Vention has spent the past year integrating a slew of new features into its web-based CAD, MachineBuilder. Now with a comprehensive set of design collaboration tools, a machine learning algorithm that suggests related parts, and extensive additions to our library of modular parts, MachineBuilder covers a growing array of use cases in tooling, conveyors, fixtures, jigs, robot cells, and automated equipment applications.
As we celebrate our second major platform release, I spoke to Max Windisch—Vention’s CTO and resident philosopher—about Vention’s progress this past year. Keep reading to find out what he had to say.
How has Vention’s growth lined up with the changes to MachineBuilder?
We’ve been very pragmatic throughout Vention’s growth, and I think that the solution that we propose with Vention is proportional to the problem that we’re trying to solve (which isn’t always the case for software companies). Everything is driven by the problem we’re trying to solve, at the stage we’re trying to solve it, and there’s always a practical measure of what it is we’re trying to build.
What can you tell me about the features you’ve added to Vention’s MachineBuilder?
This past year, we focused on ease of use. For instance, we made it simpler to resize things in the editor. Even though the purpose of our platform is to make custom machines, the truth is that some of these machines follow a reusable pattern. Users still need to be able to modify the pattern, though. Now they have the option to insert new parts, resize them, or remove some elements—all things that support the usability and modularity of custom machine design, and that make it possible for someone without much experience with CAD software to build something.
In keeping with the spirit of Vention, we added custom panels to MachineBuilder earlier this past year through our partnership with Groupe PolyAlto. The ability to design and order cut-to-size plastic panels is a useful feature on its own, but as with every new component that we integrate into the Vention library, we’ve also made sure it’s simple to use. In this case, panel installation in the builder is partially automated by surrounding geometries. After you select your panels, a contour is created automatically—no need to delineate where the panels should go point-by-point, and no specialized tools or knowledge is required.
Since Vention’s beginning, we’ve always had an idea of implicit groups in mind. Our goal is to make building in the tool as close as possible to building in real life. If you had a half-assembled machine on your shop floor, you could grab it with your hands and manipulate it without further thought. It’s the same in the virtual world: every sub-assembly is an implicit group of assemblies.
That said, we also want to give our users the ability to control explicit groups, for a number of reasons and this is what made our grouping feature such an important addition to MachineBuilder. Vention users can import CAD files, so they’re not limited to parts in our own catalog. This allows anyone to build truly custom machines since the imported part is often the thing you build around and groups facilitate a design organized around them.
Where does machine learning fit into the Vention model?
It’s not always obvious to a new user what a part—such as a specific mounting plate, for instance—is supposed to be used with. In addition to our existing part compatibility checker, we introduced a recommended part feature to help users navigate their designs more efficiently. To accomplish this, we’re using data collected from our extensive library of designs to assess the usage patterns behind these arrangements and rank suggestions based on their execution.
Traditional CAD software developers don’t have the potential to access the files that people produce in the tool, and that data is lost. Vention, on the other hand, is in a position to hold that data and augment the user experience through machine learning. There’s not that much complexity involved, but as we grow and support more complex machines, it will be a key resource for us to develop.
Have the implications of Vention’s technology for machine design become clearer to you since you were first introduced to the idea?
Vention enables users to circumvent the traditional hardware-design learning curve by giving them an accessible solution. The platform we present is closer to what people are familiar with, in that it allows them to build a machine using parts they’re already aware of—like motors and actuators—in a dramatically shortened time frame.
I was eager to prove the viability of the Vention model from the start. I knew it would be a challenge to offer a simplified tool to people who were already somewhat familiar with mechanical design and machine building, but in observing the use cases that we’ve since supported, I can now say with conviction that our solution is relevant to a great deal of applications.
Software is about abstractions. Were you looking at CAD software in this same light when you approached Vention’s MachineBuilder?
CAD software is an extremely thorough type of solution and it has years of thought processes built into it. It’s very mathematical and geometrical, and you are expected to build everything from the ground up. Because it’s so comprehensive, I didn’t even want to think about building our own version of the software; and in any case, it wasn’t the right type of solution for us. CAD software has a burdensome learning requirement, and our hypothesis was that people would come and go to Vention’s platform; they wouldn’t spend time learning to use it. It didn’t make sense to put out a software solution with the same amount of depth.
By simplifying the CAD system and raising the level of abstraction, we remove the uncertain outcome of the machine design process for those with less experience in software or custom manufacturing. Vention doesn’t pretend to support all use cases, but our mission is to grow our catalog and support as much as we can.
Have any of these use cases surprised you?
Vention essentially has one main actuator—the linear motion actuator—and I’m continually surprised by just how many things can be built with it. I think this highlights what our model delivers: a few simple building blocks can make machines that are highly specialized.
For now, we’re still in the early stages of what we aim to support. I like to say that we spend a lot of our time just adding parts. That may sound easy, but whenever we add parts, we add complexities to the mechanisms that we can handle. Software-wise, that’s where the more difficult—and rewarding—work is.