CAL - Adaptive Voxelization

One of the limiting aspects of the CAL printing process is the computational speed - the projections generated through the optimization procedure are at a much lower resolution than the hardware is capable of. This is due to the low resolution voxelized models we use. The idea of adaptive voxelization is to increase the voxel resolution only in the areas where it's needed. The ultimate goal is to use a 3D octree subdivision of a model to determine where these areas are, adaptively voxelize, optimize, and stitch the projections together. To start, I have developed a z-subdivision method - the complexity of stitching these projections together is greatly reduced compared to the full octree. This method shows promising results with optimization time being notably increased. Soon, this method will be tested on the CAL system for validation. 



As a side project for my lab, I wanted to model a novel 3D printing process. Multi-laser volume stereolithography is a nearly unexplored technology that has a lot of potential. I have been creating a basic finite difference method model of the process in MATLAB. I am hoping to add more of the complex multiphysics aspects of the technology by the end of summer. I believe that CAL and MLVS could be used together very effectively - CAL excels in high throughput while MLVS can define precise surfaces. If the model shows potential, this could provide future work for the lab.




After watching the demos and reading the design considerations of E3D's Toolchanger 3D printer, I knew that this was going to be a huge development in FDM 3D printing. I designed my own slightly adjusted motion system along with a reimagined swapping mechanism using 3 spring plungers to create a kinematic coupling. Although I haven't had the opportunity to fully fabricate and iterate on this design, I was able to prototype the motion system as part of a Hackathon while interning at Formlabs. My team created an FDM-SLA hybrid 3D printer - extruding resin using a syringe and immediately curing that resin using 3 focused lasers. Although the system had several design and fabrication flaws, it was a cool proof of concept and has me excited about the idea. I hope to fabricate this tool changing design by Fall 2020.




Early in my college career, I wanted to create a high-speed FDM 3D printer. After looking at a lot of DIY designs, I selected and synthesized all of the features of high-speed printers into this design. Thus, this printer takes after the RepRap movement, using a lot of 3D printed components and design tactics intended to reduce parts and improve ease of assembly. After a somewhat successful prototype, II quickly realized that I wanted more from this machine and moved onto the tool changer design. However, this project ended up being a great exercise in modular design, design for 3D printing, and detailed assembly and rendering.



The final project for my engineering graphics course was my first exposure to animation and rendering. The idea was to fully explain the functionality of the subject through animation - I have since found this to be an invaluable means of communication in engineering design.. I have been interested in mechanical calculators ever since seeing this video breakdown. We chose an older model calculator and animated the assembly and function of several of its capabilities. I have always had an interest in clever mechanisms and this design was full of interesting design choices.

Designed in PTC Creo, animated and rendered in 3DS Max.



Parametric designs that fully take advantage of the capabilities of modern CAD have always been interesting to me. As my 'master modeling' skills have improved, I've come to realize that any well thought out design should be able to do this. I've also adopted a form of prototyping in the way I design - starting with rough sketches, moving on to more detailed drawings, and then a mock-up CAD before developing a master model and creating components. These parametric designs were instrumental in developing this workflow. They demonstrate the thought that should go into design - there are many ways to make a model appear in CAD, but some methods are better than others.



One of my projects at Airwolf3D was to design a hook to lift a car. This project eventually spun off into an analysis of the many materials which Airwolf3D has to offer.

I ended up writing this brief article describing the testing and design process involved. I performed Fusion 360 static stress simulations to help determine my predictive order for the hook performance. After studying the design of industrial hooks more thoroughly and performing a shape optimization study, I ended up with an interesting design.



This project was the ambitious idea of UAVs@Berkeley President, Trey Fortmuller. The concept is that a quadcopter lands on a flying hexacopter, where its battery is swapped out for a fresh one. The initial idea was to use a conveyer belt system to push out the un-fastened battery case (the screw holding it to the quadcopter is loosened first) and push the new battery in from a battery magazine. The landing would be assisted mechanically (with guiding cones) and electronically with computer vision assistance. In hindsight, this is more of a graduate-level control systems project - nevertheless, we gave it a go. Although I was unable to finish the project, the team's results are detailed here.



My first taste of research at Cal involved working on a 3D printing failure and sustainability analysis for Berkeley LMAS. This research eventually culminated in this paper and this video. Additional research revolved around the recycling of 3D printed plastic - a process involving printing, making pellets from printed material, re-extruding into filament, and testing. The comparison was intended to see how once-recycled and twice-recycled filament stacked up against fresh material. Although the research was unable to be finished before the semester's end, the ramifications of reusing PLA material for 3D printing could be huge.