At the time of development, at home 3D printers require constant surveillance and hands on intervention to clear and clean the print board. We saw this as an opportunity to allow users to be able to print multiple objects consecutively with proper cleaning between prints, as well as video surveillance and the use of AI to detect failures.
What it does
Our system detects when an object has finished printing, moves the print head away, adjusts the bed, and lowers a mechanical scraping arm. Then, the arm with a sharp edge clears the board as the print bed passes underneath it. Following the completion of this task, isopropyl alcohol dispensed from the arm is used to clean the bed sufficiently and prepare it for the following print.
When a print is running, Octoprint hosts a webcam livestream, keeping the user updated on the status on their print. In addition, we've implemented the AI plug-in "Spaghetti Detective" that uses image detection to detect anomalies in the print. If an irregularity is detected, it notifies the user and stops the print, preventing dangerous accidents such as printer fires.
How I built it
Our mechanical arm is moved by a Servo. It features a tubing system running down the length of the scraper, which dispenses the isopropyl alcohol used to clear the board of all prints. The tubing system dispenses isopropyl alcohol using a diaphragm pump. We designed the parts of the mechanical arm in Solidworks and 3D printed them ourselves. We used a Raspberry Pi to run an Octoprint server, which interfaced with the printer. The raspberry pi includes a custom pcb for voltage regulation. The Octoprint server hosted the plugin Spaghetti Detector, which connected to a proxy server running an AI pipeline monitoring the print for failures through a webcam.
Challenges I ran into
Some challenges included managing our time to ensure all our mechanical parts were 3D printed on time, having adequate power deliver circuits, and ensuring we had ensuring all our wiring was in the correct order.
Accomplishments that I'm proud of
Some successes we experienced while building our project include a number of valuable features such as bed cleaning with isopropyl alcohol. In addition, we're proud to have included AI in our hack, which we didn't originally plan on using. However, it drastically improved the overall hack, maintaining user safety as an engineering priority as well as providing additional functionality.
What I learned
Our team members learned new things, especially the beginners. One member learned a lot about raspberry pi functionality and server systems like Octoprint and how they work. Another member was working with GCode, a language he was unfamiliar with to program the movement of the 3D printer. Overall, part of the learning curve was creating harmony between our various software and hardware components and merging preexisting 3D printing technology with our own builds.
What's next for AutonoPrint
Currently, all prints require a chamfer at the base in order for our scraper to cleanly remove the print from the board. However, recent materials technology developments introduce the possibility of using dissolvable printing material, which we hope to include as an auto-implemented feature to all prints using our system. Thus, all prints, regardless of whether or not they feature a chamfer, can be removed by the scraper with ease without compromising our users design vision.