Robotics

Ever since I was a kid I wanted a robot, preferably one which would do my homework.   While the robots I experiment with now are pretty limited in functionality they still let me indulge in my dreams while I learn.

I feel that BEAM (Biological Electrical Aesthetic Mechanical) robots, as created by Mark Tilden, are the best basis for handling the low level functions such as walking, obstacle avoidance, phototrophism, etc., but I feel that uControllers and memory are necessary to create a robot which can perform complex tasks in the real world.   For those of you who don't know, BEAM robots are based around something called a Nervous Network made up of little nerves known as Nvs.     Two of these Nvs make up a Bicore, which is essentially an oscillator.   Think of it this way: If a uController is like a brain then a bicore is like a spinal cord.

Description		Images
StickyBot This is one of the first BEAM robots I built, it is a two motor walker I built several years ago and recently found and brought back to life.   It's battery holders were hanging by wires, one of the hips (servo horns) was broken, and when I powered the poor guy up his brain fried!   I reused almost all the parts that were undamaged, including the IC sockets.   Sadly the circuit I originally used no longer seems to be available so I switched to another master-slave bicore circuit I found which doesn't seem to reverse as well.    I guess that counts as a partial lobotomy.    His name is StickyBot because he gets stuck a lot.
NewBot Here are some pics of my newest robot...   Yeah, it doesn't look much like a robot but this is what it looks like before it goes into final build.   Right now it is being prepared for testing and as soon as I get the body/platform built I will run wires from this circuit to the body and start testing. It is a five motor walking robot based on the schematics I found at  CostaRica Beam  and at Harold's Beam Bugs  pages.   The pictures at Harold's page are excellent as a reference for how he designed the mechanics and geometry, and I plan to base my initial experiments on his ideas and work from there towards my own preference of design.   For now he is named NewBot.
		The nine green wires coming out of the board and not connected to anything are for the sensors and controls, there will be seven sensors: 1 whisker which will make the robot reverse 1 whisker in front of each front leg which will make the robot turn and go around an obstacle 1 whisker in front of each back leg which will make the robot lift it's leg higher to climb over the obstruction 2 light sensors which will make the robot turn as it is backing up.
UPDATE - 6/19/06 - I've been working on the body and legs.    The first pic is of the control PCB which needed to be removed from the servo so I can access the servo motor directly.   As I understand servos, they use 5V to determine position with the servo centering itself when 2.5V is applied.   What I need is to switch polarity of the voltage going to the motor based on which direction I want it to turn, so I just remove the servo control circuitry.    This allows me to use the Nv neuron directly because when it is turned on the output sources current and the input sinks current, whereas when it is turned off the output sinks current and the input sources current.   The other advantage is that by removing the servo's circuitry we remove the isolation and the higher current draw of the motor when a leg is stuck becomes the feedback the Nv uses to dynamically change the timing.
		The second pic is of the servo after it is hacked.   This is a HiTec HS-311 servo which I chose because it was the cheapest one available to me.   Each servo cost about 9 dollars and the 5 servos represent more than 50% of the entire robot cost.    The current draw is 9.5mA at idle and 130mA while turning but I could not find numbers for various load conditions so I am multiplying the 130mA by 10 to get an assumed Imax condition of 1.3A.    Because of this I am planning to experiment with whether the 2 stacked 74AC240 chips can source the demanded current but I am expecting to blow those chips and then look at either using small transistors or a complete H-bridge circuit to supply Iload.
The third pic is of the box I fabricated to hold two servos at the correct angle and distance from one another.   I used 1/64" brass sheet I picked up from a local hobby shop and, for future reference, I used 1/16" diameter brass rod for bracing the skeleton.    Every once in a while in a convenience store or tobacconist's shop you'll run across a small torch that takes butane lighters to run it and costs very little, you'll definately want one of those for this kind of fabrication.    I used that torch, .050 solder, and regular flux to do the fabrication, though in the future I might experiment with plumbing solder since it won't be on circuit boards.
UPDATE - 6/20/06 - The platform is taking shape, the servos are attached to the boxes and to the perfboard which will form part of the skeletal structure of NewBot [EDIT - 7/4/06 I will not be using the perfboard as structural support due to flex and issues of difficulty changing circuitry when it is structural].   Here are some pics:
UPDATE - 7/3/06 - Well NewBot is getting closer, his body is pretty much done in terms of an initial state until testing shows me what I need to change to make him move well.   I decided not to let his circuit board be part of his skeletal structure so that I could switch in different circuit boards without changing his geometry and having to re-square him every time, but this means that he is now well overweight and I am doubting the 74AC240 chips will drive him (though I am having trouble with the circuit so I am not sure yet).
		He weighs just under 1.25 lbs and thats mostly the brass skeleton, so I am first going to experiment with amplifying the 74AC240 outputs with transistors to drive the motors and if that doesn't work robustly I will build 5 H-bridge circuits which I found at Bob Blick's website.   A qoute from his website: "This circuit drives small DC motors up to about 100 watts or 5 amps or 40 volts, whichever comes first."   I'm not 100% sure what he means by this because by my calculations 5A @ 40V is 200W, but I am assuming he means that any combined values of amperage, voltage and wattage is fine as long as none of the individual values exceed the max values.    So I am figuring on an Imax of about 1.3A @ 6V totaling 7.8W, keeping me way under the circuit ratings.   An advantage here is that if a 6V battery pack does not supply enough power I can move into a 12V battery.   At that point I would be using 35% of rated voltage, 15.6% of rated power and 26% of rated amperage.   It might be a bit of overkill but I tend to err on the side of caution.
This definately reenforces what I already knew: I need to learn more about structural engineering and materials properties so I can make accurate assessments of structural needs in advance.   Regardless, its pretty exciting to me because it means I'll learn about another circuit which will eventually help me build larger robots and hopefully bring me one step closer to my ultimate goal of building a ridable robotic spider!