Open Source Ecology Forums

Matt_Maier

I've noticed a couple of problems:

1) it's difficult to find people with extensive subject matter expertise who are also willing to give their work away for free

2) open source hardware requires actually being physically present for prototyping and testing

For example, I think the GVCS is a great idea, but I can't do the professional-level engineering required to design, prototype and troubleshoot a small aluminum factory. I'm not even sure there are many people in the world who could just show up and start making that design a reality. I bet the only way to even move forward on that design is to find several experts with overlapping expertise...but getting them in one place would be problematic.

My suggestion is that the OSE guys seriously consider developing software in parallel with the hardware and electronics. There are already great open source tools for managing collaborative projects, like ]project-open[ and I'm sure there are at least a few ways to collaborate inside a program like solidworks, but I don't think there are any good open source solutions to having multiple people across the internet collaborate on a physical design. This is mostly due to the fact that open source hardware, as a community, is still a new idea.

I think the GVCS is a perfect opportunity to start developing the software tools necessary to harness the collective enthusiasm of participants who can't actually be somewhere in person. For example, the Reprap and Makerbot communities have dramatically accelerated the development of programs that take 3D models and turn them into g-code and run the printers. I think that adding an open source physical prototype collaboration program to the list to-do list would pay dividends later as it would increase the rate at which the machines are designed.

I imagine it would be far easier to find volunteers to show up and build a machine from specifications that were worked out previously, since it takes a lot less expertise and the results are immediate. Additionally, it would ensure that all the necessary tools and raw materials were already on site, so no one has to sit around waiting for some critical part to be overnighted.

It might looks something like this:

step 1) a new project is created and titled Small Scale Aluminum Factory

step 2) reference documents are collected and stored in the project files

step 3) one or more people log in and start brainstorming designs

step 4) certified experts log in and start designing the 3D prototype

step 5) when the project lead is satisfied with the first prototype, the build files are exported

step 6) the Farm assembles all the parts and tools listed in the build files

step 7) volunteers are scheduled for an intensive building and testing session

step 8) full-time project managers record all the lessons learned from the first prototype

step 9) the results are incorporated into the second digital prototype, and new build files are exported

step 10) return to step (6)

I think the key to this process is the existence of the software allows OSE to harness the energy of expensive experts who have enough interest and free time to help with the designs, but not enough to actually devote themselves to finishing the project. The really important part is that some people already know all the "gotchas" in designing and using an aluminum factory (or whatever) but they're only necessary to a small part of the project. The rest of the project can be handled by motivated and organized hobbyists. You'd be wasting the expert if you required him to show up and build his own shop out of bricks before he could get started. Additionally, it's best to have the machine being built and used by non-experts right from the beginning because you'll run into all the stupid mistakes right away, so fixing them will be cheapest. If you can find some experts to personally design, prototype and test an aluminum factory then they'll produce a tool for other experts to use, rather than illiterate farmers.

This software tool would probably be slow in coming, but even modest gains in its capabilities would start to multiply the productivity of the project.

Or I suppose OSE could just become a general contractor.

Mike_Apostol

Hi Matt,

I couldn't agree with you more. I've already started to look into some PLM software like ARAS for project management and CAD management. For the next CEB Press version I'd like to try and implement this type of approach to it even if we don't have a server with PM software on it. We can kind of approximate the normal design approach with a wiki, just a bit more work and messy.

We're already started to amass a group of engineers and CAD designers with different specialties over in the CAD Team page (http://opensourceecology.org/wiki/Development_Team/CAD_Team) who can be asked for input on design as it progresses. Additionally we have replicators who have built these things who can also provide valuable input.

I'd like to experiment with different methodologies and tools for organizing the design process. I agree with your design approach principles but the trick is to figure out a coherent, simple, free way of organizing all the data (CAD/reference documents/specs/etc) so it doesn't get bogged down.

I'll probably be attempting to make a standard wiki template for design methodology to use in the interim but am certainly open to other methods.

Controlling who can and can't contribute has it's pros and cons and I'm a bit torn right now. I think ensuring anyone can contribute is important, but vetting out the designs based on technical knowledge and design understanding is critical as well.

Matt_Maier

There are a few examples of this idea, like DARPA's AVM / iFAB http://arstechnica.com/business/news/2011/12/darpas-factory-of-the-future-looks-like-open-source-development.ars and Local Motors http://www.local-motors.com/

Interestingly I'm having a hard time finding specifications on how much of the project is open vs. closed source and how much of the decision making is done by professionals as opposed to collaborators. It sounds like DARPA at least is basing the design decisions on a pre-defined entirely objective equation. That solution, however, presupposes software powerful enough to analyze the design based on an equation. I doubt they're going to make their META tools open source. Maybe they'll make their cyber/electro/mechanical description language open source.

I agree with the coherent, simple, free criteria. Although i think we'd need to be willing to compromise somewhat on the "simple" part. It seems like there should be reasonably obvious boundaries between the components of the toolchain. For example, only a few people are going to be at all competent at modeling the designs in CAD, so it makes sense to not require that program to be easy for anyone to understand/use. The same applies to the project management activities. On the other hand, gathering reference documents, and distributing reference documents, and building finished documents, and performing the actual tasks, should all be low-barrier-to-entry.

For example:

Document management program: wiki; anyone can browse it, anyone can edit it, anyone can pull information off of it.
This program would be used for gathering suggestions and reference documents from anyone who wants to contribute them.
It would also be used for communication between all collaborators.
It would also be used for compiling the finished documents.
It would also be used for representing the project to the outside world.

CAD program: (suggestions?); ideally this would be collaborative, anyone would be free to browse but permission to edit would be moderated.

This program would be used for creating and managing the 3D prototypes.
Hopefully this program would be able to perform at least rudimentary analysis.
It would also be used to generate the diagrams and blueprints and illustrations

system engineering program: maybe project-open(?); this would be used exclusively by the full-time project managers

This program would be used for keeping a diverse array of full and part timers on task
It would also be used for projecting future needs and capabilities
It would also be used for generating complete project templates that anyone could use to start from scratch

blade

I was thinking about similar issues. I posted on CollabCAD, which has a distributed CAD component interfaced with ERP/MRP software. It isn't as ambitious as what you're proposing, but it has some of the elements you are proposing. Unfortunately, it is not open source software though it is based on open source tools.

Howdy, Stranger!

Categories

In this Discussion

Tagged

Open Source Ecology Forums