Twin Cities Robotics Group - Member Page / Jacob Middleton

Jacob Middleton

Home | Back to Member Index | Back to Robot Index

Jacob Middleton - St. Francis, MN

Merlin

Type: Robotic pet
Purpose: Merlin was my first robot, I built him basically to see if I could design and build a "simple" robot. Although Merlin was not as simple as I had thought he would be, I did however, gain much knowledge about robotics, and a greater respect for the people who build robots, now knowing how hard it is and how much work it takes to design and build a working robot from scratch.
Controller: The controller I use for Merlin is a BotBoard designed by Marvin Green. www.rdrop.com/users/marvin
Software Development Environment/Methods: I use Karl Lunt's Sbasic for the programming language. www.seanet.com/~karllunt
Sensors: For object detection I originally had planned to use Infrared, but after building two different circuits and one kit ( none of which worked very well ) I decided that whiskers would be easier to use for object detection. So Merlin now has two whiskers. Merlin also has three cadmium sulfide photo cells ( photo resistors ) for seeking, tracking, or avoiding light. I am planning to add two more photo resistors, and a LED for light, to the bottom of Merlin so that he can also function as a line tracker, allowing him to follow a black line drawn on white paper.
Actuators: Merlin does not have any arms or grabbers.
Height: Merlin stands about three inches tall. (7.5 cm)
Width: Merlin is five inches wide. (12.5 cm)
Length: Merlin is about five and one half inches long. (14 cm)
Weight: Merlin weighs in at thirteen ounces ( 375g )
Power source: Merlin operates on two different power supplies; The first is a nine-volt battery running through a five-volt regulator for the BotBoard and the electronics. The second power supply consists of four AA batteries for the two servos.
Contruction history:"THUD-CRACK", "THUD-CRACK", "THUD-CRACK", this was the sound of MIT's Attila walking on crushed rocks in a science video I owned called "Space Age". I was fascinated with this part of the video, I watched it over and over again always finding something I had not noticed before or trying to figure out how each part of the legs worked and how I could duplicate it. I even built a mock leg to better understand the mechanics of the legs and how the robot walked.
This was a over a year ago around April '97. After researching some more on walking robots, I decided that I was going to try to design and build my own six-legged walking robot, but about two months into the designing process, I realized that I knew basically nothing about actually building or designing robots, especially walking ones! I thought that this might be a problem :-), so I decided to make sure that I could at least design and build a simple robot before I tried to build a walking one. For the "simple" robot I was going to design and build, I decided on a small wheeled one, and so Merlin began.
The first thing I did in making Merlin was go through about half a dozen different robot designs until I found one that I was completely satisfied with. This robot design used two modified servos ( which are in the front ) with two-inch model airplane wheels attached to them for the drive system. This worked pretty well when I actually built Merlin, the only drawback is that servos are not meant to be used as drive motors, and when they are they use a lot of power for continuous rotation, i.e. Merlin's batteries wear out quickly.
The next thing I did in building Merlin was to construct a BotBoard for the micro controller www.rdrop.com/users/marvin. Although the BotBoard was supposed to be one of the easiest micro controllers to use, it proved to be rather difficult for a first time robot builder like myself to get it to work, and in fact I probably would have never gotten it to work properly if Jeff Sampson from the Twin Cities Robotics group had not stepped in and given me invaluable advice on how to construct it and how to get my programs to work.
At about the same time I was building my BotBoard, I was also constructing Merlin's frame out of one-eighth inch plastic. I had already drawn all sides of the frame on graph paper, so when I actually went to build it I knew exactly what to do. Some of the main features of the frame include: two casters in the back, a cubicle built into the frame for enclosing the servo's power supply, cubicles for inclosing the servos, and a head which I later abandoned because it was not necessary. Note that I mention cubicles, what I mean by this is instead of having a flat base that you place electronics on and build a frame on top of, such as most robots, the way I built Merlin was to build the frame around the electronics and other components. I found that this was much more space efficient, allowing me to build a smaller and more stable robot.
Now I installed most of the electronics, such as the two modified servos, the five-volt regulator, and the completed and now working BotBoard. Merlin could now run forward, but he could not sense anything yet. I then started writing programs in Sbasic to try to get him to do simple things like draw a square with his motions on the floor, or sense when a switch was on or off, things that would seem to be easy, but alas, I could not get Merlin to do much with his servos, and I could not get him to sense anything accurately. Seeing that I needed help from someone that knew what he was doing, I started e-mailing Jeff Sampson who again gave me a lot of advice on my BotBoard and my programs. He also sent me web site addresses for first-time robot builders. After following Jeff's advice I got my BotBoard to accurately sense inputs and Merlin's servos to move the way I wanted.
After rigging up two light sensors the day before and cobbling together a program to use them, I brought Merlin to a robot demonstration which was incorporated into the Robot Zoo at the Science Museum. After this, I stopped working on Merlin for three months ( it had now been about eight months since I had started on Merlin ). When I began working on Merlin again, I permanently installed three photo resistors ( instead of two ), and I wrote a better program to enable Merlin to seek out bright light. The photo resistors did not all have the same sensitivity, that is, in the same light they would give different readings. To solve this problem I installed a variable resistor for each photo resistor so I could manually adjust the sensitivity of each one, instead of trying to change the program to read the inputs differently.
About a month after I installed the photo resistors, I constructed two whiskers for Merlin. The whiskers are made out of guitar string which goes through a one-eighth inch piece of metal tube. The wire is held in place in the tube by glue which fills the bottom half. The whicker sticks straight out of the tube so as to not touch it, and the tube has an electrical wire soldered to the outer side of it which goes to positive voltage. The whisker also has an electrical wire soldered to it which goes to an input pin on the BotBoard. When the whisker is bent to a certain point it touches the top of the metal tube, thus causing the circuit to be completed which applies positive voltage to the input pin on the BotBoard. That's it, the end of building Merlin, unless I find other cool things to add to him :-). If I do though, I will let you know. So long!
Operational description: I have programmed Merlin to operate in two different modes so far:
1. The first mode I will describe is obstacle detection. I will describe a scenario to make it easier to explain... One fine obstacle detecting day, Merlin is running forward when all-of-a-sudden, BAM! Merlin's left whisker runs into an obstacle. Merlin, not wanting a conflict with the obstacle, reverses his right servo causing him to turn in place to the right, away from the obstacle. The obstacle then releases its hold on Merlin's whisker. Detecting this, Merlin resumes his journey forward.
2. The second mode Merlin operates in is light seeking. I will also describe a scenario for light seeking mode... Once upon a light seeking time, Merlin awakens from a deep week-and-a-half, while his master was probably spending all his time playing some computer game to turn him on, sleep. Instantly he compares the left photo resistor with the right. Seeing that there is less resistance coming from the right, Merlin knows that there must also be more light coming from the right. Merlin then compares the center photo resistor with the right, seeing that there is still less resistance coming from the right, Merlin reverses his right servo while driving his left forward. He turns in place to the right while comparing the photo resistors with each other. Finally after turning nearly ninety degrees, Merlin detects that the center photo resistor has less resistance than both the right and left resistors. Merlin then runs his right servo forward ( his left servo having already been going forward ) and starts towards the bright light in front of him.
Future Enhancements/Plans: From now on I will probably just enhance or replace Merlin's program.
Status: Merlin is physically finished, and as for the software Merlin serves as a kind of "test bed" for when I have a "bright idea" for a new program.