Introduction: Miniature Autonomous Blimp
Autonomous drones and the like have become a hot topic lately and have many active communities working to expand their capabilities. Everyone seems to have multi-rotor drone these days, which leaves other aircraft behind. I think it's time to take back the skies, (slowly) by building your own miniature autonomous blimp! A blimp you say? Like the HINDENBURG?! Actually, the Hindenburg was a zeppelin (rigid airframe) and achieved lift with copious amounts of hydrogen gas. We shan't be using anything nearly so dangerous. This blimp is mostly latex and helium, and has an Intel Edison at the electronic helm capable of navigating on it's own. How neat!
Step 1: Parts and Materials
(3x) motor
(3x) micro servo
RC LiPo battery
(3x) propeller
50 ft^3 helium
1/8" ply wood (The vector paths and .stls for these parts are all included in the archive attached above)
(3x) 3D-printed angle bracket
(2x) 3D-printed bracket
3D-printed base piece
M2 screws (nylon)
M2 nuts
M5 x 10 screws
M5 nuts
1/8" x 2' carbon fiber tube
Attachments
Step 2: Blimp Design Overview
Theoretical
Core to the concept of blimp-based flight is passive lift. The volume of helium is less dense than the surrounding air and makes it buoyant. Due to the weight of the balloon itself, the frame, and the electronics, we'll need at least 7 ounces of lift. The amount of lift per cubic foot of helium is roughly 1 ounce, so we'll need at least this much. A custom mylar balloon would be ideal for a custom size and volume, but it is much easier and more practical to simply use a large party ballon. Of course, the ballon isn't a perfect sphere, but it is close enough in form. A three foot diameter ballon should do the job since the volume of a sphere:
4/3 * pi * r ^ 3
r = 1.5ft so
1.33 * 3.14 * 3.375 = 14.14 ft^3 which is more than enough for our payload.
Even at r = 1.25 (30" diameter balloon) we get over 8 ounces of lift.
Mechanical
The blimp consists of a wooden airframe which provides connections points for a LIDAR sensor, motors, and balloon. Carbon fiber rods extend down from the airframe to provide a connection point for gondola, which provides mounting holes for the Edison and other electronics. These rods fit into 3D printed pieces for adapting between the rods and wooden components while accounting for the change in angle as well. Although the 3D printed components add a bit of weight, they allow for more design flexibility around the air frame so that it doesn't need to be re-cut for different additions.
Electrical
The blimp has two main sensors: a forward facing LIDAR which allows for long range object detection, as well as a downward-facing IR distance sensor. The mind of the electrical system is an Intel Edison, which gathers the sensor data and sends PWM signals for controlling the tail rotors and vertical thrust rotor.
Software
The program running on the Edison is a simple Arduino-syntax-styled sketch running on the Edison. The blimp's goals are obstacle avoidance and maintaining lift.
Note: Attached to this step you'll find the .STL files for the 3D printed parts as well as the vector paths for the wooden components.
Step 3: Schematic
Power
The blimp has two power systems in order to reduce noise from the motor circuits. The first is the lower voltage system which is supplied by the LiPo battery attached to the Edison power block. This supplies the necessary voltages for the Edison as well as the 3V3 power rail for the LV connection on the logic level converter. The battery block has a mini slide switch for control on the side. The second power supply is provided by the mini RC LiPo battery which is boosted to 5V via the step up converter and switched with the push button power supply before being connected to the LIDAR, motors, IR sensor, and HV connection on the logic level converter.
Motor Control
The motors are controlled by the drivers from micro servos. It's hard to beat the compact size of a micro servo controller board. Using servo drivers instead of other H-bridges or transistors allows for single pin, bi-directional PWM throttling of the motors.
Processing
An Intel Edison connected to an I2C block and GPIO breakout allows it to communicate with the other sensors. The 3V3 PWM signal is logically high enough to directly control the servo motors, but needs to be boosted to 5V for the I2C signals and for safely reading the digital output of the IR sensor.
Sensors
The LIDAR Lite I2C pins are connected via the logic level converter to the socket on the Edison I2C block. The IR distance sensor is also level-translated for the Edison. While the IR sensor can operate at 3V3, it has a slightly greater detection range at 5V, and we might as well use the extra connections on the level converter!
Step 4: Air Frame
The mechanical skeleton of the blimp consists of a large flat wooden frame that is laser-cut from 1/8" thick plywood and consists of four main sections: the head, center frame, Gondola, and tail.
Head
The head extends away from the front of the frame and provides a mounting point for the LIDAR sensor bracket. This serves to keep the sensor from pressing into the balloon and, by having it so forward-mounted, allows it to detect objects even more closely before they might come into contact with the balloon.
Center Frame
The center frame consists of two concentric rings with mounting points at 60 degree increments along the edge. The balloon is pulled into the center of the ring and secured to the frame by pulling the tip of the neck through the gondola adapter piece.Three carbon fiber tubes extend downward toward the adapter. The force of the ballon pulls the gondola up, and the tubes distribute this evenly toward the edge of the frame.
Gondola
The gondola consists of an adapter bracket and the main electronics mounting plate to keep the blimp's center of gravity as even as possible. The vertical thrust rotor mounts perpendicular to this mounting plate.
Tail
The tail extends toward the rear of the airframe and provides screw holes for directly mounting the tail rotors to the frame. The tail serves to extend the propellers away from the balloon.
Step 5: Carbon Fiber Tube
We'll need three 6" long carbon fiber tubes to attach the gondola to the main air frame. The angle adapters have about a 1/4" depth for the tube, so it's better to cut the tube a bit short for an easy fit. Measure, mark, and wrap tape around the cut line. Cut the tube with a high speed rotary tool over a constant vacuum to suck up the loose fibers. You really don't want to breath this stuff in, it's quite nasty for your lungs.
Step 6: Balloon
Let's get this bad boy into the air. You'll need your tank and balloon ready for filling. I suggest having at least double the volume of helium, as you may wish to fill more than one balloon. It's also useful to have excess in the event of a catastrophic, balloon-overfilling event. I definitely didn't pop one or more balloons in this process. Nope. Reports of large ballon-burst noises in the office are purely rumor and highly exaggerated.
A note on accuracy:
It's important that the ballon be filled to near capacity as the relationship between the diameter of the balloon and its lift (as a function of volume) is exponential. So even a few inches decrease in overall diameter can result in a fairly steep loss of lift. A good practical way of measuring (well, accurately eyeballing) the size of the ballon is to fill it in a door jamb. This is sensible as most doors are generally 30 inches or greater in width, which is close enough to the full capacity of the balloon. Also, restricting the width of the ballon makes it much more transportable, as squishing it through at full capacity may result in a sudden loss of helium and slight hearing damage. I can definitely still hear OK, ok?
Step 7: Gondola
The carbon tubes have a friction fit for all of the adapters. The gondola plate and electronics all mount with nylon M2 screws, which are not only non-conductive, but much less heavy than their metal counter parts. You can see how the electronics rest more clearly in the 3D model photos.
Step 8: Vertical Motor Assembly
Attaching the vertical motor is fairly straightforward. The motor assembly screws into the small wooden plate, which in turn screws into the 3D printed right angle bracket. The right angle bracket in turn screws into the mid-section of the gondola plate.
Step 9: Lidar Mount
The LIDAR unit mounts to its own unique wooden plate, which is also screwed into another 3D-printed, right angle bracket. Although there are four mounting holes, two will suffice for securing the sensor to the plate. I used metal #6 screws in this case, but M5 screws will fit nicely as well.
Step 10: Servo Control Extraction
We'll need to extract three servo motor driver boards from their enclosures. You'll need a very fine tip Phillips head screw driver as well as some small, sharp snips. The case is held together with 4 long screws and pops open readily once they are removed. With access to the circuit board, you can remove the two red power lead to the mini DC motor and set it aside for other small electronic projects. We need to keep the pot attached to the PCB, but this part of the case is welded closed, so we need to carefully snip it in two in order to pop out the assembly. Do
Step 11: Modify Rotors
For each of the rotors, the new motors need to be soldered to the servo PCBs. The polarity of the wires as the attach to the board is irrelevant as the direction of rotation can be adjusted with the potentiometer.
Step 12: Download Software
You'll need the latest Arduino IDE capable of supporting the Intel Edison. The sketch and library are attached above. You'll need a console block or base block as well as two micro USB cables for the upload.
Attachments
Step 13: Operation
After turning on the Edison power supply, wait about thirty seconds for it to fully boot up. When ready, activate the motor power supply. The tail rotors should lightly power up (you may need to adjust them with the potentiometer). The vertical thrust rotor will activate for a few seconds if the IR sensor detects that it is too close to the ground or any object below it really. The blimp constantly reads the distance and randomly turns to avoid obstacles as it floats along. (Note: the design is only suitable for indoor use only!)
I built this as a basic platform that could definitely be expanded. There's definitely room for upgrades. Let your imagination run wild! The (indoor ceiling height) is the limit!
25 Comments
7 years ago
Can anyone shed any light on what the servo PCBs and potentiometers are for?
Wild guess... Convert PWC to simple voltage? Allow fine-tuning of propeller speed and direction? Does anyone know what the chip is?
7 years ago
How much cost ???
7 years ago
The LIDAR sensor mentioned in this post has been reitred from Sparkfun and not available at alternative vendors as well. Is there an alternative product we could use without much to change in other components/code etc.?
Reply 7 years ago
It's a little crude, and unless angled properly wouldn't be able to do height and such, but it may help with horizontal navigation. Hope I helped! .-.
Reply 7 years ago
You could use a distance sensor to constantly check if it's within 'X' distance of a wall or other object, and when it's that close, "tell" the little thing to turn. Any distance sensor could work for this type of project, I'd think. Maybe unsupported, but if you can code a little, it shouldn't be a big deal.
7 years ago
Looks kinda like a Steampunk Balloon, good work :D
7 years ago
It seems like the 3D stls are unusable. I tried to open them in Cura for Ultimaker2 print and I get a Warning: Object scaled down. What are the actual/physical dimensions of these stl models?
7 years ago
OMFG! The video! The video!!! Best. Video. Ever. Haha! That voiceover is perfection. I'm so glad this exists. Bravo. Next level.
7 years ago
I have never heard of Hi-Float; it's a splendid way of preventing helium balloons from losing gas, but not cheap. I wonder if you could paint the outside of the inflated balloon with some cheap nail varnish? It could be that the solvent does for the balloon, but who knows?
7 years ago
I am sorry that I did not explain myself well. My problem is that I cannot hear the voice of the presenter, because the background music is too loud; and that problem cannot be fixed by my lowering of the volume; but by the person making the video, lowering the volume of the background music.
7 years ago
Could you post it again; but lowering the music volume?
Reply 7 years ago
Respectfully the YouTube player does have a volume control. I'm sure is there any preset volume can't please everyone's needs. My keyboard has volume controls that work well to adjust the computer directly. You and others might find the information here handy
http://www.hongkiat.com/blog/youtube-keyboard-shortcuts/
7 years ago
Please check your sources! That myth has been debunked many times.
http://www.airships.net/hindenburg-paint
Reply 7 years ago
Odd article; this is the first I read the claim the dope used on the fabric skin was the cause of the accident. I have ever read previously it's that dope that created the visible flame not the hydrogen. That is not to say the hydrogen didn't burn. What is characterized as flames from the burning hydrogen was most likely was water vapor. A hydrogen flame is generally invisible to the human eye. From what I have read it was static electricity that somehow ignited the hydrogen was the root cause of the accident. Reads like that article has an agenda to make hydrogen the lone bogyman of the incident. Hydrogen can be dangerous, but is still used to launch weather ballons
Reply 7 years ago
I'll quibble with that "debunk", as it doesn't actually do so. Actual tests have shown that the coating was in fact quite flammable. Additionally, hydrogen flame is pale blue and invisible in daylight, while the fire is obviously quite visible and was bright yellow. Obviously the hydrogen also ignited, spreading it throughout the structure, I don't think anyone has ever said it didn't burn. But this dirigible would have burned had it been filled with helium, just slower.
7 years ago
Your video won me over
7 years ago
awesome video. Well directed.
7 years ago
One way to keep from popping the balloon during inflation is to use two chairs. Turn them so that the backs face each other... set the distance apart to to the rated size of the balloon. Use the chairs as a gauge to show your progress, when the balloon just touches both chairs, you are at the rated diameter. Party stores use something called hi-float to make the balloons stay up even longer, it keeps the helium atoms from passing between the latex molecules. Use just enough to coat the inside of the latex... it can add days to the float time.
Excellent design by the way... nice use of miniature parts and materials. An alternate design would be to use 3 Q24 balloons (Qualatex 24 inch) as the lift would be considerably more... and they tend to get a large "neck" before they are anywhere near popping... which makes a good visual indication. They are also cheap... usually $1.25 each as compared to the Q36 shown.
If you have a party supply store, ask for a Q24-flat.
7 years ago
Where can one buy one of these? Can a member of this site make one for me?
7 years ago
Firstly, that's an awesome project. Secondly, the video is just super creative & highly amusing. Loved it.