Over Spring break I had some time, and I was missing my MiniQuad since I never had the chance to finish testing it. Though BigQuad was nice but heavy (a few pounds!), and MiniQuad was an attempt to reduce that (~60% reduction in weight), I wondered how much I could push the scale. Obviously, there are quadrotors that scale down extremely. My goal was to get under 400 grams, and still hold a reasonable payload. And at that scale, with 10-15g motors, things get cheap. Everything is at least a third to a quarter of its large scale counterpart in pricing, which further encouraged me.

I was able to laser cut a small piece in the scraps bin that fit the dimensions for my frame. With a  200mm square profile, the frame contributes to nearly half of the weight. After my Hobbyking order arrived a few days ago, I was able to fit everything to this:

It was a modification to my original design. Though I originally planned to weigh around 400 grams, I could not find suitable propellers for the motors. These 5×3’s I believe are smaller than the recommended size, but they were the only size left that were not on back order. Nevertheless, the further reduction brings this little machine to 250 grams final weight. Probably, with some carbon fiber, I could reduce the weight to 180g or so, but the 250 grams is good for now.

Modifications from previous designs include an XBee radio over the Spectrum radio system I was using before, a digital IMU with 3-axis magnetometer, and an Arduino Nano over Uno. The switch to XBee will hopefully allow me to guide by computer, with a little FPV GUI showing important flight data stuff, and allow me to gather data remotely. The last part is crucial, as the most compelling reason I wanted to make this scaled down quadrotor was to function as a mapping drone. With (potentially) stereo cameras, its magnetometer, and a potential GPS system, this little quad will be able to determine heading and location. Some contact sensors will be placed around the quadrotor to allow the robot to bump into things and detect obstacles, and its small size will ultimately allow it to fly indoors relatively safely and navigate through halls, doorways, and windows.

I know I’ve been branching out to too many projects at this point. I have Spirit’s armor left to upgrade, MiniQuad to fly, and my robot arm left to automate. I’ve been meaning to go into more depth with bioprinting the plant cells, getting around to that drop on demand inkjet toolhead expansion, and finish upgrading the build surface. I need time to focus on a single thing. But school, and school takes priority. It’s the mindset that this coming summer will provide that extra gap of time I’ll need to get these things rolling again. The Shadow Fox project too, has been put on hold until next semester. At least temporarily, as the funds are lacking. I need to refine my design heavily before investing any time into actuator tuning and research (currently between PAMS, HAMS, and serial elastic connected motors). To aide that, I have a dozen or so sub-2$ turnigy servos to make a miniFox to experiment with serial elasticity as well as dynamic walking gaits over the summer. As for school, I’ll be busy for the next few weeks. I’m making a RoboChoir.


Edit* added a fancy 360 view. Click for gif

Here’s what I have designed so far. The head and tail are just placeholders for now, I don’t want to add too many degrees of freedom that aren’t necessary yet.

Full Render:

Side view:

Top view:

I’m working on a new quadruped over spring break, where I will have much time (alone with no food). CAD it up and make a small model if 3D printing allows. It will focus on a biometric tensegrity structure, which mimics mammalian skeletons and ligaments. Instead of direct drive in torque based joints, I will try to support structures with ligaments and partially tensed tendons.

Here’s what I have done so far with CAD. The tendons are not added in yet and I still need to design the spinal structure for the left/right motion. I also need to accommodate for the rotary sensors/flex sensors for the spine. For this robot, I want to integrate a two degree of freedom flex sensor inside the spine. I was inspired to try building a running quadruped by the cheetah robot by Boston Dynamics. The video of it’s record breaking running speed is circulating on the internet.

For now, I plan to have it electronically actuated, as pneumatics are way out of my price range. Three degrees of freedom per joint, two for the spine, two for the neck, and two for the tail. Then the walking gait generation fun begins again. And once again as inspiration, I have Zoids:

My fox project’s namesake, the Shadow Fox.

On a more relevant note, the other quadruped I’ve been working on is nearing completion. After a long lull of waiting for the university to deliver parts, I need to calibrate the sensors and servo motor settings, but then hopefully it will take its first steps.

So I’m back at school again, so I returned to where I left of with the dog. The chassis has been sitting around in CAD form for a while, two days ago I printed this thing. It’s my first time using one of the commercial 3D printers. It’s the same 3D printer I keep seeing at universities, of the Dimension brand, that prints extruded plastic with a weaker support material.

For a job the size of the quadruped, the print time was 44 hours. The chassis had to be put on its side and diagonal across the printing bed to even fit.

When it was done printing after 44 hours, it was a piece of plastic encased in a support material block. For the fancy Objet printer in the other lab, the support material seems like a powdery material you have to pressure wash off. This material you had to peel off with your fingernails and some metal picks resembling dentist’s picks.

And after some frustration and an hour or so of picking away, it looks like this:

The little gaps will need some creative sanding, and I’ll have to re-bore the holes. Otherwise, I’m getting closer to mounting the motors, legs, and valves. The next step is to 3D print the precision valve parts with the UV resin Objet printer.

I was pretty happy with my previous quadrotor build, because it was good for its purpose (to provide aerial shots with the GoPro camera). It was large but stable with its long arms, but those arms also added much additional weight to the frame. Moreover, the entire thing was made out of wood and steel machine screws.

My main goal was to make it light and small enough to work as an aerial surveillance drone that would last for practical lengths of time. Since Spirit MK2 is in fact modular and can thus house a helipad, my thought at the time was to create a larger landing platform (Spirit) with its tank tread drive modules and have it carry the quadrotor to its destination. The quadrotor would then fly vertically, take a 360 panorama of the landscape and maybe fly around any areas of interest, then return to the mother-ship (turning Spirit into an aircraft carrier!).

It would be a wonderful splicing of two of my projects (a drone within a drone). As mentioned in earlier posts, I bought a very cheap video camera transmitter/receiver combination that I have housed on this quadrotor. At 40 or so grams, it certainly weighs less than the 400g weight added by the GoPro and mount. With only 150ft of transmission range, the system would greatly extend the usable distance of the quadrotor if the video receiver was mounted directly onto Spirit and later relayed back. And with that, here’s my progress on the MiniQuad so far:

The arms are now aluminum C beams, but the mounting plates are still particleboard. It seems I’ll never fully upgrade to metal until I’m able to buy a mini-lathe and a mini-mill. Fiberglass and carbon fiber were out of the question of course, since I’m limited to the selection at the Home Depot, and I was not able to find any source of significantly large pieces of styrofoam to cut. As a result you have a (proportionally) lighter and compact quadrotor. Motors are now directly mounted to the beams, and I’ve added some plastic landing gear made of Polymorph/Shapelock/Instamorph material. Those standoffs you see there are in fact wooden dowels.

The landing gear was heavy and failed often since it was a last minute add-on with zip ties on the previous quadrotor, so this new addition is a nice feature. The whole assembly is lighter, I promise! It was worth the hours spent rebuilding a smaller chassis. Here’s the MiniQuad’s frame next to the BigQuad’s frame:

The scale may not be apparent in the picture but the diameter of the Mini is less than half that of the Big. I have no idea on how this will effect its wind stability but this smaller size allows the quadrotor to sit on the central module of Spirit without any problems.

As of now, the MiniQuad is almost complete mechanically. The two standoff totem poles in the first picture will be replaced with a lighter material, and will support the canopy of the quadrotor. I’m hoping for something fitted and nicer than last time (a plate of plexiglass) to protect the electronics from dirt and flips. The rat’s nest wiring is shown in the next picture. I’ll clean it up after everything is sorted out.

Here’s the camera mount. It’s the same Polymorph material molded around the camera. This stuff is getting really useful when I don’t want to machine anything. Great for low strength, custom mounts!

My next step in this process is reconfiguring the software on the Arduino. I’ll probably do that tomorrow, but due to sudden and odd rain (Southern California) I won’t be able to do many outdoor tests. Since my brother’s room is vacant after he moved most of his stuff out to school I’ll fly it around in there. Video transmission does work, but it can only display on screens for now. I’m waiting for a to-USB adapter so I can record/view the video from my laptop.


The x-axis is slipping in the reverse direction. Forward is fine, maybe a little oil will fix it. Y and Z axes work just fine, but the motor doesn’t like low speeds with the motor controlling program I wrote. I’ll have to change some of the pulse timing for smoother operation. I found out the set screws don’t really work with threaded rod, so I’ll have to make some solid shaft adapters. The control will come soon! It seems easier to write a G-Code parser for the Arduino right now and just have the printer blindly print files, but a nice interface like on Replicatorg might be nicer. I might just end up doing both, since writing it myself will ultimately be more customizable than modifying the source code. It’ll be important once I start using some funky tools.

3D Printing has risen to an undeniable level of popularity in the last years. From something that was hardly known to be possible to the public, 3D printing has made it to public access. Projects like the RepRap and the Makerbot have gained some momentum. My own project team, Fab@Home has done its part to contribute to this community.

So here’s where I enter. My main interest with 3D printing technology was to apply it to bioengineering. Take some cells from various tissues you want to reproduce, and culture them in a bio reactor. When the cells are sufficient in quantity, load them up in some form of semisolid gel and extrude them with extreme accuracy. Under the right biotic culturing conditions, and with the right blueprints and precision, you can reproduce tissues. Combine those tissues, and voila! ORGANS.

But it doesn’t just stop there. Of course you can satisfy the organ donor waiting list and help hundreds of thousands of people live normal lives again. But there is more. Say you have vestigial features you have in an organ that you could be better off without. Why not harvest some cells and rebuild a new, better organ to accomplish what you need more efficiently? With organs made with your own cells, there would be no chance of rejection. Organ printing can do things impossible to achieve with regular evolution.

There’s more. My brother suggested this one to me. In agriculture, highly developed breeds of corn are developed through genetic modification. Each corn plant must be propagated by hand by trained technicians. But with 3D bioprinting, there is so much more you can do. Engineer a convenient capsule to house the plant and artificially create your own seed packets for an otherwise fragile sterile plant! Print these out in mass quantities to grow crops that are otherwise unreasonable to use. Of course it’s a stretch in development, but I’m sure engineers will not fail to impress with their innovations.

Now imagine layering cells from different organisms together so closely that they develop in unison and equal to one another. Imagine a chimeral tree that produced different fruits at different times throughout the year, or an animal that contained the genetic material of different species that were otherwise incompatible. Imagine silicon printed with living cells. Transistors and neurons printed together to form viable human/machine neural networks! You can tell I’m excited about the possibilities with this technology.

So here’s my first 3D printer:

Only the chassis is shown + motors. If may not be the most efficient mechanically, but I’m proud to say that this printer was entirely of my design. Only if you count the current extrusion tool too, which I modified and developed as my assignment on the Fab@Home team. The whole thing is made of acrylic, which I got cut since I don’t own a laser cutter (though it would be nice considering the reasonable cost of these lasers).

Here are some of my CAD drawings in designing this thing. On and off, acquiring the parts (it’s surprising how difficult it is to find accurate 3/8″ aluminum rods) and designing the printer took part of my fall semester. I didn’t have time to get the acrylic and assemble the whole thing until winter break when I got home.

First iteration concept. I tried to make the extrusion surface independent of the 3-axis gantry to make sure vibrations from the motors wouldn’t perturb the printed object.

Second iteration (I discovered the scene/render option in SolidWorks)

With the stepper motors and the 1/4″-20 standard (not the expensive Acme stuff) threaded rod, the setup can theoretically reach tens of microns of resolution with microstepping. However, due to the inaccuracy of my extruder, that resolution probably won’t make much of a difference. The extruder I have is a syringe with a plunger that is driven by a drive screw that is driven by a continuous rotation servo motor.

Comically, the syringe tubes were taken from the sharps disposal section of my lab (it probably wasn’t that wise of an idea). Good thing the lab wasn’t dealing with anything medical. I will have to replace them with sterile tubes once I begin wet testing, since contamination is a huge issue.

Here’s the extruder:

Clear acrylic looks nice, but it’s terrible to photograph effectively.

I’m using EasyStepperDrivers from Sparkfun to drive the stepper motors in this project. Coupled to an Arduino coupled to ReplicatorG (for now until i develop my own), I’ll have to make a custom .xml file for the servo extrusion configuration.

The next challenge for this project is to get viable structures to print. It’s a difficult problem right now for bioprinting, as finding a 3D blueprint of an organ or biological structure requires identifying hundreds of different materials. My two material setup probably won’t be able to accomplish that, so for now I’m stuck printing with two materials.

I’m ordering sodium alginate + calcium chloride to experiment with. When mixed, the calcium causes the alginate to form closer bonds and become more solid-like. With some living cells in between, I can hopefully build up some reasonably sized structures. I intended to use the cells from a rosebush on my lawn, but my winter break is proving to be too short for me to culture the cells now. Added with an internship at an electrical/civil firm I’m working for right now, it’ll have to wait till the semester. I hope I have time.

I was debating between two styles of approaching bioprinting. Some scholars researching bioprinting use inkjet technology to deposit bio-ink laden with living cells to deposit many small layers of living cells to build up tissues. Since the volume of the drop of liquid ejected from an ink-jet is reasonably close to the volume of a cell, the inkjet can reliably deposit cells to their target. However, inkjets use bubbles generated by heat from a resistor or by mechanical shock to create such precise droplets of ink. Therefore, cell survivability is an issue. The other approach is to bundle many cells into workable unit voxels. The larger units could be placed accurately while maintaining cell survival. I’ve decided to make both systems.

Coming up next semester, (if I can get access to a biolab) actual printed cells! And a belt drive inkjet system.