Krill Labs - Spider-3

About:

Spider-3 is oficially described as an Independently Articulated Quad Limbed Reconnaissance Rover. The independent articulation referrers to its ability to individually control any of the four legs which each can extend, retract, steer, and drive. It is classified as a reconrover because it has a 2 Watt video transmitter for broadcasting live images from it’s night vision infra-red camera. And while this robot is controlled primarily from a computer console, it is still being classified under semi-autonomous due to it’s limited decision making. The onboard processor has a large number of both digital and analog inputs which are tied to position sensors, encoders, and IR range finders. These are all used to feed the robot real time data for handling various scenarios like, “He told me to turn there, but this rock is in my way, so I’ll turn here.” or, “There’s something large in front of me so I’m going to stop and wait for further instruction”.

Motor Preformance Testing:

Since I never used these motors for a drive train before, I wasn't really sure what they were capable of much less their average power consumption, so I did several tests before getting to far into this project. I didn't want to build the whole thing only to find out later that these motors can't even move the chasis!

Motor Stats:

single unloaded avg forward: 77.5 mA
single unloaded avg reverse: 83.5 mA

all four level unloaded: 370 mA all four 20deg unloaded: 560 mA
all four level 4.5 lbs: 400 mA all four 20deg 4.5 lbs: 720 mA

The current vs torque analysis was done with a fluke meter placed in series between the drive motors and the power supply. And the 4.5lbs worth of weight was a bunch of dead batteries.

Leg Construction:

      It was really the discovery of the dashboard air-mix motors and their absurdly cheep abundance that defined the overall shape and function of Spider-3. Having a servo to steer and a continuous motor to drive for each wheel is basically what allowed me to have legs instead of the more traditional convention of simply attaching wheels directly to the chassis.
      I went through multiple leg iterations; some fixed, some extendable laterally, some vertically. After tons of failure, I ended up going with a fixed length angle adjusting design. It was the most sturdy given the materials, weight / size constraints, and linear actuator strength. This implementation allows for the main chassis to be raised and lowered uniformly, or tilted asymmetrically for self leveling during a hill climb. Like the steering and drive,each leg angle can be manipulated independently from the rest.
      With all the legs withdrawn, the rover is in it’s most compact form, this is also it’s tallest mode. When all the legs are extended, the robot has a much wider stance and lower center of gravity making it less resistant to toppling over while traversing an incline.

I ended up having to change the tires to Vex Omni-wheels becasue I was getting too much drag when rotating. Also for the leg extensions, they were pretty much necessisary since the tires can't free spin or change speed.

6-16-11 - Drive Modificaiton

On my latest revision I'm using the Andymark 6in plastic onmni wheels and boy are they nice. Strong rugged and very heavy duty, but they better be for $28 a pop. I also redesigned my whole steering assembly to make it less protrusive and sterdier.

Neck Construction:

The neck can raise and lower much like how the legs can retract and extend. In fact it uses the same home-made linear actuator to accomplish this. On top of the neck there are 2 more air-mixmotors, one for x-axis panning, and one for y-axis tilting. The equipment array, or head, at the very top has a laser, an IR camera, a high intensity yellow LED, 2 16-LED-matrix spotlights, a 110 db siren, and a piezo electric indicator buzzer.

Control Boards:

The making of these boards was definitely the most tedious part of the whole project. The 4 channel driver board alone takes about 5 days to make, and I had to make 4 of them! The MainBoard was no walk in the park either. It has a programming section, a power distribution section, a TTL to RS communications section, and the extended IO was a nightmare; 6 22pin DIPS,and after soldering all those DIP seats, I later found that the chips didn’t fit because their body extends a little past the tines, so I had to saw my chips down to size!
Eventually I discovered that despite the small amounts of current drawn from each of the on-board voltage regulators, the Vregs were still getting mad hot, so I ended up using some scrap aluminum I had laying around to fashion heat sinks for all of them.

Communications:

      There were several challenges to overcome in establishing comms between a computer and the main control board. Now if we were just using a wire, like a standard serial cable between the PCs serial port and the robot, then everything is pretty straight forward. Just connect the 2 rs232 ports and swap the Rx/Tx pair. It was the introduction of wireless that made things get a little sticky.
     Back to basics here, the PIC Processor can only output 0-5v for the data exchange, while true RS232 from the PC uses -15 to 15v. The serial port on the computer is smart enough to still interpret the TTL levels and establish comms. But for wireless, we replace the serial wire with 2 transceivers, and unfortunately the transceivers are not as smart as the computer, they want the real RS levels. This is why I had to introduce the MAX232 level converter circuit to the party. But getting this circuit to work was a whole other issue. I've outlined the specifics of integrating a level converter in this link:

Convert TTL to RS using MAX232

      After getting all the communication issues worked out, I now have a 2-way full duplex 8N1-9600bps USART-RS232 450MHz wireless link straight to the robot's main processor. So you can remote into it with HyperTerminal and issue commands via the console.

Batties:

Powering my creations is always an issue. If you want to build a big robot you have to use big motors, but big motors are power hungry so you'll need big batteries, but big batteries are realllly heavy, so you'll need even bigger motors, etc etc. It's an evil cycle. Well I finally broke that cycle by buying lighter more efficent batteries. I got 2 6800mAh Li-Ion power cells from ebay at about $25 each. They are pretty small considering their capacity, and unbelievably light, in fact when you hold one you think to yourself, there is no way this little thing can output 6.8 amps for one hour (or 1 amp for 6.8 hours). Well they don't, haha, but they get close. I found their max instantanious available current is about 800mA, as is evident by the super thin wire it comes with, and the 12v output drops down to 11v after about 1 hour of 400ma usage, but then it will stay there for another 3 hours or so. They take about 8 hours to charge. Their pretty crappy actually, but for the physical size and price their kinda awesome!

Video Transmission:

I debated long and hard over the different video transmitter / receiver pairs out there. There are quite a bit on the market ranging wildly in price. It basically came down to moderately priced versions for around 100 bucks like the ones on eBay, or mega expensive ones in the thousands like on security websites. I ended up getting a 2 watt pair from eBay for $78 It seems to work fine, but man they arne't kidding about the functional distance being line of sight. Through walls it will only go about the length of a house, but out doors it deff goes a few hundred feet. I used a -8db gain patch antenna on the receiver and still didn't notice a substantial difference. The Tx module uses about 500mA at 12v. I connect the Rx module to a USB video input card so I can see the transmission on my computer.

Various Build Stages:

Estimated start of the project: June 2010

	NAME:	REMARK: