Sentry Gun!

As the second UR Robotics "Lab Collab" we decided to make a sentry gun. This was in no small part due to the grad students who have been invading our lab and stealing computer mice for the past 2 years. Card access for the floor is global, so there's no real way we can track them or prevent them from entering the space, so there's little we can do to keep them from tromping through our workshops to get to their adjacent space whenever they please, nicking office supplies and using the xbox on nights when nobody's there.

To put it plainly, we hate them. And I think we can all agree the best way to deal with something you hate is to shoot it with a bunch of nerf darts.

This project was loosely based on an existing nerf turret instructible made for the Vulcan chain-fed nerf gun, which was in turn based off another tutorial for a paintball gun turret. However, given that the Vulcan dropped out of circulation a while ago and now goes on Amazon for a whopping $400 (and also given that reviews of the gun were never very good anyway) we figured we'd go with a different model for our sentry.

Obviously electric was a must, and flywheels that we could power up, but we were also attracted to the idea of visual intimidation. So we went ahead with the off-brand AirZone Quickfire Gatling Blaster, which has a great big rotating gatling barrel. Not the best reviewed, mostly due to the darts holding poorly in the gun chain, but since we're planning to build a box for them anyway this wasn't a big issue.

prepped for dissection

We had 4 objectives for this gun:

1. Get Arduino control over the internal motors for the flywheels, gatling barrel, and firing mechanism
2. Up the voltage on them up so the gun is more powerful
3. Set up a camera to do automatic facial tracking
4. Set up the turret to let the gun aim both vertically and horizontally

This first post will focus on the gun internals, while upcoming ones will focus on the software, tracking, and turret mount.

Taking the gun apart ended up being more tricky than we anticipated. The gatling end has a ring that kind of holds stuff together, and even once we took it apart the entire front end of the gun was still adhering a bit. Rather than force it and risk the delicate clockwork around the flywheels being disturbed, we opted to simply dremel off the back of the gun to access the electronics - there was nothing we needed to get to in the front of the gun anyway, so why risk it?

Alex uses a dremel tool to cut the paneling off

Once we got the panel off, you can see all the internal circuitry of the gun, including another box which turned out to be a motor casing.

Once you take that out you can see the wire and switch assembly driven by the trigger that dictates which motors in the gun are on at which times. 

We actually opened up this motor casing, but immediately regretted it - it's simply a bunch of lubricated gearing, which is awfully easy to break and awfully hard to put back together correctly. Luckily we got it to fit back together with some coaxing but after that we just left it aside. 

The trigger assembly was an entertaining puzzle. It turns out there are several inputs to the gun. One is not trigger related, but is actually a switch in the loading mechanism (not shown) which detects whether ammunition has been added. The others are all here in the trigger. When the first switch is released, the barrel starts rotating, whereas when the others are pressed, they begin the flywheels and firing mechanism, respectively. It's a good way of making sure everything starts in the right order without needing any real electronics. 

Of course, we don't really care about how it's done with the trigger, since we just need to access it with an Arduino. So we went ahead and disabled the trigger, porting all the wires out the bottom of the gun with some documentation as to what does what. 

The power is pretty useless since we'll just be attaching it to the Arduino motor drivers we bought a while back, but at least we know it's 9V as a reference. We'll probably be upping it to 12 volts to give the gun a little extra kick and inspire greater terror amongst those pesky grads.

Next up is getting the whole thing running in software. See you then.

OWI Arm revamp

We've started work on revamping an old OWI toy arm here in the UR Robotics club laboratory at the University of Rochester. We got the box of Amazon, thinking it'd be a good example of a robotic appendage at low cost to examine for our own designs.
http://www.amazon.com/OWI-OWI-535-Robotic-Arm-Edge/dp/B0017OFRCY

We'd bought the arm for a teardown project because of its low price ($40) but we were disappointed to find that it really contained no electronics whatsoever. The motors and wires are all rigged up to metal contact switches controlled manually, and there isn't even so much as a resistor in the whole thing. This means that not only is the arm very physically weak, can't be programmed, and quickly drains the batteries if used for too long, it'll also tear apart its own gearing if you run it too far in one direction.

But hey, that's what Arduinos are for!

The first objective was to get the arm off of the manual control and into something we could configure. For the first session we didn't have any appropriate drivers on hand and couldn't figure out some of the more obscure ones lying around the robot lab, so rather than give up entirely and wait a week I figured I'd throw together my own H bridge for demonstration purposes. I found a pretty good demonstration here for building one out of power mosfets, which we happened to have a good pile of on hand (IRF9Z34 PMOS and 1D22AB NMOS to be specific, and 2N4401 NPN BJTs).

The critical thing to remember about H bridges is that they're essentially a pair of CMOS inverters, but controlled in opposite polarity from each other. Thus, when you turn one side positive, the other turns negative, and vice versa. There's a pretty good overview here if the concept seems tricky. However, this whole loopy thing only controlled one motor, which is nice to show off a concept but not so much for being actually able to program the arm.

It could waggle up and down, though!

It was pretty evident at this point that we were going to need a more capable motor driver. I ended up going with a set of Arduino motor shields off of Adafruit that had some impressive current driving capability and could also be stacked, an important feature for getting all 5 of the motors on the arm powered.

Once we got these going it was a piece of cake to rig it up to my Uno and give the arm a fair bit of individual articulation. The code is still pretty simple but can be located here:
https://github.com/URRoboticsClub/OWIArmControl

The real issue now is that the arm still has no feedback control. I'll be tackling that next time with some glued on potentiometers and maybe even a force sensor or two. After that, who knows? Chess?

Whatever. Until the next project!