"Open Source" Drive Systems

[AdamHeard]

I'm not sure how to feel about this...

sure, it would help a lot of people out, but there is really no way to prevent people from making it exactly to print and not making any changes or improvements.

On the other hand, it could cause a lot of inspiration. Over the years, I have asked for the CAD files to various mechanisms and parts from various teams to look at; originally my plan was to copy them 1:1, but I always ended up starting from scratch and coming up with my own design. In the begging I was lost, but by the end I knew my part completely.

There is potential here, but it seems like it would be too much work to implement some sort of control over distribution.

[Billfred]

I believe it was Dr. Joe who first said (or at least got spotlighted saying) that there's a high correlation between a solid drive system and an inspirational season.

A lot of people griped in 2005 with the Kitbot that everyone's robots were looking the same from about eight inches down. To me, this is a good thing--if you followed the instructions, you would have a solid drive system. This is the bar--you will move.

Now, let's say this open system is released freely with good results. Matter of fact, let's say it eclipses the Kitbot in popularity. All of a sudden, this is now the bar. You've now established the basic standard of mobility. Someone will figure out a tweak that lets them get just a bit more out of the drive system, and uses it to kick butt in one season (and I'm perfectly willing to give them a season to enjoy the spoils of their own developments). Once that spreads out, someone will come up with a new trick. This desire to keep up with (and one-up) the Joneses is what will cause folks to want to improve upon the setup as it sits. Hopefully, they'll be awesome enough to share.

This makes me wonder, what would the general aim of the drive system be? Minimize fabrication? Reduce cost? Toughness? Lightness? Any consideration for obstacles? There are a lot of paths in drive system fabrication, and there are often several right answers (see also: the 2006 world champions with 2WD, 6WD (I think), and treads).

[Alex.Norton]

Originally Posted by Billfred

This makes me wonder, what would the general aim of the drive system be? Minimize fabrication? Reduce cost? Toughness? Lightness? Any consideration for obstacles? There are a lot of paths in drive system fabrication, and there are often several right answers (see also: the 2006 world champions with 2WD, 6WD (I think), and treads).

From what I see here I don't think it would have a specific aim or that there would be one. You could upload your own drive drive that perhaps optimizes weight but is really hard to fabricate but somebody else could upload one that has a completely different focus.

I like the idea of membership but I would argue more for using whether somebody is uploading and improving more that I would use reputation (for obvious reasons). While I make a lot of CAD files and have uploaded pictures of them I just don't comment that often and as a result don't have much reputation. I would feel that I would need to start commenting uselessly of Chiefdelphi just so that I could use the CAD system.

I hope you get this going, it sounds really cool.

[Astronouth7303]

I don't usually contribute to these discussions, but I think I have a tidbit to add.

What a team gets out of this library will depend entirely on what they're looking for. If someone comes in looking for a drivetrain for their robot, that's what they'll get. If another comes in looking for inspiration, that's what they'll get. It's all dependent on their attitude.

What this tool will hopefully do is level the playing field. Teams currently have 3 options: the kit drive, AndyMark, or their own designs. The first 2 categories cover tank drive almost exclusively. If you want any kind of holonomic drive (omni, crab, meccanum, kiwi, etc), you pretty much have to do all the R&D yourself. That's a lot of work. This allows the less-than-juggernaut teams to get some of this.

The key to making this work is that everyone contributes somehow. It doesn't have to be CAD'd to perfection. Decent pictures, CAD, ASCIIart are all acceptable. If someone else can readily make it w/o much trouble, then it's acceptable.

My $.02.

[eugenebrooks]

"Open source" as it is appled to computer software means that the user of the software is allowed to compile it, or a modified version, but may not provide a compiled version to others without also providing the modified source code and granting the same rights with respect to the source code. This is a powerful tool to encourage the development of shared software in the community.

For CAD drawings, the maufacture of the part is the equivalent of compiling the CAD source. Why would we be concerned about someone making the part exactly to the provided specifications, notwithstanding the rules that FIRST applied to that this past season? Teams learn when they make something, even from public CAD drawings, so the goals of FIRST are being pushed down the road. They learn something even when they assemble something. Heck, I learned something assembling IKEA tables this weekend.

In the open source world, the key item is the source. If a user of the source code modifies it and provides a binary to others, the user must also provide the modified source and grant the same freedoms. If your goal is to cause sharing of ideas through the CAD source, you can use a slightly more restrictive variation of this concept. Require that someone who compiles a modified version of the CAD files, building a modified part in the CAD context, provide the modified CAD files to the community. This would be a CAD Copyleft, so to speak. Others can then learn from and build on this modified CAD source, providing these CAD files to the community, and the bar is raised for everyone.

Eugene

[Madison]

I think that we've seen far less exchange of specific information about mechanical systems than we have of other pieces of information because there are a significant number of factors which affect the effectiveness and implementation of such systems that first need to be clearly defined before they're useful.

The single largest problem is that very, very few teams produce production-ready drawings of even a handful of parts of their robots, let alone entire systems. One could argue that the most effective systems are those that are most likely to have near-complete models and drawings available, but it may be that these teams also use a paperless process.

Similarly, many teams do not model their robots with an attention to detail that sufficiently fills in the blanks, necessitating an unreasonable amount of effort to support sharing that work with others. Further, if one wanted to go so far as to create a shared repository of this information, there're difficulties in standardizing file formats and naming schemes, among other things.

I understand that you're proposing that some cross-section of the FIRST community join together to create drivetrain designs that address a lot of the existing difficulties with the existing efforts of teams, but that in and of itself has some of its own challenges. I couldn't, in good faith, provide to anyone an untested platform marketed as a turnkey solution for their drivetrain woes. Someone, somewhere, needs to foot the bill in producing and testing these designs in the real world, identifying manufacturing and assembly inefficiences and correcting them, and updating the models and drawings to reflect those changes. It's not an insignificant amount of work.

Beyond that, it's asking a lot of a rookie team (or many veterans) that they be able to successfully interpret modeled components or production drawings. I've worked with some pretty amazing people -- both in FIRST and in industry -- and even they've had trouble wrapping their head around how a virtual mechanical system might operate. Often during the build season, after days of working on a part or system, someone will say, "Oh, I get it," and finally understand what my work represented.

I'm not against the notion of a collaborative effort, necessarily, but I think it's worth pointing out that there's a lot more to it than throwing together some models and calling it a day. What you're proposing here is akin to bringing a product to market and requires all of the pomp and circumstance that goes along with it -- including a bit of congratulations and celebration when it happens.

[Herodotus]

Our team doesn't use any CAD whatsoever when designing our robot, and even when it comes to paper or chalk we mostly just use sketches to get the general idea. Rarely do we make detailed drawings of the robot, or parts for the robot, let alone use scale. However for those teams that do use CAD this could be very useful at building up the information level of all teams involved, so it would be a good idea really. After all, learning is one of the most important aspects, if not THE most important aspect of FIRST, and when you pool knowledge and resources like in this idea it makes it possible to learn even more.

[AdamHeard]

The most practical solution would probably be a depository of .dxf (entire drive assembly and sub assemblies) online that can be viewed with the new .dxf viewer. Teams can submit their designs (a .dxf is no more than a few mb's for an assembly that size, probably less actually) for others to look at in 3d for inspiration. They could also leave a point of contact for viewers to ask questions or even ask for the models to be sent.

So I'm seeing general interest here, so let's take a step forward. I can supply the bulk of the CAD files, as well as doing admin tasks if we get a site running. However, I'm terrible at web design. Anyone here want to maybe design a site where we can host this? I may be able to acquire hosting, but someone would have to work with me on the site design. You can PM me, or email me at ironspork AT gmail DOT com.

[Schnabel]

The TechnoKats have bassically did this with components that are on the drive train for many many years. You can veiw the prints for transimissions here for some of our past designs.

[Greg Perkins]

There are a few simple ways to make this work...

1) Create detail-less prints that can be downloaded as PWF's basically exploding the components (say a gearbox), the prints would show how the component works which would spark imagination for replication. I'm sorry but if Andy Mark gave detail-less prints and said "copy this exactly" it would be near impossible, which leads to difference and ingenuity. What they would have to do would be to take the prints, and create their own versions, using parts that would be easy to acquire.

2) Would be to create some sort of "read-only" file that you couldn't copy or modify. That would let the individual be able to virtually see it and take it apart. This way might lead to the system being hacked though, but I doubt that CAD files are #1 on the list for stuff to be hacked.

This does sound like a good idea, and I would love to be involved somehow.

Originally Posted by Greg Perkins

There are a few simple ways to make this work...

1) Create detail-less prints that can be downloaded as PWF's basically exploding the components (say a gearbox), the prints would show how the component works which would spark imagination for replication. I'm sorry but if Andy Mark gave detail-less prints and said "copy this exactly" it would be near impossible, which leads to difference and ingenuity. What they would have to do would be to take the prints, and create their own versions, using parts that would be easy to acquire.

2) Would be to create some sort of "read-only" file that you couldn't copy or modify. That would let the individual be able to virtually see it and take it apart. This way might lead to the system being hacked though, but I doubt that CAD files are #1 on the list for stuff to be hacked.

This does sound like a good idea, and I would love to be involved somehow.

I'm digging the first idea. Here's what I'm thinking: We can use the Autodesk DWF viewer to allow users to check out the assemblies, and then have a service for them to request a CAD file transfer if they're serious about building it. That way we could pick someone to host the bulk of the files on a home computer, keeping the website size down.

[AdamHeard]

Originally Posted by 114ManualLabor

I'm digging the first idea. Here's what I'm thinking: We can use the Autodesk DWF viewer to allow users to check out the assemblies, and then have a service for them to request a CAD file transfer if they're serious about building it. That way we could pick someone to host the bulk of the files on a home computer, keeping the website size down.

I'll send you some .dwfs of an updated version of our '07 base (module and assembled) which uses AM shifters (6 & 15 fps iirc) and AM sprockets. I'll also give you the contact info for if someone wants the source.