20 April 2014

MIT MechE deFlorez Competition Entry: Underactuated Robot Gripper

An Open-Source Low-Cost High-Strength Rapid-Prototypeable Underactuated Robot Gripper
by Daniel J. Gonzalez - dgonz@mit.edu
MIT MechE deFlorez Competition Entry
(Hi, Professor Hover!)

 Here is a solid model of my gripper. All of the components can be readily found on McMaster, Trossenrobotics, and other online retailers for under $500.00 total. The parts can be fabricated using only a waterjet, or using online waterjetting services such as bigbluesaw.com. With the solid model released as open-source, just about anyone can put this thing together for use in their robot manipulation project: no expert machining experience required!

The design of this gripper is inspired by the WillowGarage Velo Gripper (Work by Matei Ciocarlie): 

and the design optimization work of Prof. Aaron Dollar of Yale and Prof. Robert Howe of Harvard:

(link to paper: http://biorobotics.harvard.edu/pubs/2010/journal/Dollar_IJRR2010.pdf). 

The design of the drive mechanism is underactuated. Specifically, one motor drives both fingers, each of which has 3 DOFs. Passive compliance between each DOF allows the finger to envelop any given object that fits within its grasp without having to perform complicated grasp planning calculations in realtime. 


As of this writing, the fabrication process is nearly complete, and it should be 100% functional by the middle of next week (~4/23/2014). An Arduino will receive a grasp command from a host computer (or a button for demonstration) and command the servos to either grasp or let go of an object. The Dynamixel servo is torque-controlled and provides torque feedback, so we can apply the proper amount of pressure for a given object.
Here is an example demonstrating the underactuated closing behavior of the gripper that allows for asymmetric grasps. Using constraint tendons, the gripper acts as a standard parallel gripper until the proximal link encounters resistance. The two distal links then close around the object for a snug grasp. 
Here is a cutaway showing the direct-drive spool of the Dynamixel MX64 servo (which can output 7+Nm of torque), the cam lock mechanism actuated by an EMax digital hobby servo, and the main drive cable (Yellow). 

Here is a detailed view of the various active and passive tendons and other components within one of the fingers. Each finger is identical. 

Thank you for considering my project in the MIT MechE deFlorez Competition! 
-Daniel

20 January 2014

EtekChopper: Tearing 'er Down and Cleaning 'er Up. Also Battery Mounting & Paint Thoughts...

I've left EtekChopper outside in the rain and cold for the past few months while I was busy doing Senior things like classes and thesis and grad school apps and stuff. Poor EtekChopper! It was under a tarp, but still! Now that some of my own hardware has started to rust, and now that I know it can move around on its own power, it's time to strip it down and clean it up for a final rebuild. No rust, no janky mounts, no zipties, nice insulated electronics, etc. 

Let's begin!

I'm very pleased to say I've joined the MIT Electric Vehicle Team (EVT) and they have graciously allowed me to work on the motorcycle into their laboratory space!!! Now I can disassemble the bike without the risk of people from outside stealing parts. Plus, there's a bit more room to leave parts in a corner than there is at MITERS (which is conveniently located downstairs). 

Things I want to do over the next few weeks:  

  • Strip all parts off
  • Remove ALL rust from the frame parts
  • Design and weld steel tabs to the frame for mounting and protecting
    • Motor
    • Batteries
    • Controller
    • Charger
    • Other components
  • Re-paint or powdercoat (YAY!) frame parts. 
  • Reassemble bike
  • Fix front brake
  • Lower the bike to fit my short body. I'm only 5'8" and this is a pretty large bike. 
    • Raise front fork as much as possible (lowers front of frame)
    • Adjust or replace rear suspension (lowers back of frame)
  • Clean the rest up. 
First things first. I've started to remove everything off the motorcycle, and first comes the seat, gas tank, rear fender and light assembly, front light assembly. 
Then came out all the peripheral electronic components, Cycle Analyst, Throttle, Sevcon controller, wiring...


Then came the chain and rear wheel. And the batteries. And the motor.
The frame is SO LIGHT NOW! I can move it around without much effort. The wheel assemblies on this thing really add on the weight. 


I removed the rear shocks, footpegs, and swingarm. I'm going to have to probably get shorter shocks or make/buy my own lowering blocks. With the CNC mill working, I may just let it make the lowering blocks for me!


I decided I would inspect the front brake (currently non-functional). I opened up the oil reservoir to find a lot of gunk. Following this guide (http://www.youtube.com/watch?v=0X6BX05JAo0) I cleaned it out. I decided to leave it empty (and drain the line and caliper as well) until I re-assembled the whole thing. 

Here I realized that the clamps for the front levers were where the mirrors mounted to! this rusted and sheared piece of bolt was the old mirror! I went to town trying to remove it. I Dremel-ed a slot on the top and tried to use a flathead screwdriver to remove it. I ended up breaking the screwdriver! Even after torching the area (The aluminum clamp would expand more than the steel bolt) and wailing on it with WD40, the thing would not come out. 

I ended up drilling a hole into the bolt and using a screw extractor, which got a little out of hand when the thing wouldn't budge, until it finally...
CRACKed, taking a piece of the clamp before it would move. I used a Dremel to remove as much bolt material as I could (I feel like a surgeon now). I then squeezed the sides of the bolt together and pulled it out. The threads seemed mostly intact, though...


And sure enough, I could get the mirror onto there! I can see behind me! Turns out there are two axes of rotational adjustment: The angle of the brake lever clamp on the handlebar (adjusts it vertically) and the angle of the mirror on the clamp itself (adjusts it horizontally).

Now if I can only find where I put the clutch clamp, which can hold my left-hand mirror... :x
 Now it's time to take apart the front fork. First comes the wheel...

Then the forks and handlebar. Easy enough.

At this point, I need to finalize where I need to weld tabs that will hold my batteries, controller, fairings, DCDC converter, and provide additional bracing for the motor. 


The motor mounting tabs were easy enough to design/make. I just cut some steel L-channel to size, and used a grinding wheel to shape the channels to fit the frame tubes and other bits. These will come out really nice with a MIG welder. 


Now, the batteries. I've decided to NOT use these two 12S8P packs, even though they have a nice BMS and pakage. The truth is, these batteries are unsafe in this cramped mounting position. If I move it far back enough to clear the front wheel and fender while the fork suspension is fully compressed, I cannot mount the plastic rear fender (which protects the components in front of the rear wheel). If I move the batteries far enough forward to mount that protective plastic fender (I have to squish the plastic battery case a bit against the steel frame), the batteries come dangerously close to the front wheel and fender. With a max front shock compression of ~5.5", there's no way the battery can clear the front suspension. Something has to change. 

Enter the smaller 12S4P pack that is available to me via an extremely generous donation. 4 of these packs is equivalent to the two 12S8P long packs I've been testing with, but they are each in a package that is half the size as well. This leads to more (and better!) options for storing the batteries on the motorcycle frame. 


After playing around with them, I found I was able to fit eight (8!!! Double the original capacity!) of the 12S4P battery packs in the front of the motorcycle in a clean and consistent space. Look at how elegant that is! 
I will only be using 6 packs though, for a total of 24S12P. (150% original capacity, 288 cells, 720 Ah, ~ 45 miles of range!)

To fasten these packs I need to design steel mounting brackets to be welded onto the frame. I'm going to add about 0.5" of padding/buffer/armor to each side of the battery to provide some cushioning and protection from rocks/shrapnel that may penetrate the battery pack and short something. I can't design the specifics of this until I have the 6 battery packs, so I will simply design it for a box of 3x2 of the 12S4P packs, with an additional 0.5" on each size. 

I plan to cut cardboard into a box of this size, then design mounting channels in the frame using steel L-channel to allow that box to easily slide into the channel from one side, then clamp down using webbing or some other type of clamping method. 

An interlude: I have to thinking about the painting process. The problem with the process is many things have to happen quickly after eachother, or the frame will oxidize or the primer/paint will not hold corrctly. I'm going to follow these basic steps:  

  • Sandblast frame to remove rust+old paint (start the oxidation clock)
  • MIG Weld tabs to frame in proper areas
  • Move frame to a dry, warm, and ventilated location and suspend frame 
  • Clean entire frame with acetone or denatured alcohol or mineral spirits. 
  • Coat frame with primer (RustOleum automotive primer for metal)
  • Wait the specified primer recoat time (about an hour acc to the label)
  • Spray LIGHT coat of color paint (RustOleum Semi-gloss black enamel). Wait recoat time. 
  • Spray main coat of color paint. Wait recoat time. 
  • Spray final coat of color paint. Wait recoat time. 
  • FINELY sand/polish the frame to make shiny. 
  • Coat with a clear gloss paint (RustOleum Clear Gloss Enamel). 
  • Final surface polishing. 
  • Leave in warm, lit area for some time (on the label). Maybe use a hot lamp?

All that must happen with very minimal interruption, especially the painting portion. Oh boy, what did I just get myself into... But first I need to finalize my battery mounting.

See you next post!

17 January 2014

EtekChopper: Further System Integration and Testing!


Last we left off, the batteries and controller were ziptied to the frame for temporary testing, and the motor was mounted with a waterjet aluminum sprocket at only two points. The point of this is to get the bike to the point where I can ride it around and prove that everything will work. Then, I will take everything apart and put it back together in a clean, final, ziptie-less fashion. 

The electronics and wiring were certainly a mess, but they will do for now. At least I have some idea for how many wires I'll have running around. 

Next I decided to mount the new 56-tooth rear sprocket. I have drilled the proper mounting holes, but the main hole where the axle passes through is too small. 

Using the indexing head on the mill, i was able to mill out the proper sized hole to mount the rear sprocket. 

And on she goes! With all the torque my friend Adam and I can muster. That's 100ft-lbs, right? 

I mounted the motor sprocket (I flipped it over from the way it is shown here). I cut a piece of keystock I found lying around MITERS to the proper length, and cranked down the beefy setscrews to the shaft of the Etek. 

Next came installing the new chain. I moved the rear wheel as far forward as I could in the chain tensioning slots (to ensure a loose chain) and wrapped the chain around the entire front and rear sprocket. I placed the master link on one end, and pulled the other end on top and marked where I was going to cut it to size with a black sharpie. 

The chain cut easily enough with an angle grinder, and the pin (shown intact above) fell out when I cut down on one en long enough. 

Now all I had to do is place the master link through the chain...

Et voila! The chain is attached and I can move the rear wheel, and the motor turns!

The chain looked straight enough, but I am still worried about it not actually being straight, or it contacting the rear swingarm while running. I may want to make the front motor mount adjustable to ensure minimal misalignment and stress on the chain. 

For now, this standoff on top and spacer on the bottom should suffice. The issue with this is when there's tension on the upper part of the chain, the whole mount will deflect and possibly oscillate. That's why there's a third (not visible in this picture) hole at the bottom left of the motor mount for putting a third brace of some kind. I will weld a tab onto the frame and drill a hole in that tab for the third standoff. 

I installed the Cycle Analyst's shunt resistor on the negative terminal to the controller. I love working with large power bars, because that means my terminals and rails can be minimally structural! Look at those aluminum/copper bus bars! :D

The speedometer of the Cycle Analyst works with a hall sensor and a magnet. The sensor is attached to the front fork with zipties, and the magnet is epoxied to an unimportant part of the front brake disk.

After tying the Cycle Analysts's B- shunt probes and B+ probe to the appropriate wires, I turned the motorcycle on to find this! Way cool! I ran the initial setup routine, which included inputting the front wheel circumference and other options. 

To test the speedometer, I lifted the front wheel up and employed the help of fellow MITERS-folk Charles Franklin who is making a really nice scooter of his own! His scooter theoretically goes about 20-25 mph, so if we push his drive wheel into mine and run it, my speedo should read about the same. 

And so it does! (It reads negative Ah used  because I had the shunt terminals backwards...) With everything in place, it's time for a quick test of my system in the parking lot outside MITERS!

Hand-me-down helmet, no jacket, no boots, no lights, no front brake (The rear drum brake works fine though!). Safety third! (Please don't do stupid things like I do, or you may end up like this poor bastard). 

Well, here goes nothing! 

It all works alright, but there seems to be some shaking/vibration when there it too much torque being applied by the motor. I'd attribute it to loose electrical connections, or the motor mount needing further constraint. Either way, I'll tackle the issue later. For now, the entire system works well enough for me! 

The chain sound is normal, though it sounds pretty nasty. Turns out the original internal combustion engine was enough to drown out the sound of chain. Other electric motorcycles sound the same, so I'm not too worried about it. That may be why Zero Motorcycles use belt drive for their ECycles. When I kick it up and rev it though, it sounds soooooo cool! :D I can't wait to get on this thing and really feel the amps working underneath me. 

I purchased an inexpensive cover for it before the fall 2013 semester got tough, and here it remained until January 2014. Next up is tearing it apart and cleaning/welding mounting tabs/repainting everything!