A Tensegrity Micro Quadcopter, Part 2

I've been playing around with my tensegrity quadcopter a bit more and I've come to the conclusion that this configuration works best. The video above shows a tensegrity frame made out of 20cm carbon rods with 3D printed end tips, string and 3D printed string tensioners holding it all together. This is an advancement on my last post where the frame wasn't tensioned properly. I'm quite impressed with my design for the tensioners, which was done using OpenSCAD. Due to the weight restrictions, I needed to find a way of tensioning the wires without adding too much weight. The whole thing needs to weigh about 50g and my final evolution of tensioner is only 0.2g. I think I've got about six of them tightening up the frame so this one doesn't wobble like in the last video.

The tensioner in OpenSCAD. About 32mm long and 2mm thick.

The model is called "tensioner2" in my Tensegrity GitHub Repository.

All I do is to hook one end around the string, wrap the string around the notches on the body and hook the string into the other end. I can wind it around as many times as I need to and I can tension the wires as much as I need for very little weight penalty.

In the video above, you can see the tensegrity quadcopter hitting the wall a few times and surviving a lot better than before now that it's tensioned properly. Also, I now understand why it is so hard to fly. At the end of the video you see it roll quite hard to the right. What I actually did was to give a quick left yaw input and you can see it clearly roll right quite violently. I think it's the weight distribution of the frame that's causing the adverse aileron/yaw coupling, but I'm going to need to switch to a proper flight controller to do any more experiments. At the moment I'm still using our modified HubSan X4 flight controller which I can't program.

This video shows the same basic configuration as the last post, where I've put a small, orange, quadcopter frame inside the tensegrity frame. It's just a small micro quad hanging onto the carbon rod frame using some 3D printed clips (the "motorclip" model in the same OpenSCAD file as the tensioners). I have been experimenting with attaching the motors directly to the carbon rods to save weight, but actually ended up making it heavier!

The orange quadcopter frame that I'm using is a model I download from Thingiverse called a "Micro 105 FPV Quadcopter": https://www.thingiverse.com/thing:1221911

This frame only weighs 5.8g! However, the way I'm attaching it to the carbon rods using the motor clips is causing it to snap the motor supports when I crash.

This is the clip that attaches to the motor and the carbon rod frame.

The four motors are placed into the quadcopter frame as normal, then the four motor clips are fitted onto the motors. The cylinder part of the clip then attaches to the carbon rod and holds the quadcopter inside the tensegrity frame.

What I did next was to try to get rid of the orange quadcopter frame, as you can see in the image below:

This was a complete hack, as I didn't have access to a 3D printer and made some motor holders using plywood and some old 3D printed motor mounts. It's held together with screws, which clamp the two bits of plywood either side of the carbon rod with a notch to locate them on the rod. I still had to add a piece of foam across the two rods to attach the flight controller, though.

I now have the motors attached to the rods directly, so how does it fly?

Badly. It's too heavy. Each one of the motor mounts weighs 3.4g, so they're adding 13.6g. The entire orange quadcopter frame which I'm trying to get rid of only weighed 5.8g! I then went and designed a 3D printed version of the mount (it's in the OpenSCAD), but that also weighed 2.6g, adding 10.4g. This needs some more thinking about and some work. I can get it to fly, though.

With the motors attached to the rods directly, it's imperative that the motors are placed at the same positions on the rods to form a square and that the motors are pointing upwards and not at an angle. Once I had that sorted, I could get it to fly briefly, but not well enough to make a video. This is still a work in progress.

What we've learned so far is that this looks like a viable method for making a quadcopter which sits inside a protective shell. I haven't seen any of the vibration problems that some other people have found with motors mounted directly to the frame [link], but it does suffer from adverse yaw/roll coupling which makes control very difficult. Tensioning of the wires is necessary for the frame to maintain integrity during crashes and during flight it prevents it from oscillating and causing control problems. Still to be investigated is whether having the motors attached to the frame can reduce the vibration that the flight controller sees when the motors are running. However, this is going to have to wait for an improved design of motor support which is light enough to fly. This seems to be the limiting factor of the tensegrity design, namely that the added complexity of the tensegrity frame adds to the weight of the quadcopter. I think a few more design evolutions are needed.

GitHub repo for the OpenSCAD models used here: https://github.com/drones4good/Tensegrity