Under Construction
I have been experimenting with amateur robotics for a long time. My first robot was named "Rob Robot" and was constructed in 1963 for an elementary school science fair. Rob looked like most of the science fiction robots of that time. I improved Rob the next year and won the local science fair.
The idea for building Rob came from an construction article in the March 1962 issue of "Popular Electronics" on building Emily the robot. Rob had and improved two transistor control circuit that allowed it to follow a white line on the floor or follow a light beam. Rob won third place in the state regional contest. I later learned that the judges thought my exhibit was the best one but they did not believe I had done all the work. Well at least I know it was the best.
Rob spent several years as a remote controlled high school basketball mascot. In 1987 Rob was modified to add a block catcher and a light detector. The catcher and light detector allowed him to compete in our robotic club's annual contest. That year the robot had to find a white block in a eight by eight foot black robotic play pen and put the block in a six inch square hole under a light bulb. Rob was able to get one of the four blocks to the hole almost every time.
The next robot built was named "Jr" by my wife. It started out as an early Motorola 6800 training computer that was modified to drive two stepper motors. At first it was just programmed to wonder around and avoid objects using just a front bumper. A robotic hand and a remote control was added.
About this time I read an excellent robotic article titled "A Mobile Cognitive Robot" in the June 1977 issue of BYTE magazine. The article inspired me to start working on an intelligent robot. (Look up the article. It is still relevant)
The 6800 computer was modified to have 2k of memory instead of 128 bytes and four parallel ports were added.
The goal of the new robot was to have all of it's systems controlled by the computer with the remote control only giving the robot commands. The robot was finished in 1981 just before my daughter was born. It was a good thing that it was finished then because life as I knew it changed drastically.
Jr. has a hand that opens, closes, rotates and lifts from the floor to the top of the robot. An attachment for the hand allows Jr. to launch paper airplanes. Another attachment allows him to hold a felt tip pen. A pen holder was also built into the center of the base of Jr. The pens are raised and lowered using the hand lift controls.
Locomotion is provided by two stepper motors. The stepper motors drive two large toothed gears with timing belts. The large gears are the wheels and the belts are the tires. Jr. moves forward, backwards or rotates either direction
Jr.'s program has several modes of operation. In the first mode Jr. is operated using a remote radio link. All of his actions are controlled using a modified calculator that sends hexadecimal commands to the computer. The second mode is a numerical control mode where the hexadecimal commands are entered directly in memory and Jr. executes them in sequence when commanded by the remote to run a stored program. There are looping commands and commands for how long or how far for the different actions. The last mode is a teach mode. In this mode Jr. remembers everything it does and saves the commands in the program storage area. Then Jr. can repeat the actions at the command of the remote control.
Jr. has logged a lot of time visiting local schools, launching paper airplanes at coworkers and making a mandatory appearance at each of my daughters birthday parties. (until she entered middle school)
In 1983 one of my coworkers (Roy Miller) challenged me to build a useful robot that would cut grass. A task he hated to do. This intrigued me again and I started on another robot. For the first year all I did was talk with Roy about the robot during lunch. We discussed how to do the mowing, locomotion, navigation etc.
Then in January 1984 the local home brew computer club reorganized to become a computer experimenter and hobby robotics club. I attended to organizational meeting. This group Triangle Amateur Robotics (TAR) encourage me to continue and they have been a sounding board for my ideas. Back to the lawnmower.
The lawnmower has been a very big project. It is presently in its third platform. The first one was a three wheeled unit with three wheel drive. That platform was unstable and flipped over too easy. The second had four wheel drive with four wheel steering. It had traction problems. Even though it had four wheel drive, uneven ground would lift two of the wheels off of the ground. The third platform (I call it Robotic Lawnmower Test Bed has four wheel drive and four wheel steering with the front two wheels attached to a bar on the front of the chassis that pivots to keep all four wheels on the ground.
The robot moves forward and backwards with it's wheels pointing straight ahead. When it turns it stops and uses a single motor to turn all of the wheels tangent to the center of the robot. Then the robot rotates in place without changing location. I did the steering this way to simplify navigation.
A Toro 12 volt electric push lawnmower deck was found and incorporated into the third platform. It saved a lot of time designing something to cut the grass. Work on this project has been very slow. Typically I work on it a couple of months each year just before the annual club contest. That is unless I am working on a robot to enter in the contest.
At present the lawnmower is controlled by a Radio Shack Color Computer. An interface was built to drive the four drive motors and the steering motor. One of the drive motors has an encoder to determine how far the robot turns or moves. An invertor was designed to provide 120 volt ac to the computer and disk drive.
Most of the work up to this point went fairly fast. Two to three years. Then there was the problem of navigation. I am determined to make the robot self guiding. That meant that the robot has to know where it is and be able to plan the route to take to mow the grass. After several years of experimenting with lasers and beacons a transit was designed for the robot in 1992. The transit turns around and locates infrared beacons. It determines the direction and the computer calculates the distance to them. The present transit is not accurate enough to do the navigation for mowing.
In the future a rf modem is planned and possibly a complete redesign of the chassis (It is too big.). All of the navigation problems could be solved with a differential GPS system but I just can't afford one. The on board computer needs to be upgraded. The 64k Color Computer does not have the trig. functions or the accuracy needed for the navigation calculations. Or the navigation and route planning can be done by a remote computer and commands transmitted to the mower. Then there are the club contests.
In 1987 a tele operated robot was built to evaluate using a video camera to locate the blocks in the club vision contest. A Radio Shack Iron Claw remote control toy was gutted and hard wired to a remote control panel. An old surveillance camera was attached to the top of the robot. The robot is operated remotely while the operator looks at a video monitor. It is quite educational to operate the robot seeing what the robot sees and try to locate blocks, pick them up and put then in a small opening. I determined that it took too much computing power (in 1987) to use the video camera to find the blocks and navigate. However, the robot has seen a lot of action with people operating it at various club functions. For the years we ran the vision contest we let visitors compete against each other using the robot to perform the contest. This robot has outfitted with a new lighter video camera in January 1997. The new camera just needs a single small cable to provide power and carry the video. This is quite and improvement over dragging a coax cable and a grounded drop cord. The robot was upgraded so it could be used by Explorer Post 2350 at the Boy Scout Mall Expo.
In 1988 an old Radio Shack remote control tank was converted into a test bed for evaluating sensors for the vision contest. Using a differential light sensor a Light Tank was developed. The light detector uses two light sensors connected in a half bridge circuit. The detector ignores changes in ambient light (like someone turning on a video camera light) but detects edges of sudden light change (such as the side of a light bulb or the edge of a white block). The first time this detector was demonstrated at our robotics club the light tank immediately charged the president of the club and started chasing him around. He was wearing black pants and the room had white walls. The detector can spot a flashlight at 30 feet and a white block in a dark area at 6 feet. The Atom Ant robot described below uses two of these detectors.
In 1988 another contest robot was designed for the block finding vision contest. He was named "Atom Ant" because it looked like it had two big eyes. This robot was built at the last moment and used a Motorola 68HC11 single board computer. I had the computer at home to learn how to program it for a project at work. The example programs were boring so I bolted the computer to some wheels and a couple of motors and it was a lot more fun to learn to program. Then just a couple of weeks before the contest I realized that this platform might be able to do the second level of the vision contest. In the second level you had to collect all four blocks and put them through the opening.
The robot has one drive motor and one steering motor. Two light detectors are attached to the drive motor portion of the robot so they turn with the drive wheel. One light detector looks for the blocks and the other one looks for the light above the opening. These light detectors were developed using the Light Tank described above. There are two other sensor inputs. One detects a block in the block catcher and the other detects the wall when the robot enters the opening.
Atom Ant was able to do the second level of the contest by locating one block at a time. Capturing it in the wire block catcher and pushing it into the opening. Navigation is simple. Start out turning the detectors left looking for a block. When seen turn on the drive motor. With the drive motor going turn right until the block is not seen then turn left until the block is seen. The robot turns it's head back and forth as it advances. When the robot runs into the block it then starts looking for the light above the opening and repeats the same behavior as when it is looking for a block. When it runs into the wall it backs up and starts looking for another block. After the fourth block is inside the opening the robot backs out and stops.
Well it was able to get all four blocks a couple of times during the contest.
In 1988 and 89 a robot was built to put eye drops in my dogs eyes. My dog's tear glands were damaged due to an eye infection. The vet wanted us to put drops in the dog's eyes every couple of hours during the day. Well that is hard to do when you are away at work for 9 to 10 hours a day. It should be easy to put eye drops in a dog's eyes with a robot. The first problem was that the dog hated to have the drops put in it's eyes. This was solved with a dog biscuit reward for each application. Next a dog biscuit dispenser was designed. The dispenser is operated by a computer. The computer sounds a tone then pushes a dog biscuit out just a little bit. When the dog tries to get the biscuit out of the dispenser it touches a micro switch with it's nose. The micro switch lets the computer know the dog is there and ejects the dog biscuit all the way out. For a while the computer would sound the tone every couple of hours and we would go over and put eye drops in and let the dog get the dog biscuit. Next a frame was built to hold the dog biscuit dispenser up high on a slanted surface. The surface was designed so the dog is standing on it's tip toes and lifting it's head up at a 45 degree angle when it reaches for the dog biscuit. After the dog became accustomed to the frame and the angled surface a Plexiglas top was mounted that wedged the dog's head down when the dog reached for the dog biscuit. As the dog became accustomed to the top pins were added to the sides of the wedge. Slowly more pins were added until the dog's had to wedge its head into the fixture to be able to get the dog biscuit. Two small holes were drilled in the Plexiglas cover right over the location of the dog's eyes when the dog is wedged into the fixture. A eye drop dispenser was designed that squeezed eye drop bottles. Finally the whole thing worked. The system waits two hours then sounds a tone and presents the dog biscuit part of the way out. When the dog squeezes it's head in the fixture and bites at the dog biscuit it's nose hits the micro switch. The eye drop bottles are squeezed releasing the eye drops in the dog's eyes. The dog biscuit is pushed the rest of the way out. It actually worked. The vet was real excited and was setting up a visit for me with the vet school. Then it happened. A scientific breakthrough. A new eye drop that lasts 12 hours. Well it was a terrific educational experience and the dog was fat and happy.
maze rat
Since 1980 members of the Triangle Amateur Robotics Club (TAR) have been giving robotic presentations for the Parents for the Advancement of Gifted Education (PAGE) Saturday morning program. The PAGE program was developed by a group of Raleigh NC parents that wanted their gifted children doing something better than watching TV on Saturday mornings. In the early days we demonstrated our robots to the class and talked about robotics. The students enjoyed the demonstrations, but it was hard to keep the student's attention when we talked. We thought the class would be better if each student could actually built a robot they could take home when the class was over.
We developed a wire control device called the TAR Rat (Rat). The Rat basically was a wooden base with two wheels, a couple of motors and a battery pack that was attached with wires to a box with two switches. The students enjoyed making the Rats and over 100 of them where built in the classes. After the Rats where built the students put their Rat's through several exercises where they simulated robotic tasks. The Rat went through several design iterations. Plans for the latest version.
Photo 1-
Last version of the wire controlled TAR Rat that was used to teach gifted children.Photo 2- Several versions of the TAR Rat including the one on the bottom right that solves mazes.
This section under construction more to come soon!
The TAR Rats took too much out of class preparation time. The Organic Robot presentation was developed to use instead of making the TAR Rats. See The Organic Robot Web Page for the lesson plan. A condensed version of this lesson plan was a feature article in the March, 2001 issue of Circuit Cellar magazine.
Photographs of the first Basic Stomp contest winner. See the contest rules.
The TAR Rover has a group project. Several members of the TAR club and I built a Mars Rover replica in the Ken's Robots workshop. It has been a very big and very interesting project. It is also the first major robot the club has built together. Check out it's web page.
Organic Robot Trainer was first developed to use to demonstrate the flow-charted program the students wrote during the Organic Robot Presentation on an actual robot. The robot's design was changed to make it easier to assemble and is now used with the Duke University TIP program. See link below for a description of the robot and photographs.
Robotics Class. 2003 was the 5th year I taught a robotics class for the Duke University Talent Identification Program (TIP). The classes ran for 3 weeks with the students in class 6 days a week for 6 hours a day (3 hours on Saturdays). During the first week and a half the students are taught how to build amateur robots from scratch by building a robot of their own. During this introduction period each student built a copy of the robot that I designed for the course. With a combination of instruction and learning by doing the students learned how to read mechanical drawings and schematics, use simple tools, and do mechanical layout, sawing, drilling, soldering, wiring and programming. During the last week and a half of the course the students are divided into four teams and each team builds a fire fighting robot with identical materials given to each team. The classes have been a great success and everyone had a great time. Thanks for the help of Bob Hamlin who helps teach the course and Nancy Womack (98 class), Kevin McGowan (99 class), Lori Webber (2000 class), Dennis Chin (2001 class) and Dana Scott (2003 class) the teaching assistants.
KensRobot 
KensRobot 2000/11999 was my first year attending the International Fire Fighting Contest. I was thrilled when my robot "KensRobot" placed 8th in the senior division. The robot used two Parallax Inc. Basic StampII computers networked together for control. It's sensors included two homebrew IR range sensors, two line sensors, two Hamamatsu Corp. UVtron flame sensors, a Precision Navigation Inc. Vector 2X digital compass, two bumper sensors, and a sound sensor. It used two Hitec RCD Inc. RC servos converted for continuous rotation to drive the wheels and a RC servo to turn the sensor head. A fan was used to extinguish the flame. The robot ran in the non dead reckoning mode and was able to return to the starting circle most of the time.
My "KensRobot 2000" was similar to the previous year's robot. It placed 13th at the 2000 contest with a score that was eight times better than the previous year's robot. There was more competition at the 2000 contest with a lot of outstanding robots participating. The robot used one Basic Stamp II and two Basic Stamp SX Parallax Inc. computers. All three computers where networked together. One of the SX computers was connected to three homebrew IR range sensors and two bumpers. This computer accepted commands like "Follow_Right_Wall" or "Forward_to_Wall." The Stamp II did the signal conditioning for the Hamamatsu Corp. UVtron room flame detector. It also measured the frequency of the sound sensor circuit to determine when to start and it operated a Precision Navigation Inc. Vector 2X digital compass. This computer accepted commands like "Turn_North" or "Turn_South_West". The second SX computer was the master. It sent commands to the other two computers and it monitored the homebrew candle light detector, two Sharp GP2D05 IR detectors (used to detect the candleholder) and a floor line sensor circuit. The master computer also controlled the fan. All three computers controlled the two Hitec RCD Inc. converted drive servos using a Scott Edwards Electronics, Inc. Mini SSC II serial servo controller. This robot also ran in the non dead reckoning mode and was able to return to the starting circle all of the time.
"KensRobot 2001" placed third at the 2001 International Fire Fighting Contest with a score that was six times better than the previous year's 13th place score. The robot uses one Basic Stamp II and three Basic Stamp SX Parallax Inc. computers. Three of the Stamps are networked together and one Stamp operates the three high-speed homebrew IR range sensors. The networked locomotion, navigation and sensor Stamps all communicate with each other over a 5 bit parallel buss and one handshake line. The locomotion computer uses the range information form the IR computer and accepts commands like "Follow_Right_Wall" or "Forward_to_Wall" from the navigation computer. The sensor Stamp does the signal conditioning for the Hamamatsu Corp. UVtron room flame detector. It also measures the frequency of the sound sensor circuit to determine when to start and it operates a Precision Navigation Inc. Vector 2X digital compass. The sensor computer accepts commands like "Turn_North" or "Turned_South". The navigation Stamp is the master. It sends commands to the locomotion and sensor Stamps and it monitors the homebrew candle light detector, two Sharp GP2D05 IR detectors (used to detect the candleholder) and a floor line sensor circuit. The navigation computer also controls the fan. The locomotion, navigation and sensor Stamps share one pin on each Stamp to control a Scott Edwards Electronics, Inc. Mini SSC II serial servo controller. The controller is used for the two Hitec RCD Inc. RC servos drive servos using a Scott Edwards Electronics, Inc. Mini SSC II serial servo controller. This robot also ran in the non dead reckoning mode and was able to return to the starting room in the arbitrary starting point mode all of the time.
"KensRobot " placed third in the Expert Division at the 2003 International Fire Fighting Contest. I modified my 1999 contest robot for the new Expert division rules. The robot used two Parallax Inc. Basic StampII computers networked together for control. It's sensors included two homebrew and one Sharp IR range sensors, two line sensors, two Hamamatsu Corp. UVtron flame sensors, two bumper sensors, and a sound sensor. It used two Hitec RCD Inc. RC high-speed metal gear servos converted for continuous rotation to drive the wheels and a RC servo to turn the sensor head. A fan was used to extinguish the flame.
The International Fire Fighting Contest is a great event. It is very exciting to be at one place with hundreds of people that want to talk robots, attend great seminars and actually talk to some of the legends of the amateur robotics community. Hope to see you there next year.
FIRST Robotics: I have been mentoring team 587 from Orange County North Carolina for the 2001, 2002 and 2003 competitions. In 2003 our FIRST team did great at the nationals. They were fourth in their division (all of the teams were divided into four divisions.) For all of the 291 teams that qualified for the nationals our team posted the third highest score for a single match and ended up with the 18 highest average, round score points. Average point's rank teams.
Our team was invited to the nationals because we did so well at the Richmond regional. The team was in 6th place at the end of the qualification rounds. Our alliance lost in the final elimination round due to weak batteries and we ended up the semi-finalist alliance (second place alliance). The team also won a General Motors design award.
I get a lot of e-mail asking where to get information on building amateur robots. I enjoy getting the messages and please keep sending them. But I thought I would add a resource section to this page. The following are good resources that I have found.
"The Robot Builder's Bonanza: 99 Inexpensive Robotics Projects" by Gordon Mccomb, publisher TAB Books Inc No, 2800 ISBN 0-8306-0800-2 This is a very good beginners book with a large reference list appendix. It is also a must for the experienced builder's library.
"Nuts & Volts Magazine" 430 Princeland Ct., Corona CA 91719, http://www.nutsvolts.com/ Good parts and kit resource and it has a regular robotics article and several robotic projects each year.
"ROBOT Science & Technology" magazine, 888 510-7728 http://www.robotmag.com This is one of two amateur robotics magazines I know of.
"ROBOTICS Digest" magazine 719 254-4558, email 102505@compuserve.com This is one of two amateur robotics magazines I know of.
"The Robot Builder", Club newsletter for the Robotics Society of Southern California. It is well worth the membership to this very active club just to get the newsletter. RSSC, PO Box 26044, Santa Ana CA 92799-6044
"CIRCUIT CELLAR INK" magazine 800 269-6301, http://www.circuitcellar.com Good electronics magazine for people that like to build things. They have several good robotics articles a year.
Mondo-tronics Inc., 524 San Anselmo Ave 107-11, San Anselmo CA 94960 (415) 455-9330 e-mail info@mondo.com http://www.robotstore.com is their web page. They where the first robot store and they have the only amateur robot catalog I know of. Lots of kits, books and neat stuff.
KELVIN, 10 Hub Drive, Melville, NY 11747, 800 535-8469, http://www.kelvin.com/ Good parts resource. They have supplies for school science classes.
"Mobile Robots: Inspiration to Implementation", By Anita Flynn and Joseph Jones, publisher AK Peters, Ltd. This is a good book for the serious robot builder. It goes through the steps on how to build two computer controlled robots. Order form AK Peters below.
AK Peters, Ltd. 508 655-9933, http://www.akpeters.com/ Publisher they have several of the best amateur robotics books and they sale the RugWarrior robot.
http://www.frc.ri.cmu.edu/robotics-faq is a great web page for robotics frequently asked questions.
Marlin P. Jones Associates, PO Box 12685, Lake Park, Florida 33403, 800 652-6733, http://www.mpja.com, Good parts and kit resource.
Edmund Scientific, http://www.edsci.com They have optical supplies and science supplies.
Wirz Electronics, http://www.wirz.com/ PIC stuff.
Parallax Inc., http://www.parallaxinc.com Basic Stamp computers, serial displays, robot kits and other neat stuff.
Scott Edwards Electronics, Inc. 520 459-0623 http://www.seetron.com Neat displays and stuff for micros.
Acroname Inc. 303 258-3161, http://www.acroname.com/ Robotic sensors, books and robot kits.
Hobby Masters 919 870-1121, http://www.hobby-masters.com/ Excellent local hobby shop in Raleigh NC. Gives discounts to students and teachers doing robotic projects.
Robot Books, http://www.robotbooks.com/ They have a neat web page and they sale robot kits, books and robot toys.
GoRobotics.net, http://www.GoRobotics.net/ and RobotGeeks.com, http://www.RobotGeeks.com/ two great web pages by a member of our robot club.
MrRobot, http://www.mrrobot.com/ They have a neat web page and they sale robot kits, robot parts, books, compilers and microcontrollers.
H & R 800 848-8001, http://www.herbach.com/ A good source for surplus equipment, motors and sensors.
MCM Electronics 800 765-6960, http://www.mcmelectronics.com/ A good source for electronic and VCR parts, and timing belts.
Jameco Electronics 800 831-4242, http://www.jameco.com/ A good source for electronic components, microcontroller chips, robotic sensors, electronic compass and electronic kits.
I will keep adding more as I have time and find good ones.
Last modified 8/13/03
URL of this page "http://users.aol.com/kensrobots/kensrobots.html" or "http://kensrobots.com"