Older blog entries for EdwardRupp (starting at number 5)

Wow it's been a long time sense I last wrote an entry. Looks like the last time I was still getting ready for the Trinity Fire Fighting contest that was going to be held locally by Acroname. Well to keep a depressing story short. Fritz my fire fighting robot wasn't as well tested as should have been. I had programed him to avoid walls. As a result he did a great job moving about my house finding a candle and putting it out. However I never tested it in the event arena. I was a little concerned that the much tighter confines would be a problem. It sure was, Fritz had a serious case of "Claustrophobia"! He just couldn't find a way to get away from those bad walls! Next time he will be programed to wall follow. This will probably mean adding a few extra sensors.

The next item is the new robot club in Denver, the Denver Area Robotics Club challenged my club Front Range Robotics to a little inter club contest. Foolishly I accepted to come up with a new contest.

I wanted to create something that was more intellectually challenging and constructive than the traditional head to head combat like stuff such as sumo. I thought also it would be nice to come up with something that would allow for a wide variety of sensors. Most of all I wanted the robots to actually manipulate their environment. Most hobby robots don't ever seem to get past the wander around and avoiding bumping into things level. That's of course a great way to start into robotics but after that's mastered I think a bot should do more. Being most people think of robots as things that should do useful work, why not a good simulation of work for a contest. Thus the "Find a Object" contest.

The short of the idea is there are eight cylindrical objects of varying but well defined characteristics. Of the eight, two are "Bad". The idea is to keep a robot from simplistically sweeping the field and not really knowing what its doing. Thus the total of points for all the "good" objects is equal to the total negative "bad". The objects are made out of PVC pipe. The objects are characterized by height, color, weight and "taste", which is a couple of copper bands with a fixed resister. There are several levels of difficulty from simply having the objects set all around the bot in a ring where the bot simply has to push out the correct ones to the most complex where the objects are clusterd and the bot has to take the objects to a deposit location. If interest check out the rules at the FRR http://groups.yahoo.com/group/FrontRangeRobotics/ web site

My bot to date can find and push out the objects correctly most of the time but is still undergoing lots of work and modifications to do better.

The very latest robot built was for the Critter Crunch, a yearly event held at a science fiction convention in Denver every October. Most of the bots are simply remote controlled stuff meant to push the other guy off an arena. However there is a effort by fellow FRR club member Dennis Clark to get autonomous robots to this event. This year there were four competitors, but maybe next year there will be more.

My bot was a tracked design using IR proximity for sensors. It worked pretty well, however the robots had trouble seeing each other. Lots of this was due to flat black paint on the bots. Probably sonar will be much more popular next year!

Robotics can have such distractions! My major project is still to build 3 identical robots who's ultimate task is to cooperatively collect and sort colored M&M candies, peanut M&M's specifically. Thus I'll be able to state that I eat the labors of my robots! Also along the way, design these robots are do other interesting things such as compete in the Trinity fire fighting contest. However my efforts have been slowed a bit by a I/O board design that proved to have a great deal of noise on it and cross talk on the serial lines.

Then to soothe my great disappointment (it was a heck of a lot of work!) I bought a little toy walking robot bug. My thoughts were ``this looks like a nice quick weekend project!''. How wrong I was.

The toy bot in question was the model I.B.34 Insecto-Bot. A little fellow so ugly it has to be cute. I mistakenly assumed this toy would use two motors on cams to move the legs and thus allow a fairly simple brain transplant to more intelligently control them. Unfortunately it turned out to only have one motor. True to classic cheap toys (it after all, only cost $15) it used a blizzard of gears and a cleaver set of gears that engaged when the motor reversed. Thus it could walk forward quite well and very badly turn left only. Thus to make the little guy move I bought three $20 mini servo motors.

I now have new respect of Dremel style motor tools! I've done a tremendous amount of surgery on the little beast, and I must say its coming out very nicely. Two big problems were of course coming up with motor mounts and connecting rods and cranks to the legs. The second big problem, the battery compartment!

The most evident of the modifications, the original battery compartment only had 3 AAA batteries. For my application I needed to have two independent batteries. This is due to a walking robot having to do a lot more movement with the motors. That uses a lot more draw current, and generates more noise. A wheeled bot, with careful design can get along just fine with one battery pack. Also I needed four batters for each pack because the rechargeables have a lower voltage. Thus the need to stuff EIGHT batteries where only three had lived before! Lots of careful mototool work and cleaver hinge design latter, they fit. Though he now has a bit of a beer belly.

Next I have yet to build is the PC boards to control the little guy. One thing I decided early on was I wanted him to do something more interesting than the usual run around and avoid bumping into things. I found a little recording module at Radio Shack. It can record up to 20 seconds. The thought is to have the bot walk up to something and attempt to circumnavigate it. Then conclude the shape of it, and vocally announce it. I'll give it a simple vocabulary of words and have the processor start the play back and blank the audio for only the words needed. This will make for a stilted speech, but should be useful.

The sound module itself didn't produce enough volume to be heard well but very close up. I happened to have a little 1 watt amplifier kit I had never assembled. It works well with the output from the recorder. And as luck would have it the little speaker in the toy robot (he originally made random sounds)is just the right eight ohms the amp needed. Now the little bot can down right shout!

I hope to have the electronics done in the next few weeks, in-between the usual work on the other robots. Presently my CNC is cutting out a sensor array board for the more complex robots.

The sensor array board idea uses IR and ultra sonic. The ultra sonic section is a Devantech SRF04. Latter designs I hope to build my own ultra sonic hardware directly into the board. The IR section is the new idea. It uses 4 IR sensors and 3 IR LED's Using hardware PWM to adjust the brightness of the LED's. The twin IR sets are pointed 45 degrees from the board, producing a large field of view. The upper IR's are sensitive at 38khz and the lower set at 56.9Khz. One IR LED is centered between the IR sensor sets with the other IR LED's flanking either side of the IR sensors. With this setup it is possible to see seven possible angle hits. Along with the crude changing of light level I hope to use in software, a system where the number of pulses returned verses transmitted is also used to give a ruff calculation of distance. I realize it will be very crud and heavily influenced by the color of objects but should still be useful. If the idea works I'll use a simpler one frequency, IR only setup for the walker.

I hope all of this isn't interpreted a crowing. I like to hear and hope to hear more about what others are doing, and hope my writing is also interesting.

I now have a crude video system for my robots. Found a color CMOS video camera at Best-Buy for $50. It's meant to be a kids toy, however a little hacking and now I can watch what my bots looking at. The nice thing about this design is it has a built in transmitter. The toy has a plastic shell to make it look like a camcorder. Upon opining it up, only 4 Phillips head screws. I found a single board camera similar to those advertised for around $30 to $100. A separate board has a small RF transmitter. The transmitter board is about 2 X 2.125'', the camera is 1.125 X 1.5''. Also the camera has a adjustable lens. The camera has a part number on the board, TJ-194VO. The camera comes with a little receiver that can be switched for 2.4 or 2.4835 Ghz signal from the camera. The receiver has a VF signal and audio RCA jacks for hooking up to a VCR.

I made up a blank SIMM style card to hold the transmitter and camera, this way I can simply pop it on to a buss board on my bots. The board material was the normal epoxy fiber glass for PC boards, only this stock had no copper on it. Drew up a simple layout that has a cutout in the center to accommodate the transmitters RF shielding box. This way the combined boards are thin enough to allow other boards to sit next to it on the SIMM bus. The camera is mounted on top with thin sheet aluminum to allow the mount to be bent, changing the view angle.

The video quality is definitely not as good as the manufacture suggests on the box, however its not too bad for a tiny camera and transmitter combo for the money. Its been very entraining watching this video. In this case the camera was mounted on a walking robot. This definitely can induce motion sickness! A wheeled robot would be a better choose. Also fun to hear the broadcast foot steps!

Here some data from the instruction sheet: Sensor = CMOS 250K pixel Lens View angle = 45 degrees RF transmitter and receiver frequency 2.4 - 2.4835 Ghz Free space operation distance 15 meters Transmitter power 9v

I found it works ok up to 50 feet, even with several walls and machinery in-between the transmitter and receiver. But it definitely varied as the bot moved. As long as you don't expect too much its probably worth the money.

Fun simple and quick project. Now I have to figure out how to have a PC munch on the video data. Then send some hints to the bot about what to do next!

Went to the Critter Crunch this Sunday in Denver. It was a lot of fun. Got to see and get a better fell for how the robots were made and operated. I was surprised that the majority of the robots were remotely controlled by a tether. Very few radios, also surprised to see a couple controlled by IR. There were only a couple of autonomous bots and those were made with Legos!

Us Front Range Robotics guys did our little sumo robot competition demo. This went well though my Cygnus X-3 seem to have troubles with its IR floor sensors and so kept frantically backing up, thinking it was close to the edge. However my Cronos 2 once again dominated the compatition. This was actually a surprise to me as I had expected Dennis Clark's Godzukey to have some new software tricks to get Cronos. However brute traction still won the day!

We also demonstrated our walkers. These were a big hit. People love to see lots of stuff moving about, so nothing like a walker for that! Dennis has two walkers, the newest is also the smallest. The crowd loved it when it walked off the edge of the platform did a somersault landed of all feet and kept skittering on!

I'll have to decide if I should continue work on Cygnus, he is a cute robot, but is so hampered by the chose of motors. It uses motors and gear train that was used to eject CD trays. The motors run too fast and have too little torque. I may try to add a extra gear step, but this may prove too difficult in the present layout. The other thought is make a new version that uses a hobby dual motor gear train.

The Front Range Robotics club meet this Saturday, unfortunately the turnout was light, so my new walker didn't have anyone to compete against. Also there was only one other sumo robot for my sumo's to go against, and that one was still having problems. However it was still enjoyable to see fellow robo folks again and exchange ideas. Next Sunday the 28 is the Critter Crunch in Denver. For more info see http://www.milehicon.org/. Some of us plain to be there to see the robots compete and we will be doing a sumo demo, and show our other robots off. I've never been to one of these events before, from what I've been told the competitions there are mostly remotely controlled robots similar to Robot Wars or Battle Bots in nature. Only these bots are much smaller, 20 and 2 pound classes. This event has been going on even before the more famous shows by about ten years. Should be a lot of fun!

Front Range Robotics meets on the third Saturday of the month at the Coffee Connection, on the corner of Drake and Shields, next the Pulse fitness center in Fort Collins, Colorado. Meeting starts at noon and goes tell we can't talk no more! The email list serve is FrontRangeRobotics@yahoogroups.com.

Also at the meeting I showed my prototype SIMM based motor control board. This board I call the kitchen sink motor control board, because it is one very busy board! It has 2 dual H-bridge controllers, 2 16F84 processors and a 7 channel power transistor array. It worked very well for the demo. I had 4 brush motors, 5 servos, a small fan and a bunch of LED's going through a simple test program. The next version I hope to make double sided as the present single sided board has way too many jumper wires!

I finished converting a hobby worm gear kit with a servo motor and controller. It works well, but could use some more work. The walker is moving very nicely now. Speeds up to a smooth run and turns nicely. If it sees something straight ahead or if the bump switches hit something, it will slowly backup and turn on its axis an then speed up going forward again. I'll definitely have to continue my Idea of multiple processors, because just the code to make it walk uses up all the memory of the 16F84. My plan is a dedicated SIMM style sensor board with its own processor, separate boards for a central processor and modify the existing program for a processor board dedicated to only moving the legs.

The center lifting legs motor is very stressed by this design. To keep the motor from burning out I've cut off the back case of the servo. Glued copper strips to the control chip and to the motor case, then use a very tinny fan to help cool the thing! My plan is to continue to work on a worm gear drive and also some linkage system to help reduce the stress on the system.

I've been observing the behavior of my new walking robot and want to improve it. The design is a classic 3 motor, 6 leg robot.

One motor is used for lifting by moving the center legs. This motor is the one that requires the most power. About half the time it spends trying to hold half of the robot up in the air for that sides corner legs to move forward. This requires a large holding current. It would seem a great solution to use a servo motor that had a worm gear in it. The lovely thing about worm gears is its very hard for loads to transmit back though them. Thus the motor lifts the robot, at the top of the travel it should be able to "relax" and the load will lock the worm gear, thus keeping the side of the robot in the air without further need of power.

I haven't been able to find a ready made solution so I converted a cheep worm gear box motor with a hobby servo just the other weekend. Works pretty well. I used a Tamiya worm gear kit bought at the local hobby store. Gutted a old Futaba servo. Used the pot, controller and the motor. The cheap worm gear motor was very inefficient so did some machining to adapt the Futaba motor, it was worth it. It works pretty well!

The walker design is based on the two center legs being mounted directly to the servo output shaft, no connecting rods. I'm thinking of trying a separate pivot for each leg and a connecting rod system. However it still wouldn't really solve the power lifting problem. This is because I want my robot to have a variable walk that will lift the legs only as far as is necessary to clear the ground for a fast walk. Then using sensors of a still undetermined nature, detect that the ground ruffness needs for a higher leg lift for clearance. Thus slowing the walking speed.

This weekend I finished the designing for a SIMM based motor control board. The board has two 16F84 processors controlling two L293 dual H-bridge chips. It also has a 7 channel Darlington transistor array. Totally this board can control four DC motors, five servo motors and 7 high power outputs, which could also be DC motors or a unipolar stepper. Also the four DC motors could be replace with two bipolar steppers. Lots of capacity! It also takes a long time test! The layout didn't prove as neat as I hoped. This ones defiantly a prototype. The next one will use double sided PC board. I had planed on double sided but ran out of stock and its hard to find .047'' PC board. Everyone seem to sell .062'', which won't fit in a SIMM socket.

X
Share this page