Robo-Magellan Robot

Team Members (from Left to Right): Scott Barlow, Tyson Messori, Terry Cooke. Club members Patrick McCarty and William Hsu also helped out with the robot.

Video of Winning run (1:32)

Gold Medal (Received 2/22/2008) - better late then never. The medal contains a microcontroller and LCD that scrolls “RoboGames 2007 - Gold - RoboMagellan - Tyson Messori/Scott Barlow - Spybot”.

June 17, 2007 - Spybot competed in the RoboMagellan event this weekend, held at the International RoboGames event in San Francisco. After two problematic first runs, Spybot completed a nearly perfect last run in a record 1:31 which was substantially lower than the next runner up time of 2:12. Spybot navigated the course using only wheel encoders, a digital compass, and nine sonar sensors, and then homed in on the cone with its camera. The on board ETek GPS was not used because we found it was giving highly inaccurate measurements that constantly drifted, despite indicating a low HDOP (we later found out this may be due to solar flares). The College of Engineering covered the success of the project in an article on the CENG website.

Final Senior Project Report

• Mechanical Construction
• Electrical Construction
• Software
  • ARM Microcontroller (Slave) - GNU Arm C++ program to read sensors and transmit data packets to the mini-ITX computer.
  • Master Control Program - Main C++ program run on the mini-ITX computer under Xubuntu Linux. Handles navigation and obstacle avoidance.
  • GPS Class Library - C++ library used by robot to send commands to the GPS and interpret NMEA responses.
  • GPS WaypointLogger - C# 2.0 Windows GUI app used to test the GPS and log cone locations with a laptop.
• Linux Configuration - All the steps needed to customize a fresh install of Xubuntu.

Browse our Trac website to see the latest code and other files.

This is a senior project for Scott Barlow and Tyson Messori. Autonomous navigation is becoming more important for both military and commercial applications. A competition called Robo-Magellan is now being held twice each year which entails building a robot capable of GPS waypoint navigation, obstacle detection, and color tracking. For our senior project we will be building and programming a robot for the competition held in mid-June of 2007 at Robogames. The robot will use an RC monster truck chassis, with a custom built computer enclosure mounted on top. It will use GPS, camera, compass, encoder, and ultrasonic range finder data to autonomously guide itself along a trail of orange traffic cones. This robot will be completely autonomous, running a course of approximately 500-1000 feet and avoiding obstacles along the way.

Many of the components (chassis, sensors, etc) are from the previous Spybot project.

• Add transistor for the MaxSonar, which runs at 5 volts, since the ARM7 runs at 3.3 volts
• Rear Bumper
• CMUcam3 box
• Power Board
• Mount Wireless Router
• Mount Second Fan

• Bumper/Buttons
• Sonar “Groups”
• Encoder (with SPI)
• GPS interpretation code (Patrick)
  • GPS Class Library (C++) - used by robot to send commands to GPS and interpret NMEA responses
• GPS Driver/Interpreter/Wrapper/Parser
• Compass (SPI)
• ezLCD Driver/Wrapper
• I2C LCD
• Servo Controller Software

March 21, 2007 - After three hours of work the front sensor array mount has been re-made to the proper width (the old one was .25” too thin). Also the front sensor array holes have been enlarged to properly fit the sensors. - Tyson

March 22, 2007 - After four hours of work the front sensor array has been bent and mounted to the new sensor array mount. - Tyson

March 23, 2007 - After seven hours of work, the main chassis is almost complete. All that is left are the front and rear bumpers, along with several small internal motor cover components. - Tyson and Scott

April 10, 2007 - After five hours of work, the front bumper bracket is complete. Also, the chassis transmission gear box has been rebuilt, and modified so that there is less play in the gears. The main spur gear was also replaced with a new one, making the drive train much stronger and better balanced.

April 11, 2007 - Several hours were spent finishing the front bumper.

April 13, 2007 - Work was done to try and finalize which version of Linux will be installed on the mini-itx pc. An effort was made to install Puppy Linux, which is similar to Damn Small Linux. - Scott, Patrick and Tyson

April 15, 2007 - We successfully installed Puppy Linux on the IDE-to-Compact Flash card and set up a VNC server on it. We can now directly access the robot from any wirelessly enabled laptop. Work was continued on the camera turret box and top plate. We also finished building a door to cover the LCDs. - Scott and Tyson

April 18, 2007 - Several hours were spent finishing the camera turret. It was difficult process mounting the turret on a thrust bearing - mostly because the thrust bearing was slightly out of round (it was bought for $5 at a hardware store). - Tyson

April 29, 2007 - After roughly fifteen hours of wiring and coding, the sonar sensors are working as well as the magnetometer.

May 7, 2007 - Patrick posted the first version of the GPS WaypointLogger (v1.0).

May 8, 2007 - 50% of rear steering servo bracket was completed at Terry Cooke's house (Tyson, Terry)

May 9, 2007 - Rear bumper was completed (Tyson)

May 12, 2007 - GPS antenna was spliced back together but still does not work. LCD02 was found to be damaged. 1 Ping sensor was destroyed when short circuited by a loose battery prong. Packet communication from ARM7 board was completed. It was decided that a new GPS (or antenna), 20×4 LCD, and PC power supply will be needed.

May 13, 2007 - Patrick posted a new GPS WaypointLogger (v1.1) as well as an improved version of the SerialHex Tool (v1.1).

May 14, 2007 - A new GPS, PC power supply and LCD04 screen was purchased. Work was done to test serial communication between the micro controller and PC.

May 19, 2007 - We successfully got the communications working completely! We then wrote a simple obstical avoidance program and tested it actually running on Spybot. below is a video of this test.

May 23, 2007 - Patrick posted a new GPS WaypointLogger (v1.2) for the ETEK GPS, as well as an improved version of the SerialHex Tool (v1.2) now capable of transmitting data and configuring options.

May 23, 2007 - Tyson Posted a new video from the point of view of the robot's CMUcam3. The video was taken from a digital camera resting on top of the CMU camera box. The CMUcam's pan servo settings were configured for a smaller pan increment(2). The values that were used were: Pan increment: 2, Pan Range Near: 6, Pan Range Far: 14. These values were found after extensive testing.

June 7, 2007 - Tyson posted video of our test runs, performed just prior to our senior project review.

June 10, 2007 - Patrick posted a new GPS WaypointLogger (v1.3) that makes it easier to collect and see waypoints.

June 17, 2007 - Tyson posted the video of our final WINNING run at the 2007 RoboGames in San Francisco. Our first two runs were problematic, first because there was a simple mistake in the coordinates we entered, and second because the course was sloped, which threw off our compass measurements (which assume the robot sits on a level field). The last run however was nearly perfect and was completed in a record 1:31 which is substantially lower than the next runner up time of 2:15. Spybot navigated the course using only wheel encoders, a digital compass, and some sonar sensors, and then homed in on the cone with its camera. The ETek GPS was not used because we found out that it was giving us highly inaccurate measurements that constantly drifted, despite indicating a low HDOP. In the future the robot would benefit from a better GPS unit and a 6-axis Inertial Navigation Unit to account for slopes.