Saturday, January 17, 2015

Attrition: The Unspoken Hobbyweight

Before I get into this long long post, I want to mention the existence of my Facebook page. There, I make smaller posts on more incremental work (which is where I've been more recently).


So if you want to see more consistent updates, I suggest you go there and follow :P
_________________________________________________________________________________

Prelude:
Long long ago, I built a hobbyweight with the help of fellow bot people like Brian Schwartz and Casey Kuhns. This was back in 2008 which for me was high school and my resources were limited so I would like to take yet another moment to thank those mentioned above. However, I did not blog about it because my blog did not yet exist. Later versions of the robot were also gone unmentioned (such as my Motorama '13 and Motorama '14 posts). I find it now necessary to log the development of this power-dense combat robot.

2008-2009:
This robot was a drum spinner (as was the flavor of the month) and was a wonderful blend of scavenged robot parts. It used a set of 5:1 'mabuchi' gearmotors I scored from ebay, a set of BB 12-45 escs I bought off a leaving bot member, and batteries given for doing a favor for another bot member. I bought a 40 mm copier stator to wind the brushless motor and had machining help as mentioned above. Furthermore, I had machining services donated by local companies to make my drum, an odd combination of 4130 for the drum body and A2 tool steel.



Please for heavens sake do not weld those steels together. EVER. Looked okay but was difficult to heat treat you might imagine...



Either way, this was my first big bot I made and it debut at Motorama 2008 where it went 3 wins and 2 losses to Igoo and Surgical Strike.




It had many notable flaws:

  • Banebots escs overtemp shutoff from the high current draw mabuchi gearmotors
  • poor detail mechanical design and material choices allowed for bent frame and lockup
  • weight distribution made for poor inverted driving (and poor ability to self-right)
  • additional precautions to secure connectors also required

 It was also the first robot I built featuring a custom made brushless-in-weapon system. I'll attribute the original idea to Corona and Skittlebaru but none have gone as far to put the esc inside. Will bales "Fluffy" came out later utilizing my old 5" drum idea having everything except the drive motors inside the drum. Lets just say there was a reason I didnt finish that version of Attrition. The things learned from this robot would carry in designs to come. Additional pictures from long long ago can be found here:


2012-2013:
By now I was in college at Georgia Tech. My activities at local events and Dragon Con had interested enough friends to justify the "Georgia Tech Fleet" to drive from Atlanta to Harrisburg for Motorama 2013. I dug Attrition out from storage and decided now was time to address its shortcomings.




I started from scratch. Time to apply new knowledge on mechanical design, materials, and even 3D modeling (since by then I had my SolidWorks professional certification). We would decrease the footprint to save weight and increase the durability of outer armor. For the same goal, we would also downgrade the weapon assembly. The gearmotors would have to go; Instead I picked up four higher reduction planetary units with 'long can 400' motors attached. This new version went by many names such as Tofu or War Machine but decided to keep it as Attrition to indicate its lineage.

Let's do some planetary gearbox math:
A single stage of a planetary gearbox can be calculated by:
(Ring + Center)/Center:1

Pinion :11 T
Sun: 17 T
Ring: 46 T

Stage 1: (46+17)/17:1
Stage 2: (46+11)/11:1

Overall: 19.24:1
This number was good according to my calculations should yield ~10 mph with my long can 400 motors.



To make everything compact, the keyword was 'integration'. I built the gearboxes into the frame rails to save space. The electronics were repackaged into a small cube to reduce wire space needed. I was able to find  the perfect battery to sneak into the remaining space.





Drive test showed the robot was fairly nimble. I dont remember if I had switched to Banebots RS395 motors yet or not. Either way this video showed that I could not use round belts like I usually favor. On high acceleration moves like direction changes, the belts would slip. This made driving with the drum impossible as I later discovered. Later i switched to waterjet machined XL timing pulleys.




The drum was simple; I cut the original drum from Attrition reducing its length from 8" to 5.625". The motor was a bit more challenging. To simplify the design I kept the esc out of the drum. This was a decision made based on 'Cake's matches where the drum would break free and eat its signal or power wires. I also decided to not route wires inside the shaft. This was so I could preload the inner race bearing assemblies in the shaft for higher rigidity and high frequency performance. As such, the wires would exit with a new concept I had involving a large ring bearing on the motor side.









The original concept was to build the internals from several stacks of waterjet material. However the drum was a bit warped so any precision machined parts would not readily fit inside. I ended up wrecking a motor in an initial test and remachined the entire assembly from a single round of aluminum. Some fully assembled pictures.



Now, time for the money shots! The LED vag was a nice touch too...




This version debut at Motorama 2013 where it faced a number of tough opponents including Zandor and the undefeated multi-time champion Cataclysm who won 2nd and 1st respectively. Attrition went 3 wins and 2 losses for 3rd place, going undefeated for 3 matches until he met Cataclysm on Sunday.





Somehow, Attrition came out in good shape. Aside from blemishes on the top and bottom plates, the majority components including the drum were fully operational. The source of my demise was actually lack of spares. After my match with Zandor in the quarter-finals, I was down one drive motor and went into the Semi-finals against cataclysm with only one drive side. This put me at a huge disadvantage and lost decisively. Shortly after, I lost the losers bracket finals to Zandor because of my lack of drive side. Originally the culprit was a broken magnet. The replacement given by another builder ended up failing also because of pinion slip.




2013-2014
Losses at Motorama 2013 were disappointing to say the least. This year's emphasis would be into the drivetrain. I wanted faster and more reliable. Now at MIT, I lost a valuable resource at the Georgia Tech Invention Studio. However my research afforded me new capabilities I would try and leverage when possible. In addition to remaking many outer armor plates, I also swapped the Banebots RS395 motors for the original long can 400 motors and remade the center timing pulleys from 3D printed material. I believed that high starting torque was the culprit for pinion slipping.




Another decal with lewd LED placement was needed.



As fate would have it, I was wrong. At Motorama 2014, Attrition went 3 wins and 1 loss in a round robin format for 2nd place. Agility was useful in my first match against the pneumatic flipper Jumbonator but I lost a drive side again after a win on Upchuck. This was a fatal blow against the strong drive of Isotelus Rex as I could not maneuver for a good hit. The FDM material actually split and rendered the center pulley useless. This was from the high stress concentrations near the 6mm drive shaft to the ~25mm center pulley. This could easily be fixed with a larger diameter intermediate made of metal.




However, another defect surfaced that I had not predicted. The A2 teeth, after years of battle, finally broke during the match against Isotelus. I suspect this was due to fatigue as nothing on Isotelus could have done that. However I note that points were given to him for 'damage' to my robot as the judges pointed out when Attrition had finally spun down after the match. Closer inspection of the welds show little to no penetration into the wall; it was no wonder they had broken.

Present Day:
No place in 2008. 3rd place in 2013. 2nd place in 2014. This year we pull out all the stops. At Motorama 2015 we will go for the prize!

I here conclude this post as a documentation of my first 12-pound combat robot. In another post, I will summarize the efforts to prepare Attrition for Motorama this year which includes professionally made drums and extensive drive modifications.

Video of all Attrition's matches can be found as a playlist on my Youtube. I have placed the direct link below.

Friday, September 5, 2014

Bug Love Robot: MegatRON for Dragon Con

I sense a change in the winds! Seriously, its getting cooler. That means Summer is ending but more importantly signaling the arrival of Dragon Con, my annual pilgrimage to Atlanta, GA to visit friends and fight robots at the Robot Battles and MicroBattles Contests. Last year I brought DDT and Dead Meat (formerly Turboencabulator) but with Dead Meat, well, dead I needed another entrant to take over the main event. Luckily I had built a mega version of my original beetleweight, Ron, and could think of many ways to make him worthy of the stage.
_________________________________________________________________________________

After Motorama
MegatRON was pretty sad offensively  after Moto. In its fight vs Shaka, the dustpan was wrecked. Although Adam helped reweld the pan for the rumble, the pan was once again split open by Hyperactive and . To make matter's worse the saw arm motor smoked when I was using it as a hammer. It had a pretty sad ride back home next to the nearly victorious Uberclocker. Maybe some of that luck will rub off on it next time?



Uh... a bit more than rub off.
_________________________________________________________________________________

The Love Robot
To satisfy the rules of Robot Battles, the robot had a tip speed limit for all spinning weapons. This meant that the saw cutter could no longer be used with its current configuration. Instead I opted for a lifter and clamper mechanism similar to that of Uberclocker but utilizing the same hardpoints on MegatRON and general dustpan shape. After a first-pass concept, I flushed out the below design:



Looks too much like Uberclocker to me although the components are different. This design uses dual Banebots P60 132:1 gearmotors with RS550 motors. At ~10 inches from the pivot point, this configuration should have tons of lifting torque at the tip and more than enough speed to boot.

Additionally, I was feeling extremely lazy at the time. Not only was I taking courses and doing research but I had an industry short course of my own to teach and a conference the following weeks that needed a poster that I had not made yet. In short, I was strapped for time. If I didnt have the gearmotors on hand for the project, It couldnt be done.



This design turned out to be more feasible. Say, what is that beastly box object tied to the main lift pan?

It is a Tonegawa PS-050 super ultra mega cray servo! I found this sifting through old FIRST and decided to give it new life. This guy would make the top clamp arm at over 1200 oz-in of torque!

However, the servo has plastic gears which are still concern for stripping, especially in this high torque application where backdriving is a possibility. Similar to the lifting pan, I will integrate a slip clutch to relieve the servo under torque but also use spring steel strips as the gripper mechanism. My hope is that the flexibility of the steel is enough compliance to prevent damage to the servo.

The real challenge was converting the servo horn into a more useful medium. I had two options. The first involved milling a relief into the face of a shaft to match the servo horn geometry such that the horn could essentially sit in the relief and transfer torque by twisting against its features. The second option involved modifying the horn to essentially emulate these servo spine shafts from Servo City. I decided towards the latter for ease of support. Below, I describe the steps I used to create the part.

First I turned the horn into a round. No particular diameter, but the more nylon meat the better.



Next I counterbored both ends of the ceramic-coated aluminum shafting. One end should press fit the nylon servo spline but the other end diameter mattered less since that feature is only meant to pass the head of the servo horn screw. In this case, I needed a through diameter for the head of an M4 screw.




I pressed them together and drilled a #2 radial set screw for good measure. I could now assemble the servo shaft assembly.


After machining a few thousand standoffs, the weapon comes together with some long tie rods.


The top clamp servo is also mounted with standoffs. Now I can add the clamp clutch and clamp arm!



Pretend-o-bot! Looking good so far!

Very little would change with regards to electronics. I removed one of the 85A bot bitz escs since a single esc could power both lifter motors. To connect them together I utilized what I call the "shotgun connector".



I mean a double barrel bullet connector could be a shotgun right? Id love to see someone scale this into a six-shooter or something...

The servo is beastly enough that it would need far greater current than the 500mA or so from the ESC BEC. However, with a 12V max limit, I would need a method of stepping down the main battery voltage. Luckily, Charles had experiemented with DC-DC converters for herpy nano bike. He simply used a BEC and changed the feedback resistor. This tricks the converter into supplying higher voltage but at the expense of current.

After those changes it was time for a test! Seems to pass.




The robot wheelies very hard on accelerating which makes scooping robots or object very difficult. However should it happen the pan turns out to be incredibly quick! It is a standalone flipper!



Later that evening, Charles finished up his modifications to Uberclocker so we decided to spar.



From the video, you can easily see how detrimental the wheelies are. Most times, I jump right into his grabber. However, should I keep mobile the robot is still very maneuverable and has lots of flipping potential.

Aside from the expected scuffs, the robot took no damage and performed as expected. From this point on, I decided to tidy up the wiring and take the finish shots.





And this concludes the build report for.... what am I naming this thing again!?

Charles and I agreed the robot looked like a bug, with the red grippers being eyes or the spring steel strips being antennae. This immediately turned towards one of our favorite inside jokes, "bug love tree". This phrase was uttered by a professor during a robotics competition using broken English and has been cherished by us ever since only to rival that of "control systems potato". Change tree into robot and ship it. "Bug Love Robot", perfect.

How much love with Bug Love Robot give at Dragon Con? Read the event report to find out!