Project data
Authors Lafe Spietz
Status Designed
Completed 2020
Instance of Geometron Trash Robot
OKH Manifest Open Know-How Manifest Download
Upload your project too!

The project documented here consists of three parts, which will ultimately be documented separately but which are presently going to be in one article, focusing just on the Trash Robot to get started. Geometron is a geometric programming language. It uses symbols, written in the language itself, to represent geometric actions either by a physical machine or by a geometric virtual machine if it is in software. Trash Robot is a system for using upcycled materials to build robotics for manufacturing in a "two and a half d printing" system to be described below.

What is documented here, specifically, is a printer for printing coins out of clay built from a Raspberry pi, some simple electronics, and some cardboard and plastic trash. The details of the Geometron language are documented elsewhere, as is the social media documentation software used to self-document the system.

Problems Addressed and Methods used[edit | edit source]

Most fabrication technology imposes strict requirements on what it can work with, limiting choices for material used as input. The ideal input material for sustainable industrial production is a waste stream with the minimum possible modification. While technologies like CNC machining and laser cutting can take a much wider range of input materials than 3d printers, they are still limited to materials with a fairly uniform shape and texture. Also, the consequences for using a wrong material in CNC or laser cutting can be high: the laser can start fires easily if we change materials and don't get the settings just right, and a CNC can cause dangerous failures if used machining the wrong types of object(e.g. stone). We therefore seek to find technologies which can interact in the most general way with work materials, poking, pushing, gouging, chipping, burning or melting with a hot tool, etc. We also want our technology to have some "give" to it, where if a tool is pushed against something harder or softer than expected it at least does no harm.

Fabrication technologies are generally divided into additive and subtractive. For example, 3d printers add material, and CNC machines remove material. What is presented here is an example of "transformative manufacturing", where a material is manipulated with a tool, rather than added or removed. The two examples given here are clay and polyethylene. By poking a nail into soft clay, one can displace the material without adding or removing any, and then it can be baked to make a hard solid object shaped by the poking(or pushing or gouging or melting) process. While it is possible in this way to make complex three dimensional shapes, we present it primarily as "two and a half d printing", where we can vary depth of a depression poked into a surface, but mostly just control the x and y position. Using a tool to modify a material by poking at it opens up a very wide range of materials which can be manipulated. A heated tool can reshape thermoplastics without melting the plastic other than right at the location of the tool. A vibrating tool could chip hard and non-machinable materials like stone. All of this is pointed toward reducing the loop size of upcycling waste materials into fabricated products. Rather than turning plastic waste into pellets, then filament, then a printed object, transformative manufacturing can use the minimum possible time, energy and most importantly complexity to impose order on waste material.

The robot presented here uses thin plastic sheet to create some springiness to the action, which leaves a huge margin of error compared to the build plate of a 3d printer or the uniformity of a machinable sample. Tools are held in place with magnets, allowing modular swapping for any tool.

A problem with most automation control technology is that it either requires users to use some technology that does not allow for direct machine control or requires the user to learn far too much in order to control it. In the example of 3d printers, users design parts using 3d design software, then a separate slicer program turns that into the code which actually controls the motors. This works if the whole system is designed for the purpose of just 3d printing, but if a user wanted to do more general automation with the stepper motors in the printer, that direct control is not part of normal operation. In the opposite extreme, CNC machines often require real coding which looks much like most computer programming in that it requires complex arithmetic, extensive use of English, and memorization of technical jargon in order to use. Replacing conventional computer code with the symbols of Geometron allows people with no advanced math skills and no familiarity with English or any technical jargon to fully control automation systems of any kind.

Another problem addressed in this work is the replication of open hardware. Most open hardware still has a largely centralized model of replication in which people share the designs in a centralized repository, share it using mass media, and then hope other people will copy it. But how well is this really working? If a new open hardware technology is shared on a popular tech blog and is shared 10's of thousands of times, how many of those readers actually replicate the system? Very few. Our approach is to decentralize the information documenting the technology to the point that each robot contains all the information required to replicate itself in a web server users can see over a local wifi network. Thus the information can be edited by users as they modify the system. This whole system self-replicates, with code copying itself from a central code repository to the robot, from the robot to a local hard drive of a laptop, and from there to a central repository again, replicating the entire system globally and locally without any central repository.

Finally, a problem with open hardware is how to finance it(given that investors are usually looking to build IP). The main approaches are to build machines to sell for profit, to build machines with a grant of some kind from either a government or non profit, or to fund out of "spare funds" of hobbyists. We take a different approach with Geometron, building media which is designed to transmit symbolic capital, essentially building its own economic system. By printing coins that carry meaning, we can use those to circulate within a community to stimulate commerce and also non commercial productive activity(donation of waste materials, labor on the robots etc), and by building our own micro economy, the machine essentially self-finances, building its own incentive system to encourage replication. Examples of media with symbolic capital are objects like belts in martial arts, or medals and certificates. There is a non-zero financial value to a piece of paper with ink on it or a cloth belt with coloring, but the main value in paper certificates or martial arts belts in what they mean, which is a non-numerical and non-financial form of value. The Geometron Trash Robot system can be used not just to directly fabricate media, but to make design files in SVG format which can be sent to a laser cutter or in .stl format which can be sent to a 3d printer. Laser cut media can then be used to replicate itself onto cardboard with magic markers for even lower cost transmission of symbolic capital, and 3d printed objects can be used as stamps to copy symbols into soft materials like mud or used to make stamps to replicate on paper, again lowering the cost of symbolic transmission.

Project goals[edit | edit source]

  1. Build fabrication technology from trash, which can take trash as its feedstock to produce useful products entirely from trash
  2. Build a geometric language for programming fabrication machines which allows users to program automation systems without any math, English, or understanding of computer programming(all programming is symbolic)
  3. Build a platform for control of robotics which also serves as an information server on a local network, fully documenting its own replication. This system has capabilities for creating text documents using Markdown, ability to create vector graphics for technical illustration(also in the Geometron language, all running on the server), and several other open and free document creation capabilities. Users can log onto the local network, even without global network connectivity and share and remix instructions for machine replication. The Geometron server system all runs on top of Apache on the Raspberry Pi.
  4. Build self-replicating physical media. The media produced by the fabrication system must contain information, the equivalent of marketing materials in a for profit venture, which communicate information to users which will convince them to replicate the system, in order to stimulate growth. To this end, the machine creates "coins" made from clay, which can be printed with arbitrary symbols. These coins are aesthetically pleasing, and can use the "coin" metaphor to create value which can be shared similarly to currency(but with no monetary value)--much like "challenge coins". An operator of a Trash Robot can keep printing coins until they find one that people want. When people want more coins, eventually they'll exceed the demand of one printer and this will stimulate the construction of further printers.

Bill of Materials[edit | edit source]

  1. Raspberry Pi and associated parts(screen, keyboard, mouse, power, SD card) $50-$150
  2. 1 custom circuit board $1 in batches of 10 or more from any board house(Eagle files provided)
  3. buttons and headers $2(digikey, newark, allied)
  4. 4 pin ribbon cables $3 (Samtec)
  5. 40 pin raspberry pi header $1(Pololu robotics)
  6. MP6500 stepper motor controller board, 3 x $6 = $18, from Pololu Robotics
  7. 12 Volt barrel connectors $1 in packs of 10(Amazon)
  8. a few square feet of corrugated cardboard scrap
  9. thin HDPE sheet, as in a milk or juice container
  10. a stack of small rare earth magnets
  11. one small nail
  12. duct tape
  13. glue
  14. three DVD or CD drives of the kind used in desktop computers. $5 on eBay x3 = $15

Tools[edit | edit source]

  1. soldering iron
  2. solder
  3. pliers, wire strippers
  4. multimeter (need to measure both voltage and resistance)
  5. screwdrivers, both tiny flathead and medium size Phillips head
  6. box cutter
  7. Elmer's glue
  8. Gorilla 2 part epoxy
  9. duct tape
  10. ruler
  11. markers and/or paint pens
  12. scoring tool like a fork

Construction[edit | edit source]

  1. gather materials
  2. build electronics
  3. set up raspberry pi server
  4. build robot mechanicals from cardboard and plastic
  5. finish assembly, build enclosure, test, operate, share

It is not our intent to describe every step of the construction here, since that is a duplicate effort. The system is intended to be self-documenting: each instance of the robot has a webs server which shares the full replication path with all local users over a local network connection.

Page data
Type Project
Keywords trash, robot, upcycle
SDG Sustainable Development Goals SDG09 Industry innovation and infrastructure, SDG11 Sustainable cities and communities
Authors Lafe Spietz
Published 2020
License CC-BY-SA-4.0
Language English (en)
Issues Automatically detected page issues. Click on them to find out more. They may take some minutes to disappear after you fix them. No main image