HOW WE USED GENERATIVE DESIGN TO 3D PRINT A DRONE
Why redesign the housing?
The Hubsan X4 is a very small and lightweight remote-controlled quadcopter, which has been designed to be as light as possible in order to maximize flight time. There are however limitations of the manufacturing process they used, and we decided to see how much lighter we can redesign the outer shell of the drone using 3D printing.
The original frame is injection molded, a commonly used fast and cheap material, but there are design considerations that fight against the ideal drone shape, all walls have a minimum thickness so the hot plastic can fill the shell mold properly. The original shell also had been designed in 2 parts, requiring screws and additional assembly time to fasten them together.
In such demanding conditions, it seemed like the perfect project to explore generative design and printing the organic shape with our U3DS 3D Printer.
from Autodesk’s Generative design page
What is generative design?
Generative design is a co-creation process, between designers and the computer. You enter a set of starting conditions and the computer iteratively tries to produce a 3D form that fulfills your goals.
The model is constantly being analyzed for its strength, so any parts that are not contributing will be removed. This process is very similar to genetic processes, so it should not be a surprise how organic the output often looks.
How the quadcopter frame was created using generative design
For the drone, our goal was to hold the motors, the circuit board, and the battery securely in place while minimizing weight.
After modeling their elements, ‘Must keep’ geometry was created around it to give the software a starting point and to have surfaces to attach forces to. These forces are the basis of the generative model as these are the inputs for the analysis to determine whether there is enough strength at any location.
Setting the correct forces is the most important task when setting up a generative design, setting them too high and you’ll use way too much material, set them too low or miss a force and the design might not be safe, or work at all.
In-flight a drone has almost no stress going through it, it weighs only 60 grams, and when it’s hovering there's an equilibrium between the forces keeping it up from the motors and gravity pulling it down. The real challenge is landing! Unfortunately, this software doesn’t have support for impact forces, so we had to compensate with higher than typical forces through the corners, and set the safety factor high enough to cover even the worst crashes.
To encourage an aerodynamic shape, we also added some blocking shapes to limit the possible design. And to force the motors t knot directly connect with each other, but through the PCB and battery in the center.
Initially, the generative design was set up using defaults to make sure the geometry was 3D printable, but the minimum feature size was very safe, which led to a pretty simple design, which did naturally reach some design features that engineers use in day to day life, thickening the edges of the part, Using similar reasoning behind I beams, having the most material as far from the bending axis as possible to maximize strength.
For this project we wanted a design more traditionally associated with generative design, something that would be impossible to make with traditional manufacturing methods, so we reran the generative design simulation with no manufacturing considerations and just a 0.8mm minimum feature size.
You can see how the design both respects the starting keep and blocking geometry
3D Printing the shell
After designing the new shell came the more simple job of printing & assembly.
The model was sliced in our slicer with automatic supports and was printed in our Iron Grey as it captures detail insanely well, but more importantly for this project reasonably strong to survive falls and very rigid, which means that drone behaves well in sky not flexing as it flies which would cause issues.
Here is the shell before removing supports and post-processing. The print cost only 500INR of material and took a little under 4 hours.
To save weight the original drone directly soldered the motors wires rather than using connectors that you’d see on larger drones. This meant that assembly included the tricky task of soldering in between the struts of the 3D print. If weight reduction wasn’t the goal it would’ve made a lot of sense to add JST connectors to make future work easier.
