Rapid Prototyping Approach
OSE rapid prototyping refers to 1-day builds and rapid 1-2 day prototyping runs that follow weeks of preparation. This is a technique that we follow at OSE pending complete preparation of designs and fabrication procedures. We have demonstrated proof of concept on the 1 day build with the 1 Day Build of the CEB Press, and we have demonstrated the 1-2 day build of a machine prototype with the Ironworker Prototype 3.
We have learned over the years that we can either spend the time building or spend an equivalent time - and typically lesser time -designing for a rapid build. This is especially true for heavy machinery - where it is easier to prepare everything on paper prior to a build, up to language agnostic instructionals. This level of preparation allows builds to happen once, and only once - instead of parts being assembled and reassembled repeatedly, or mistakes being made repeatedly. This is especially true for down-the-road prototypes - where we are building on prior experienc - and less true for initial prototypes, with which we have little experience. For the mechanical part of the GVCS - since we are constantly reusing similar parts and have a lot of collective experience with it - it is critical that we spend more time on paper - because we have gained the ability to plan out builds rather meticulously by this point in the project.
This is feasible because we are simplifying the build stock and build procedures to the simplest possible design which still allows for peak performance.
Then after a build - we can iterate rapidly to make improvements - because our working materials make up a reconfigurable Construction Set.
The key to the above is module-based design. While it is much more difficult to build an entire device, it is much more tractable to build a module (or component of that machine). Only in the absence of skill or under time pressure do we not go through the full design process on paper, when we attain a lower level of quality, while still achieving a proof-of-concept.
Because we are working on many proofs-of-concept that have never before been attempted, it is more important to have something done rather than having something done perfectly. That is - done is better than perfect. This approach is taken because the thing at hand may not work - so it is important to determine early on whether something works or not - in order to take mid-course correction. This way, innovation happens in the span of days, not months. While we begin with a study of industry standards, our methods are nontraditional - so it is typically difficult to find fabricators and experts who do not have bias towards the status quo when evaluating our methods. Yet the project has depended on breaking convention, so this introduces a challenge to the review process.
Wikispeed has attained a 100 mpg car proof-of-concept in 3 months, as opposed to the years of typical development cycles. See Wikispeed.
OSE has attained a 12 hour build of the Ironworker Machine Prototype 3, compared to the 6 month build of Prototype 2 - by radically shifting the design approach. Yes - let me repeat - 12 hours instead of 6 months - for a proof-of-concept machine that can cut 1"x8" steel.
And this is only the beginning...
It is important to point out that the GVCS is being designed as a Set, not as individual machines. That means that a product ecology must be considered.
Numerous people have commented Why don't we build something right the first time instead of rushing? Our response is that the essence of rapid prototyping is to fail more to learn more - or otherwise spend months instead of days learning something new. Our response is further - Yes, we would gladly do that if our timeline was 100 years." However - our criterion for success is attaining robust functionality of the entire set by the end of 2015. This calls for a velocity that most people cannot wrap their head around.
The requirements of GVCS being a Set means that any machine built informs the building of other machines. It is important to note that we are currently at the phase of showing proofs of concept - not finished machines. Finished machines come after many iterations that build on a whole set of proofs-of-concept. The typical process is:
- Partial machine proof of concept
- Whole machine proof of concept.
- Rapid iteration of Construction Set design method for functional prototypes
- Field use and further refinement of functional prototypes
- Build optimization to 1-day production runs with Rockstar Team
- Optimization of machines for 1-day build and high performance
- Development of language agnostic instructionals for 1-day production runs by non-Rockstar Teams
- Performance optimization
- Final product release
Each one of the steps above poses unique design challenges. So next time you see a broken machine at Factor e Farm - do not lose hope - just recognize that we are doing the required due diligence to pack a significant number of innovations into a single machine - tho OSE Innvoation Stack. And ask yourself if you understand which of the above 10 or so prototyping stages we are at. And notice that because we are using a lifesize Lego Set for construction - it is easy to iterate on day - not month - timescales.
When we say OSE Innovation Stack - take LifeTrac as an example. Nobody in the world is using modular wheel units that can turn, for example, a microtractor into a trencher. Nobody is using a stackable power unit that can be replaces in minutes. Nobody is using quick connect valves - where valves can be switched between machines. Nobody is using a scalable framing system which allows one to convert a 27 hp microtractor to a 200 hp bulldozer. We have shown proofs of concept of each of these elements.
If you are getting involved in OSE, it is important for you to recognize that you are entering the project that runs at Marcin's pace - and that your role is to support that pace as much as possible for the reasons stated above.