Twin Cities Robotics Group - Member Page / Jeff Sampson

Jeff Sampson

Home | Back to Member Index | Back to Robot Index

Jeff Sampson - Minneapolis, MN
Web Site: www.pobox.com/~jsampson

Yard Rover

Type: DC motor differential drive ("Big" DC motors!)
Purpose: Outdoors experimental testbed
Controller: Quark V40 MSDOS Single Board Computer from Megatel
Software Development Environment/Methods: Will probably use ASM or Turbo-C under MSDOS
Sensors: None yet
Actuators: None yet
Height, width, length:
Weight:
Power source: 2 - 12V 4AH NI-CAD packs. It has a 5Volt 4Amp and a 12V 10Amp DC-DC convertor.
Speed: Fast enough to be dangerous.
Contruction history: This project started in 1985. I was bored, so I dug out some parts. I had these big DC motors out of 1/2" tape drives. I went down to Northern Hydraulic and bought some 12" pneumatic tires and wheels meant for go karts. These had a hub and bearings and a sprocket for #35 chain. I built the frame out of plywood and 2x4s. The drive train is #25 chain for the primary drive and #35 chain for the secondary drive.
Then it sat around until around 1992. I pulled it out and put my model airplane radio control system on it. I built a custom speed controller with a single series pass transistor. (For each motor.) This gave me variable speed but no reverse. I ran it around my yard and up and down my sidewalk until a jumper lead popped off and fried the drive system. Back into the garage it went.
Then a couple years ago I found out about the Twin Cities Robotics Group. I updated it with two 12V 4Ahr ni-cad packs. A 24V 6Amp AC power supply. (So I can run it when the battery is dead.) I added a built battery charger. It still has a series pass transistor for PWM motor control, but I added a relay to reverse the direction. I added a V40 MSDOS single board computer. It has a JDR plugable prototyping board to do all the interfacing. Then I got side-tracked and started the Mars1 robot.
Operational description: I haven't decided what it will do yet.
Future Enhancements/Plans:
Status: On-hold, waiting for completion of Mars1
Lessons Learned: Eventually you have spent enough just thinking about something and you just have to sit down and build it. This was my first real robot project.
More Info: www.pobox.com/~jsampson/yr

Mars 1

Type: Stepper motor differential drive
Purpose: To someday compete in the "AAAI Life on Mars" contest
Controller: 68HC11E2 - Marvin Green's BOT1 Board
Software Development Environment/Methods: Currently using Karl Lunt's S-BASIC compiler. When I run out of code space, I will switch to ASM or a bigger computer. The digitizer board development is done in 8051 ASM.
Sensors: Uses a Circuit Cellar digitizer board and CCD camera to do laser ranging
Actuators: Custom arm mechanism
Operational description: I saw the results of the 1997 AAAI "Life On Mars" competition. I thought to myself "I can do that well". (I probably even told anbody that would listen. :-) So I started to design my rover entry. I decided to use stepper motors for their inherent postitioning capability. I designed up an arm to grap and lift the balls. I also decided I needed a hopper to hold several balls. I needed a way to locate the balls so I set out to design a laser ranging/vision system.
I was waiting because the rules for AAAI '98 were supposed to be published "just after the first of the year", they weren't. I waited and nothing. I got discouraged and did not get much done. Then we were invited to a robot exhibition at the Science Museum of Minnesota. So I threw everything together and got it done tthe day before the exhibition
A quick hack of software and I put it on the floor. I turned it on... and the motors just buzzed. The arm motor didn't have enough torque to lift the arm. The vision system would return ranging information, but I didn't have software to use it. So it sat on the floor at the exhibition and looked dumb.
Since then I have updated the stepper motor driver circuit and the arm motor driver circuit. Rewired the whole thing. It now zips around the floor. (Even on carpet.)
Now I just have to finish it.
Operational description: I haven't decided what it will do yet.
More Pictures
1 2 3
Future Enhancements/Plans:
Status:
Lessons Learned:
More Info:

Research 1

Type: Research Platform
Purpose: A prototype for Research 2
Controller: MSDOS 386 CPU card for passive backplane
Software Development Environment/Methods: I haven't decided yet. Will probably be a combination of ASM and Borland C.
Sensors: Whatever I can get my hands on.
Actuators: None planned.
Height, width, length: 15-1/2" high and 16" diameter
Weight: Only weighs 30 lbs. so far. :-)
Power source: Two 12V 7AHr Gelcells as a 24V supply.
Speed: I don't know yet.
Construction History: I decided I needed a platform large enough to attached lots of stuff to. I just started bolting together parts from previous robot projects. I decided to use an MSDOS type computer which gives me tons of development platform options. It also gives me virtually unlimited I/O capability.
Operational description: I haven't decided what it will do yet.
Future Enhancements/Plans:To be determined.
Status: On-going project for the foreseeable future.
Lessons Learned: Small diameter wheels only work well on a smooth, hard, flat floor.
More Info: I will add a link to my web page when I start documenting this project.

Research 2

Type: Research Platform
Purpose: A prototype for Research 3
Controller: Probably Pentium laptop and 68HC12
Software Development Environment/Methods: Will probably be a combination of ASM and Visual Basic.
Sensors: Laser vision/ranging
Actuators:
Height, width, length: ??" high and 16" diameter
Weight:
Power source: Two 12V 7AHr Gelcells as a 24V supply.
Speed:
Construction History: I found some nice wheels...
Operational description:
More Pictures
1 2 3 4 5
Future Enhancements/Plans:
Status: Just getting up to speed (2/15/00)
Lessons Learned: This is the first time I used PIC-Servo controller chips. They work beautifully. Large wheels with precision bearings work a lot better than small wheels.
More Info: More Pictures: www.pobox.com/~jsampson/r2

The Earth Explorer

Type: Research Platform
Purpose: Outdoor explorer machine
Controller: IBM clone motherboard and one or more slave processors
Software Development Environment/Methods:
Sensors:
Actuators:
Height, width, length:
Weight:
Power source: Two 12V 17AHr Gelcells as a 24V supply.
Speed:
Construction History: This machine is evolving out of the Yard Rover
Operational description:
More Pictures
1 2 3 4 5 6 7
Future Enhancements/Plans:
Status: Just getting up to speed (9/22/00)
Lessons Learned:
More Info: The Earth Explorer Web Site

Research 3

Type: Research Platform
Purpose: The follow on to Research 2 and Research 3
Controller: Mini RoboMind 68332 controller and a laptop computer
Software Development Environment/Methods: GCC C on Mini RoboMind and Visual Basic on the laptop.
Sensors: Compass, gyroscope, 16 sonar ring
Actuators: None
Height, width, length: 36" high and 18" diameter
Weight: Heavy
Power source: Two 12V 26AHr Gelcells as a 24V supply.
Speed: Way too slow
Construction History: This robot used the drive train from Research 2. I used the Mini RoboMind controller as the low-level controller. I wrote motor control routines to do PID control of the two Pitman motors of the main differential drive. I created a Robot Bus to communicate with the sensor module. The Robot Bus used a multidrop asynchronous serial interface which ran at 38.4Kbaud. It used a simple diode isolation on each slave transmitter. The sensor module had a Gyration gyroscope, a Devantech compass. I had another device on the Robot Bus which was the sonar ring. I used 16 sonar modules that BG Micro was selling as surplus. They were a custom variation of the Sensecomp modules (the company was orignally Polaroid). These modules are limited to a maximum distance of 10 feet. I later added a laser triangulation TV camera system to the Robot Bus, but I didn't get it integrated.
Operational description: The Mini RoboMind as the low level controller coordinated functions at the hardware level. It directly controlled the drive motors. It also polled the Robot Bus interface to get sensor readings from the sensor module. It polled the sonar module to get the range values. Sensor data was compiled and sent to the laptop where it was logged to the harddrive. The laptop would send commands to the low-level controller to initiate moves and would also request status.
Future Enhancements/Plans:Retired
Status: Retired. Some of the parts were recycled on the Red Robot.
Lessons Learned: A serial Robot Bus works well to allow distributed computing. However you have to deal with bandwidth limitations. I was never satisfied with the PID controller code that I wrote. Later when I was writing code for my Atmel XMega motor controller I found that I needed to divide down the final output value to get reasonable control. Otherwise the control is too twitchy. But I haven't gone back to this code to see if it will calm down this motor controller. The sonar ring is a good investment as long as you can get cheap sonar modules.
More Info: www.pobox.com/~jsampson/r3

Red Robot

Type: Two wheel differential drive
Purpose: Outdoor/Indoor robot
Controller: NetBurner MOD5270 for low-level controller and a laptop for high-level controller
Software Development Environment/Methods: NetBurner's C programming package and Visual Basic on laptop
Sensors: Originally it had a gyroscope, 3-axis accelerometer, Devantech compass, GPS. Since I needed some form of proximity detection I add some of the sonars from Research 3 robot. I also added a sensor head on a pan/tilt unit. The sensor head has an AVRCAM, a Devantech sonar, two different range Sharp I/R range sensors. Software for the sonars and sensor head have not been implemented.
Actuators: None
Height, width, length:
Weight:
Power source: Two 12V 15AHr Gelcells as a 24V supply.
Speed: Controlled at 3 feet/sec max.
Construction History: This was my first all metal chassis. It uses angle iron from an old bed frame and is welded together.
Operational description:
Future Enhancements/Plans:Get the sonars and sensor head integrated.
Status: Currently waiting to get its electronics package back from the Black Robot. That electronics package will eventually live on this robot. The Black robot will be the next generation of control architecture.
Lessons Learned: Big motors with built-in gearboxes and encoders are the way to go. Trying to connect all of your sensors to a single processor that is running an RTOS can get messy. Pnuematic tires can be problematic.
More Info: Various pictures until I get a web page created

Black Robot

Type: Two wheel differential drive
Purpose: Indoor/Outdoor robot
Controller: NetBurner MOD5270 for low-level controller and a laptop for high-level controller
Software Development Environment/Methods: NetBurner's C programming package and Visual Basic on laptop
Sensors: Gyroscope, 3-axis accelerometer, Devantech compass, GPS. I added a bumper sensor made from PVC tubing and it has microswitches. I am adding a web camera on the Asus laptop and running RoboRealm software.
Actuators: None yet.
Height, width, length:
Weight:
Power source: Two 12V 12AHr Gelcells as a 24V supply.
Speed: Controlled at 3 feet/sec max.
Construction History: This robot was created for a couple of reasons. I needed a robot that weighed less than 50 lbs to meet RoboMagellan requirements. The other reason was to experiment with less drive train slop. The Red Robot has to much backlash and has a few tolerance problems. This robot has wheels mounted directly to the output shaft of the motor gearbox. Most of the backlash on this design comes from the final output gears in the gearbox and is tolerable for the moment.
Operational description:
Future Enhancements/Plans:I plan to create a next generation electronics package for this robot. It will move the IMU type sensors to a combined module. This generation will go back to a serial Robot Bus. This will eventually use a custom motor controller which will have its own Robor Bus interface.
Status: This is the current active robot project
Lessons Learned:
More Info: