Mobile Rescue Robot based on the RoboCup Rescue (NIST) Standards

Transcription

1 University of Manitoba Department of Electrical & Computer Engineering ECE 4600 Group Design Project Progress Report Mobile Rescue Robot based on the RoboCup Rescue (NIST) Standards by Group 05 Justin Dyck Josh Greenberg Morgan Sitter Joseph Ethier Craig Nemeth Academic Supervisor(s) Dr. Zahra Moussavi Industry Supervisors Dr. Ahmad Byagowi Younes Sleep Technology Date of Submission January 12, 2015 Copyright 2015 Justin Dyck, Joseph Ethier, Josh Greenberg, Craig Nemeth, Morgan Sitter

2 TABLE OF CONTENTS Table of Contents 1 Introduction Project Progress Chassis (Morgan Sitter) Motors (Morgan Sitter and Joe Ethier) Speed Controllers (Morgan Sitter and Joe Ethier) Batteries (Morgan Sitter and Joe Ethier) Robot Control Unit (Craig Nemeth and Josh Greenberg) Hard Real-Time Tasks (Craig Nemeth) Soft Real-Time Tasks (Josh Greenberg) Operator Station (Justin Dyck) Future Work Chassis (Morgan Sitter) Motors (Morgan Sitter and Joe Ethier) Batteries (Morgan Sitter and Joe Ethier) Hard Real-Time System (Craig Nemeth) Soft Real-Time System (Josh Greenberg) Operator Station (Justin Dyck) Conclusions Appendix A Updated Gantt Chart Appendix B Updated Budget i

3 1 Introduction 1 Introduction Our objective is to design, build and test a remotely operated robot capable of aiding in search and rescue relief efforts. The final product must be capable of traversing rough terrain to locate and identify potential victims. The robot consists of six major subsystems: mechanical design, motor and control, battery and power distribution system, command and data handling, vision system and operator station. Currently we are testing the battery, motors, motor controllers, and control hardware. Software for the operator station is being interfaced with the Linux system, and the Linux system is being transferred to a physical implementation on the UDOO. Finally, the Arduino s scheduling software and remaining device drivers are being developed. The project is approximately three weeks behind schedule due to funding difficulties resulting in a revised budget of $ (see Appendix B). With a minor redesign to reduce costs, we expect to complete the project on time. 2 Project Progress 2.1 Chassis (Morgan Sitter) The robot has been redesigned to weigh 20 kg instead of the original 30 kg estimate. 1/8-inch aluminum will be used instead of 3/16-inch to accommodate the lighter robot components. We have designed six separate aluminum plates which will be bolted together with steel brackets to form an enclosed casing for all components. An opening in the centre of the chassis will allow the manipulator arm to move through the body of the robot. 2.2 Motors (Morgan Sitter and Joe Ethier) The drive motors provide power to both the tracks and the flipper arms. In addition, four G15 Cube servos have been supplied by our advisor for use in the manipulator arm. Four geared motors will be used to drive the tracks and flipper arms. The motors need to handle the worst case scenario 1

4 2 Project Progress of climbing a 45-degree incline. We previously completed a preliminary design and one motor was ordered but later realized that the motor could potentially draw too much current from the battery. Furthermore, adding a gearbox to each motor would increase the weight and complexity of the design. As a replacement, we then found lighter, geared motors that draw less current. We have ordered these motors and they are expected to arrive in two weeks. 2.3 Speed Controllers (Morgan Sitter and Joe Ethier) Each of the four motors requires an electronic speed controller (ESC) to control the the drive speed of the motors. We originally purchased a SyRen-10A ESC rated at 30 V maximum. During testing, this ESC did not work with our 29.4 V peak battery. After contacting the supplier, we discovered the SyRen-10A has an undocumented 28.4 V shield preventing it from operating at the maximum voltage of our battery. The preferred solution was to buy new speed controllers. We were able to get a refund on the previous controller and have ordered two new ESCs (Sabertooth Dual 25A 6V-30V Regenerative Motor Driver). These new ESCs can each drive two motors simultaneously. Finally, a motion control algorithm will be designed by Joe Ethier and Morgan Sitter and passed on to Craig Nemeth for implementation on the Arduino. Currently the algorithm is in the beginning stages of development. 2.4 Batteries (Morgan Sitter and Joe Ethier) The main 29.4 V peak, 12.6 Ah lithium polymer battery is used to power the drive and flipper arm motors. The battery has arrived and has been tested with the first ESC. The secondary battery will be a lithium polymer battery specified at 11.1 V, 3 Ah and a maximum current discharge of 7 A. This is sufficient to power the UDOO and its peripherals. 2

5 2 Project Progress 2.5 Robot Control Unit (Craig Nemeth and Josh Greenberg) We defined tasks the system had to handle and classified them as periodic or sporadic. Next, we decided upon the periods and priorities of these tasks, dividing them into hard and soft real-time. Finally, we researched devices capable of handling these tasks and selected the best candidates based on efficiency, functionality, and cost. The UDOO single-board computer was chosen over microcontrollers and FPGAs as the central processing unit because it contains the low-level interface of a microcontroller, the high processing power of an ARM quad-core processor, and is simpler to implement over an FPGA. We have acquired the UDOO and software development is ongoing. We then designed a lightweight communication protocol to handle the transfer of data throughout the system. This protocol encompasses control and status messages between the operator and the hard real-time system via the soft real-time system. 2.6 Hard Real-Time Tasks (Craig Nemeth) The Arduino Due on the UDOO has been chosen to handle all hard real-time events. The nested vector interrupt controller (NVIC) has been utilized to implement a state machine to handle all tasks required by the system. The hard real-time processing system consists of a device to detect orientation and acceleration of the robot and a device to detect processor failure. The BNO055, an accelerometer, gyroscope and magnetometer (AGM), was selected to to detect orientation and acceleration. This AGM was chosen because it has an on-board processor for data gathering and uses all of the sensors to calibrate each other, providing an accurate and efficient input to the motion control algorithm. The AGM has been acquired and a software library has been developed. The watchdog timer, which is built into the UDOO, was selected to monitor the processors for failure and restart them if necessary. The hard real-time processing system interfaces with the electric drive, soft real-time processing 3

6 2 Project Progress system, and power system. Driver software has been developed and tested for the ESCs. Quadrature encoders, with built-in decoders, have been selected for positional input to the electric drive. These high resolution encoders can determine the direction of rotation and offload some of the processing. Interprocessor serial communication has been tested between the soft and hard real-time systems. Temperature and voltage sensors have been selected to monitor the power system but have yet to be interfaced. 2.7 Soft Real-Time Tasks (Josh Greenberg) Robot Operating System (ROS), a heavily backed open source robotics platform, was chosen to handle the soft real-time tasks since it provides a number of libraries that are directly applicable to the project. Significant literature review has been dedicated to understanding ROS. Among the soft real-time components is the vision system. Two cameras were specified and purchased, and a proof-of-concept video streaming server was implemented in ROS. This work was done on a virtual Linux machine using GStreamer, a multimedia library which simplifies the task of media streaming. In order to reduce processor load, a lightweight operating system was chosen. Ubuntu Core was chosen for its kernel size and minimal footprint (no graphical user-interface). Currently, all the major parts required for the soft real-time tasks have been received and development is ongoing. 2.8 Operator Station (Justin Dyck) The operator station provides an interface to control and monitor the robot remotely. A Windows PC coupled with a gaming joystick serves as the hardware platform. Custom graphical user interface (GUI) software is being developed for the PC allowing the operator to observe on-board cameras, sensors and other indicators, and monitor system health. The Logitech Extreme 3D Pro joystick has been selected, tested, and confirmed to perform all functions required to operate the robot. This device has four analog axes, one hat switch, and 4

7 3 Future Work twelve buttons, all of which have been interfaced with the Simple DirectMedia Layer (SDL) library. The custom GUI software is being developed in C++ with the Qt application framework, allowing for streamlined development and cross-platform compatibility if desired in the future. Subcomponents of the software which are complete and ready for testing include the TCP network socket interface, data packet generators and parsers, joystick handling thread (SDL context), text indicator displays, console display, and various configuration dialogs. 3 Future Work 3.1 Chassis (Morgan Sitter) The chassis design and layout of components still need to be finalized with construction to begin immediately after. 3.2 Motors (Morgan Sitter and Joe Ethier) The motion control algorithms for the drive motors still need to be designed and tested. 3.3 Batteries (Morgan Sitter and Joe Ethier) The main battery will need to be tested with the newly purchased motor controllers. The secondary battery needs to be ordered and tested with the UDOO and its peripherals. A 10 A current limiter circuit will be needed at the power terminals of each motor controller to limit the current drawn by the 25 A ESCs. A capacitor will be added at the terminals of the battery to handle burst currents if all drive and flipper arm motors are activated. 3.4 Hard Real-Time System (Craig Nemeth) Future work on the hard real-time processing system will involve the development of a real-time operating system, which is currently in progress. The NVIC state machine is a contingency in case 5

8 4 Conclusions the operating system fails to materialize. Device software remaining includes the drivers for the quadrature encoders, temperature sensors, voltage sensors, and watchdog timer. The encoders are the only devices yet to be acquired. They have been ordered and are expected to arrive next week. 3.5 Soft Real-Time System (Josh Greenberg) The UDOO must be loaded with a pre-configured image of the operating system and tested to ensure all hardware integration works as expected. The Wi-Fi module must be configured as an access point and tested. Additionally, both camera streams need to be tested and fine-tuned for quality and reliability. ROS must be configured to route messages and data between the operator station and the hard real-time system. 3.6 Operator Station (Justin Dyck) Subcomponents which are incomplete or currently in development include the UDP network interface, video data parser and displays, graphical indicator displays, supplementary control panel, and 3D visualization implementation. 4 Conclusions Apart from the electric drive, other systems have not experienced major technical issues during development. Despite the delay incurred by funding gaps, we still plan to complete the project with our revised budget of $ by the expected due date. 6

9 Appendix A Updated Gantt Chart Figure A.1 shows a Gantt chart depicting the project timeline. Our original Gantt chart has been updated to indicate our progress on certain tasks. Green indicates completion, red indicates delay and blue indicates future work. Currently, we are about three weeks behind schedule due to delays in funding. Software development is nearing completion and the majority of parts have been received or ordered. In order to save time, we will be combining the component and system assembly phases of the project. 7

10 Component Task Name Duration Phase 1 Research, Design & Procurement 130 days August 1 September 1 October 1 November 1 December 1 January 1 February 1 March 1 7/20 8/3 8/17 8/31 9/14 9/28 10/12 10/26 11/9 11/23 12/7 12/21 1/4 1/18 2/1 2/15 3/1 3/15 12/20 N/A RoboCup competition research / 7 days requirements gathering ALL Design robot architecture 14 days CHASSIS Design robot chassis 10 days POWER Design power system and drive mechanics 18 days CMD Design and develop command & control 99 days system VISION Design and develop vision system 99 days OPS Design and develop operator interface 99 days ALL Order/Manufacture Long Lead-Time Parts 14 days ALL Order/Manufacture Short Lead-Time 30 days Parts Phase 2 Component Assembly & Testing 18 days 1/7 CHASSIS Assemble chassis and track system 3 days CHASSIS Test chassis durability and mechanics 2 days POWER Assemble drive-train and motor/battery 7 days system POWER Test power system and drive-train 6 days OPS Test operator station functionality 18 days VISION Assemble main and secondary camera 4 days modules VISION Test vision system 7 days CMD Interface command & control system with3 days other modules and sensors CMD Test command & control system 3 days Phase 3 System Assembly & Testing 25 days 2/1 POWER Install power system and drive-train onto 5 days chassis POWER Test chassis/drive-train/power system 5 days CMD Install command & control system onto 3 days chassis VISION Install vision system onto chassis 3 days CMD Test robot-operator communications 3 days OPS Test control interface and functions 5 days ALL Adjust design parameters based on 4 days system tests Phase 4 Live Testing & Fine-Tuning 14 days 2/14 N/A Create mockup arena 3 days N/A Test different mission scenarios 6 days ALL Adjust control parameters as needed 4 days N/A Final project sign-off 0 days Fig. A.1: GANTT Chart for the mobile rescue robot design project. 8

11 Appendix B Updated Budget Table I provides a breakdown of our updated budget. Our initial budget proposed two different design options based on how much funding we could receive. The first option involved machining the chassis ourselves and the second option involved purchasing a pre-built chassis, alleviating us of much of the mechanical design work. The two design options were projected to cost $3575 and $7990 respectively. After an unsuccessful funding campaign and further design iterations, the budget was reduced to $ $500 is covered by the Department of Electrical and Computer Engineering and $200 is covered by Fort Garry Fire Trucks. The remaining cost of $ will be absorbed by the team members or by other organizations should we receive funding in the future. Most of the parts for the project have already been received or are on order. The parts that still need to be ordered include the tracks, track wheels, wheel bearings, flipper arms, flipper arm wheels, secondary battery and charger, sensors, connectors, and LED lights. We expect to have these remaining parts ordered by January 16, 2015 and delivered by the end of January. 9

12 Table I: Project Budget Item Part Number Supplier Unit Cost QTY Total Chassis Aluminium Plate N/A Speciallaser Tech $ $95.00 Robot Tracks TBD TBD $ $ Wheels TBD TBD $ $ Wheel Bearing Sets TBD TBD $ $40.00 Flipper Arm Rods TBD TBD $ $40.00 Flipper Arm Wheels TBD TBD $ $15.00 Motors, Controllers and Batteries 24V LH Drive Motor AME RobotMarketPlace $ $ V RH Drive Motor AME RobotMarketPlace $ $71.00 Dual Motor Driver RB-Dim-23 RobotShop $ $ Rotary Encoders RB-Sbo-31 RobotShop $ $78.12 Cube Servo Motor G15 Supplied by ECE $ $ V 12.6Ah Battery PL-25.9V12.6Ah Battery Space $ $ V Battery Charger CH-L25928 Battery Space $ $ V 3.65Ah Battery TBD TBD $ $ V Battery Charger TBD TBD $ $20.00 Robot Control Unit UDOO Board UDOO Quad Supplied by ECE $ $0.00 AGM BNO055 Supplied by ECE $ $0.00 Various Sensors N/A N/A $50.00 N/A $50.00 Wires/Connectors N/A N/A $ N/A $ Primary Camera Logitech C525 BestBuy $ $70.00 Secondary Camera PS3 Eye Supplied by Team $ $0.00 LED Lights TBD TBD $ $20.00 Operator Station Operator Laptop N/A Supplied by Team N/A 1 N/A Joystick Amazon $ $33.00 Subtotal: $ Shipping, Taxes and Duties: $ Total: $