Typical devices include wireless serial Modems, Ethernet and _ _69 - Fig.4 Screenshot of Interface Software (pc Side) This interface manages the communication protocol and makes the necessary settings for serial port. It also sends the operation modes of the vehicle which are: Driver Mode Remote Control Mode Autonomous Mode Interface sends the joystick data in "remote control mode". The connected joystick is a standard game console from Logitech, which can be seen in figure 2. Without a joystick, acceleration and steering angle commands can be sent manually. In autonomous mode, GPS coordinates of the desired location are sent to vehicle. An emergency button also sends emergency signal to the vehicle, which is interpreted as brake and stop on the vehicle side. Additionally, the interface takes the GPS position data from the vehicle. On vehicle side data from the sensors is gathered and processed by running algorithms on PXI processor. Then processed data is sent to Microautobox (MABX) via CAN protocol. MABX is used for driving the actuators. The screens hot of the interface which is designed for MABX is illustrated in figure

3 ..,-... _ (,0 '-""_ti"".:. =....'i-"""". "'.. i "EE 1 t:;:... ' m :: :j,-"":.., ::: I oci'o 1:9 I I r-- r I r- SONY-XCI- It is a smart Image Processing SX100 camera, which can process images with its own processor and sends processed data via UDPIIP. C. Additional Power System Fig.5 Screenshot of Interface Software (Vehicle Side) B. Sensors and Controllers In an UGV, there should be some sensors for localization and mapping. More information about sensors that are used in intelligent vehicles can be found in [5]. The sensors that we have installed on the vehicle are listed below: Sensor Quantity BrandIModel Type Laser 2 meo-lux Scanner Laser 1 SICK LDR- Scanner LRS-3100 Laser 2 SICK-LMSI51- Scanner Camera 1 SONY-XCI- SX100 Ultrasonic 10 Bannersensor QT50ULB GPS 1 Trimble SPS851&SPS5 51H Digital 1 KVH Compass Azimuth 1000 IMU 1 Crossbow VG700AB-201 Because of additional sensors, actuators and computers on board an extra isolated power supply is needed. When calculating the additional power requirement the worst case (all the sensors, computers and actuators are using maximum power at the same time) is considered. For this aim 4 piece of l2v/50ah lead-acid batteries are used by connecting them serial and paralle1.12v and 24V is enough for all the additional components. Appropriate fuses and connectors are also designed for each new electrical component as a power module. Figure 6 shows the new battery pack and power module. Fig.6 Additional Batteries and Power System D. Vehicle Interface Hardware The most important property of an autonomous vehicle is its drive-by-wire capability. For this aim, an electric circuit is designed and produced. Schematic and PCB layout of the circuit is illustrated in figure 7. These sensor data should be processed for localization and mapping. According to these processed data, trajectory planning and tracking operations should be done. All these tasks need huge amount of processing power. That's why the tasks are distributed between 3 main computers on board. These computers and their duties are given in the table below: Computer Description Duty NI PXI- It has 2.26Ghz Locali zation, 8110RT quad-core mapping and path processor processor and a planning PXI-7954R powerful FPGAmodule FPGAmodule on PXI1000B with several chassis ]fo cards. Dspace It has 800Mhz Path tracking, low Microautobox processor and level control several ]fo (throttle, steering, 07 interfaces. brake), wireless communication Fig.7 Vehicle Interface Circuit Schematic and PCB The aim of this circuit is to switch the acceleration signal to electric motor and brake lamp signals between pedals and low 566

4 level controller (MABX). According to the position of the mode selection switch which determines the mode of the vehicle (autonomous or classic mode), desired electric signals are given via the pedals or MABX. Installed components are illustrated in figure 8. Mode Selection Switch MABX Hardware A. On-Track Tests This part contains a few tests to get to know about vehicle's dynamical behavior. Aim of these tests, are getting the basic information of the vehicle, such as maximum velocity, acceleration and deceleration capacity, minimum turn radius as illustrated in Figure 11, characteristic velocity and dynamical response of vehicle while turning which is shown in Figure 12, lateral and longitudinal forces as shown in Figure l3 and Figure 14. Final aim of those tests is to estimate dynamical and kinematic model of the base vehicle. Some of those parameters are measured as below; Minimum turning radius: 3 meters. Braking distance from initial speed at 30 km/h: l3m. etc. Vehicle Interface Circuit Fig. 8 Vehicle Interface Hardware N. MECHANICAL MODIFICATIONS In order to convert a conventional electric car shown in figure 9, into an autonomous one which can be seen in figure 10, some mechanical modifications must be done as well as electrical modifications. Fig. 11 Geometry of a Turning Vehicle [6] Understeer Path Neutral Stee Path Oversteer Path Disturbance Force at Center of Gravity Fig. 9 An Interior View of the Base Vehicle Fig. 12 Olley's definition for understeer and oversteer [6] Fig. 10 Interior View of the Base Vehicle after Modifications Fig 13 Cornering model with tractive forces [6] At first, mechanical limits of the stock car should be understood well to get design limits. This can be done by a set of tests which are taken on-track and off-the track. 567

5 Fig. 17 GPS Antenna and Digital Compass VertIcal Fig. 14 SAE Vehicle Axis System [7] Once we have this information, we need some other data like required torque in order to turn steering wheel and force requirement to actuate brake pedal. That's why some off-the track tests have been done. B. Off - the Track Tests Torque, required to turn steering wheel is gathered by reverse engineering techniques. That data is gathered by using a torque meter. Dynamometer is used to measure the necessary force amount to be able to push the brake pedal. C. Using Computer Environment in Mechanical Design Process Basic measurements, consisting of simple dimensions of the vehicle help to build a CAD model of the vehicle. Once we have these dimensions accurately and translate those data to CAD environment, any modification to be made can be designed, simulated and get prepared for easy production, using CAD and CAE software. D. Braking Mechanism Design An autonomous car should be able to do what humanbeings act while driving. These actions have to be taken by electro-mechanical systems. The system below is designed to actuate brake pedal of the vehicle. Our design had to be detachable so it could easily be converted into a conventional car when it is in manual mode. Final design is as seen below in Figure 18 and Figure 19. Fig. 18 Brake System Desigo Overview Fig.19 A View of Brake System After Modification Fig. 15: Basic CAD Data of the Base Vehicle Each component's CAD data are gathered precisely to keep the new designs accurate as illustrated below in figure 15, figure 16 and figure 17. This design contains a servo motor which is used to control the brake pedal by a linear guide. Mechanical movement is transferred by a set of gears. E. Steering Mechanism Design A servo motor without a gearbox is used to actuate steering wheel. The reason for using that kind of servo motor is that, again manual mode of our autonomous car. Thus, it is possible to operate steering wheel by hand as in a conventional vehicle. Views of the steering system assembly, in CAD environment and after modification are illustrated in figure 20 and 21 respectively. Fig. 16 Ultrasonic Sensor and UDAR 568

6 After various design try-outs, final design is obtained as in the figure above. Fig. 20 A View of the Steering System Assembly, in CAD Environment Fig. 23 FEA of the Cotter The same procedure is followed to analyze strength of other critical elements taking part in the mechanical design as shown in figure 23 and figure 24. Fig. 21 A View From the Steering System After Modification F. Mechanical Simulation and Strength Analysis of Designed Mechanisms CATIA's DMU Kinematics module is used to simulate dynamical behavior of designed mechanisms. In Figure 22, finite element analysis (FEA) results of the sheet metal, which is used to mount steering motor is shown. Fig. 24 FEA of Pinion Gear of Brake System All these strength analyzes of designed parts are done using ABAQUS. V. CONCLUSIONS Conversion procedure of a conventional automobile into an UGV is illustrated in this work. Conversion is studied in 2 main parts, as electrical and mechanical modifications. In literature, it is not easy to find a paper which gives enough information about autonomous vehicle conversion procedure. This study tries to fill the gap in this area. Initial autonomous applications that we have tested with the vehicle are autonomous parallel parking and obstacle avoidance basically. Global path planning considering non-holonomic constraints, mapping, localization and robust path tracking subjects are being studied using our autonomous vehicle for future applications. ACKNOWLEDGEMENT The authors would like to thank to State Planning Office (DPT) for their support under the Grant Number REFERENCES Fig. 22 Implementation of FEA for Each Designed Part [I] DARPA Urban Challenge Team MIT, Technical Report, [2] Stanford's Robotic Vehicle "Junior:", Interim Report, [3] Team Caltech, Sensing, Navigation and Reasoning Technologies for the DARPA Urban Challenge, [4] [5] Vlacic L., Parent M., Harashima F., "Intelligent Vehicle Technologies, Theory and Applications", Elsevier, 200!. [6] Wong, J.Y., "Theory of Ground Vehicles"; John Wiley & Sons Ioc., 3rd Edition, 2oo!. [7] Gillespie T.D., "Fundamentals of Vehicle Dynamics", SAE Iotemational,