CubeSpawn, an Affordable, Distributed, Open Source, FMS (Flexible Manufacturing System)

2»

Comments

  • CubeSpawnCubeSpawn Posts: 32Member ✭✭
    I have been out of circulation for a while, My partner and I were accepted into the PlugAndPlay Tech Center in Sunnyvale, California for a 90 day Bitcoin Accelerator/Incubator, but I am hopeful that this effort will yield a revenue producing company that can help fund the army of DACs we need to get all manner of services baked into the 'net Autonomously - and process trash to energy, and trash to trinkets, among things!
  • harveybc0harveybc0 Posts: 29Member
    edited May 2014
    @CubeSpawn you will need a system to classify, access and trace the evolution/forks or derivated designs and versions, this system has to be infinitely scalable, and you will need a coordinate system for navigating in it. The designs are shared resources, pretty cool.
    I think all the concepts used in my project can be used in yours.

    By the way, Great Project, do not surrender until you finish it. If you need something, tellme.

    http://forum.ethereum.org/discussion/882/congratulations-humanity-you-have-reached-singularity#latest
  • harveybc0harveybc0 Posts: 29Member
    edited May 2014
    You will also need a way of make connections betwen related designs or parts, i mean creating connections from one taxonómical category to other for compatibility or for use in a bigger machine, i do that too with my categories as they are represented by the nodes of a neural network called the AI taxonomy. The coordinate system i use is the turing tapes mentioned as they are coordinates for a fractal object (the category).
  • CubeSpawnCubeSpawn Posts: 32Member ✭✭
    @harveybc0 I have read through your project and while I do not pretend to follow your logic in its entirety, I have a general feel for the desire to harness the computational power for narrow, evolutionary AI.

    I applaud your efforts, although I believe that complex things can be achieved using a simple, incremental approach to building hierarchical (but stupid) logic that initially asks higher logic for answers to undefined problems, at the top of this hierarchy are human brains (for now)

    As the systems are used in the field, and encounter more complex problems, hand built solutions to the range of common problems a machine cell encounters will accumulate, making them incrementally "smarter". (meaning a nuanced response tree with branches for each variant of a problem)

    While a long and ambitious project, this would provide narrow AI with an ideal comparative environment for training solutions, since a "good enough" solution would be an ideal comparitive standard for machine generated solutions to "compete" with practical logistic, mechanical, workload routing, and a plethora of other problems the system must solve as it grows more complex.

    To address your other observation regarding a system components heirarchy, again I have a simpler preliminary solution - versioning and dependencies, much like github, or perhaps even adapted to github to keep track of system parts, variants and modified versions as well as dependencies. (sophisticated CAD can do this too)

    The Big goal is to make creating physical objects as easy as running a compiler, in fact, you could say CubeSpawn aims to be an Atomic compiler, the way you'd implement one before you had direct molecular placement technology ;-)
  • harveybc0harveybc0 Posts: 29Member
    edited May 2014
    I will use your project if i find a way to link our category systems for the storage of reusable designs of parts like actuators or structures used for hardware interfaces to use Singularity experts in field devices. I am very interested in your approach to the design and implementation of the project.

    I also applaud your efforts, this is a very good project, and thanks for your information.
    Post edited by harveybc0 on
  • CubeSpawnCubeSpawn Posts: 32Member ✭✭
    I am initially applying the "rule of five" wherein any range of functions (Stepper motors, for example) can decompose into 5 general classes (tiny, small, medium, large, huge/or custom) these then provide correlation to actual NEMA sizes, and minimize part counts by exclusion of subtle, but non-significant differences, this will speed recursion, since it enforces a simpler model on procurement and manufacture and reduces part counts.

    I like the fractal concept, and think at a more advanced iteration of the system, that will come into play, but for now all solutions are a compromise between existing systems and their size relationships and a more theoretical design framework.

    Your framework sounds useful and interesting - I will monitor whatever you publish and look for parallels going forward - James
  • GesamtGesamt Posts: 5Member
    edited June 2014
    These concepts and working examples are great! I've been thinking about these similar ideas for some time as well, but it's really cool to finally see different people actually working on them.

    The idea of a parametric and version controlled distributed manufacturing system, with a form of DAC to control and search, where the network can accept 'voted' contracts and inputs which provide conduits to algorithms for analyzing CAD or manufacturing inputs as well as process data and material science libraries, to form and then optimize evolving taxonomies of all characteristics of manufacturing and design... would be spectacular.

    If there was a way to form patterned relationships in databases of 3D cad data just as a simple and overused example, like taking an open set of models like Trimbal 3D Warehosue, and then referencing patterns in parametric design similarity, there could be some amazing implications for that ability across more sophisticated sets of data related to materials science, fluid dynamics, etc, to enable a kind of Ai for designing based on a limited input parameters over huge sets of data.

    Almost as if a spider of Google or Wolfram Alpha knowledge inputs were possible, then a lay-person could request "Jonathan Ive high performance motorcycle" and also specify "Advanced Search" to narrow the mashup parameters like "hydroforming + CFD exhaust analysis and optimization suggestions engine performance and acoustic dynamics". At first a crude 'search and blend' system would be terrible and reference design parameters which don't satisfy any of the needs of any intended ideas of the user, but eventually, the AI system would improve and search on top of these underlying manufacturing DAC could enable a form of design, research, and system implementations (like enterprise resource planning and PLM) that are optimized for working examples, per search, and equates to a 'boilerplate' which satisfy these basic search inputs. Then the physical designs and related manufacturing system can output the formats needed, produced by these algorithms, and then users could still manually reworked designs into something where all process concerns are known and almost ready to manufacture in 'production'.

    This is a broad poke at a huge bubble of concepts which I'm sure are the basis for a lot of similar science fiction writings I haven't read. But it's cool to see what you guys may be underpinning for the future!

    Some day, this kind of search-design AI system could uncover some effective patterns and find efficiencies in process we haven't discovered yet. These autonomous design search engines could satisfy rules of the existing distributed manufacturing network, material science and general physics and give humanity a tool for developing a physical reality which is representative of a collective human design intelligence.

    Hopefully these kinds of ideas doesn't produce results like the main page of YouTube, but I imagine this capability will happen, and projects like CubeSpawn and others are the beginning.

    Thats my rambling about similar ideas.. but good luck, I look forward to seeing more progress! And keep up the great work!

    I also imagine that visualizing these systems for manufacturing will also be an important driver to identifying areas where DAC and processes can improve. I've animated this webGL model of my own cnc router and enclosure I've built, just shows the range and modes of function. This is a brutal hack I've been putting together, but there are far more sophisticated platforms which will do a better job of visualizing distributed manufacturing capability once manufacturing can be setup on line and not simply negotiated by email, phone and skype. http://media.klampwave.com/cncenclosure_rough_edge_9.html
  • CubeSpawnCubeSpawn Posts: 32Member ✭✭
    @Gesamt
    Thank you for your thoughtful post! if you haven't done so, you may wish to take a deeper look at ROS (Robot Operating System) it has some very nice tools that can simulate a 3D model, from a sketchup file if desired.

    It will require a bit of learning curve (of course, what doesn't?) but you can draw in sketchup, convert to a dae, write a metafile (a URDF - Universal Robot Descriptor File) to tell the simulator how all the "joints" move (linear in a CNC router) and animate it from a control stream (such as G-code, in your example)

    this is VERY powerful open source software,

    as an update to this overall project, I am working on getting the designs for the hardware and open control stack published here: http://www.openbuilds.com/builds/cubespawn-3-axis-mill-module.758/ and
    http://www.openbuilds.com/builds/cubespawn-ultimaker-3d-printer-module.782/

    I will also be placing all the 3D models for the parts here: https://3dwarehouse.sketchup.com/search.html?backendClass=entity&q=cubespawn
  • CubeSpawnCubeSpawn Posts: 32Member ✭✭
    I have condensed all the relevant links to one external page for easy reference. http://goo.gl/N9N86f Also, we will be crowdfunding for a new cube design shortly, and have entered the NASA Tech Briefs contest here: http://goo.gl/xXlmfT
  • CubeSpawnCubeSpawn Posts: 32Member ✭✭
    An update: I have found a partner to help build out the rest of the prototypes, pictures of the project are here:

    https://picasaweb.google.com/103828779781480193226

    and more specifically, the latest activities are here:

    https://picasaweb.google.com/103828779781480193226/DougSShop?noredirect=1

    A major revision of the website is underway as well.
  • CubeSpawnCubeSpawn Posts: 32Member ✭✭
    Much progress! two experienced engineers have become involved in the electro-mechanical portions of the system, we have two ROS experts onboard, MachineKit is in the mix and after a meeting with the creator of MTConnect, a ROS implementation has turned up - the technical parts of the system are enough to blow away any resistance to progress!!

    once the technical implementation wraps up, the Ethereum DAC design will begin in earnest!!

    -James
  • CubeSpawnCubeSpawn Posts: 32Member ✭✭
    A small update: Everything takes much longer than anticipated, but I have never stopped pushing this along, the ROS/Machinekit/MTConnect solution is coming together and the hardware is finished to the degree that documenting how to build one is now about all that is left. I do expect another 6 months of work until the DAC becomes the primary focus, but it surely will!

    Latest Pix are now here:

    https://picasaweb.google.com/103828779781480193226/DougsShopPhase2?noredirect=1

  • StephanTualStephanTual London, EnglandPosts: 1,282Member, Moderator mod
    @CubeSpawn - love the new pix/videos! Keep it up!
  • CubeSpawnCubeSpawn Posts: 32Member ✭✭
    I have posted most of the basic models for the prototype build-out here:

    https://3dwarehouse.sketchup.com/by/CubeSpawn

    so that others can begin building inter-operable Cubes and we can get the automated factory of the future into ALL the hands that need it most!
  • CubeSpawnCubeSpawn Posts: 32Member ✭✭
    Another Small Update:
    Picasa is going away at some point, so the pix have morphed to google pix:

    https://goo.gl/photos/cbL7RQsjnGirBNya7

    I have Blogged a bit regarding progress and status:

    http://cubespawn.blogspot.com/2016/03/marching-onward.html

    Much work has been done, though it looks small from a distance - I am looking forward to beginning on the two final foundational elements: a DAC to manage the system growth, and deep learning to let the system improve its own performance.

    Then I can rest ;-)
  • CubeSpawnCubeSpawn Posts: 32Member ✭✭
    Just a little Bump: the project is alive and well, looking for investors and expertise on creating a manufacturing token on Ethereum within the next year, to allow arrays of machines to pay for materials, energy, and royalties on digital templates for parts, assemblies, and machines!:

    New Pix here: https://goo.gl/photos/cbL7RQsjnGirBNya7
  • ECOdECOd Posts: 4Member
    Free Litecoin faucet up to 1 000 000 Litoshi per day!! :)
    http://baglitecoin.com/?r=41988
  • CubeSpawnCubeSpawn Posts: 32Member ✭✭
    edited September 7
    Another little bump, 5 3D printers in the system, Milling machine design being tested, lathe on the second version all that is left for the genesis machine set is the pallet movers, loading robot and pick & place Cube ... https://goo.gl/photos/cbL7RQsjnGirBNya7 Building hardware IS hard, but not impossible!
Sign In or Register to comment.