#ping sensor

Meet Arsonbot. 

He just returned home after spending two years buried in boxes in my storage unit.  I figured it was time to bring him home and commence repair work.

Arsonbot is a scratch-built robot I made during college, but never did finish.  His chassis is based on a Tamiya tank tread kit.  The computing power is provided by a BASIC Stamp 2: HomeWork microcontroller, connected to a Pololu H-bridge module for driving the motors.  There is a Parallax PING sensor for measuring distance, attached to a small servo to allow for looking left and right.  The only other sensor in board is a simple bumper to detect collisions.  The rest of the chassis is composed of assorted Erector Set/Meccano parts and some plywood, plus an antenna for decoration.  Oh and on top sits a small fireball cannon.

The part I never finished in the hardware was hooking the fireball cannon to the microcontroller.  It’s based on something I saw in Make: Magazine years ago, designed to launch flash paper/flash cotton by way of a glo-plug (one from a Nitromethane powered radio control car).  The idea is that the fireballs it launches leave no smoke or ash, and disintegrate within a second of leaving the barrel.  Relatively safe, considering its name.

Now that he’s out of storage, I can redesign the electrical system so I don’t have to rely on three separate battery packs.  In general, operational time has been rather limited due to the power consumption of the motors.  Plus, he can’t drive in a straight line; there’s a bit of listing from the tracks being slightly mismatched and the spokes add a significant amount of friction to the entire drive-train.

He’s programmed in PBASIC over a real RS232 serial connection.  Can’t say I ever wrote a good AI for him, just a few test programs for the critical functions as I added each new component.

Here's the Owl V.02: http://pastebin.com/YejZyZ3Z

There are QUITE a few new changes.  First off, you'll notice that PubMain is a bit chubbier.  I moved all of the ping information and the serial terminal access to the Main function.  Instead of cog Servo, it's called GotoAngle.

And then we also have this global variable "Target".  We set it to 0 first, and if Target = Angle, then we add 5 to target.  Angle is another global variable that in the Compass module is set to aziadjust (value of compass reading in degrees).  So if the target was the same value as the angle measured, then we'd increment the target by 5.  So 0 - 5 - 10 - 15.

So here's the GotoAngle function.  The main difference here is the lack of ramping but now the new incorporation of error terms.  So essentially you see that 1520 was the stop pulse width for the rc servo this time around.  And depending on the difference between the target, and actual angle, we'd send a different pulse width.  So the larger the difference, the faster the servo would move, and the smaller the distance, the slower the servo would move.

So this section you can see the ping portion is removed, and we have Angle being set to aziadjust.  It's pretty barebones to be honest.  I also deleted the AziTerm portion of the code because we didn't need it!

So in between working on construction around the house, construction in my sleep, working my day (night) job, and recording all of our podcasts etc, I've also been trying to get in some electronics hackery.  Remember this project idea from two weeks ago?

http://tymkrs.tumblr.com/post/22782566064/project-thoughts-w-the-ping-distance-sensor

I have started work on this with the help of quite a few folks in the Tymkrs IRC.  The first step I already had was knowing that I could get each individual step working on its own.  Then I got each step working on its own but in the same code as the other modules. This is actually a bit more difficult for a newbie than it sounds. I had to figure out where I could cut, how to rearrange code to adjust for the serial terminal, etc.

So so far, each module is working on its own, and the next step is to start getting them to work together.  JohnS_AZ suggested getting the RC servo to calibrate to North.  So that's what I did, made it so that after initiating the modules, the RC servo would calibrate according to the angle of direction (0 being north, 90 being east, 180 being south, and 270 being north) to north.

At this point I'm being suggested to make it so that as the rc servo gets closer and closer to north, it slows down.  That requires a bit of thinking so it might not get done as fast as the last bits have.  And I also have to figure out how the hardware will go together.  Egads!

Alrighty!  So we continue with the OBJ that a lot of people use when playing with the Ping sensor.  This again can be found in the PropTool 1.3 Library folder called Ping.spin.  

We talked yesterday from outa[Pin]~ to the end of cnt2 := cnt.  Now we're onto the microseconds equation:

Microseconds := ( ||(cnt1-cnt2) / (clkfreq / 1_000_000) ) >> 1

This essentially starts by getting the absolute value of the difference of cnt1 and cnt2.  That gives you the number of clock cycles the ping's chirp took.  Now if you want to use microseconds to represent that, you take the number of clock cycles in 1 second (clkfreq) and convert that to number of clock cycles per microsecond.  Divide the two numbers, and through magic, you end up with the microseconds that ping chirp took.  Another way to look at it here:

Then the >>1 divides the whole thing by 2 with bitshifting magic because the time that the ping took represents going to the object and back, but we only need the "To" section to determine distance.

And next!  Conversion equations from microseconds to distances!

So I thought, well since you can see distances from this, maybe I could make a rudimentary photograph?  This would require both up down and left right motion.  I believe Roy has mentioned that Jessica Uelman has already done some form of photography (depth map?) with the ping sensor, so I know it's possible!

But I thought I'd think through the process before doing it.  So off the top of my head, I was thinking I could use the compass module to measure the degree of rotation, and then with each 10degrees, get the Ping distance.  Do a 180 degree sweep on one Y-axis level, then go up.

Do that for like...20 (or more) levels up, and color coordinate according to a gray scale.  So 0-5inches would be light grey whereas 100-127 inches would be black?

The above is an example of Jessica's project showing how the ping sensor can show a rudimentary map of what's around it.  

I'll see what I can cook up :p