NASA Grant: Location-based, Auto-stabilized, Handheld, 3D Construction Printing tool

Based on current precedent, I set out to build a better construction 3D printer, and ultimately utilized human movement and Computer Vision to do so. Mark I was originally developed for my Brown/RISD MADE Graduate capstone project.

Complimentary Materials:

Mark II was made possible by a generous grant from Brown University and NASA RISG. In this iteration, I implemented a Computer-Vision-based auto-stabilization system to integrate with the existing AR interface.

To accomplish this, I taught myself Unity, Oculus XR, C#, python, IoT development, the basic tenets of rheology, and honed my geometric dimensioning and tolerancing acumen. I also learned how to stream live data over UDP and HTTP, developed a working knowledge of stepper motors, drivers, 3D Printer boards, and the Marlin firmware. Finally, I used existing NASA research as a basis to reproduce an extrudable sulfur and regolith-based “MarsCrete”.

Mark I

Mark II

Mark I

The concrete extrusion tool utilizes Computer Vision to know where it is in space, and shows the operator in real-time where they need to go and what they need to build, layer by layer, through a simple Augmented Reality interface. 

Click to pan 3D model

To design and build a successful prototype, I thoroughly researched construction printing, rheology, computer vision techniques, ergonomics, and inverse kinematics.  I also extensively tested different extrusion materials, and human motion ranges.

Mark II

The extrusion tool now utilizes the Oculus’ locational data to provide real-time autostabilization through stepper motors in a Delta formation, all while continuing to utilize CV and AR to direct the operator and live-correct for their imprecise movements. 

Click to pan 3D model

I built on what I learned with Mark I, and delivered the location of both the intended geometry and the extruder head location, utilizing a Delta 3D printer board and firmware to close the gap.

Since early 2020, construction costs have risen 31% and over 1.2 million construction workers left the industry.  This has increased industry adoption of 3D Construction Printers (3DCP), and boosted companies like ICON3D, COBOD, and WASP.  This push has also led to adoption by SpaceX and NASA for future Lunar and Martian missions.

Leading 3DCP companies advertise improved costs, time, and efficiency, but they do not take into account issues like long setup times, logistics with large machinery, complex interfaces that necessitate operator specialization, and limitations in print area size.  These issues ultimately prevent 3DCP from gaining any practical advantage over traditional construction, and relegate it to niche status.  But what would it look like to overcome these issues?  We’d need to create a tool with similar print accuracy that is quick to set up, mobile, easy to control, that works independently of other systems, and that is infinitely scalable.

I spoke with many field experts to gain valuable perspective and insight into what the needs of such a tool might be, and how best to achieve those needs.  I also connected with a multitude of prototyping experts on how best to physically and digitally meet these requirements.  

From these conversations, I realized that I would need to create a tool with similar print accuracy to current 3DCPs that is quick to set up, readily mobile, easy to use and control, that works independently of a larger system, and is infinitely scalable.  My goal therefore became designing a better 3D Construction Printer for mass adoption by the industry that is smaller, positionally-aware, and that utilizes a simple visual interface.