My good friend Erik Schluntz and I came up with this really cool idea: We should build room-sized 3D printers using winch robots (also called cable robots or rope robots) instead of using big 3-axis gantry systems. The idea has a lot of merit: eliminate bulkiness, vastly-lower robot cost, insanely-big workspaces, and super-easy installation and calibration. We thought a bit about what it could look like as a business (eg. launching via Kickstarter), but opted to pass for the time being. In the interim there are some supremely-entertaining possibilities to scale it up to the size of a football stadium... If anyone can get us access to a stadium, we'd be game for the mother of all weekend hackathons! Anyway, we figured we'd expand a bit on the idea and examine some earlier, related systems.
The basic idea: Attach a few winches (motors with cables) into a room's periphery and run cables to a central "stage" that contains a fusion deposition molding (FDM) print head. The robot could auto-calibrate its kinematics and perform precise pose estimation using depth cameras (eg. Kinects) on the stage in conjunction with "odometry" from the motor controllers at the winches. Fine movements could be achieved with feedback control over pose rather than relying exclusively on odometry alone at the winches. Using my masterful GIMP skills:
I can imagine three potential challenges / limitations:
(1) How do you get material to the print head when working with such a large workspace and with such large structures? I think that could be trivially solved by having the robot pick up pre-positioned material canisters from elsewhere in its workspace.
(2) How do you prevent previously-printed structures from interfering with the cables? As long as the stage has sufficient mass such that it can be controlled exclusively by overhead cables, I think you can follow the class FDM approach: Build up a structure from overhead via successively taller layers.
(3) How do you print fine features? There are plenty of large-scale FDM printers, but they tend to have poor resolution. To address this, you could attach a high-precision arm or 3-axis system onto the stage for fine-feature printing (eg. any of the ones discussed here or here). Alternatively, I think it's probably possible to do precision control (ie. servo feedback) by doing pose estimation with a depth camera (eg. Kinect) while using a smaller FDM print head -- either as the main print head, or have a secondary fine-resolution one on the same stage. Using the depth camera for precise pose also has the secondary benefit of eliminating a lot of systemic errors caused by using odometry alone for control.
As we talked through the idea, it hit me:
Why stop there? We could literally build stadium-scale 3D printers. We could even reuse the existing SkyCam winch robots that are used for overhead filming during football games!
Erik and I floated the idea of turning the room-sized 3D printer into a quasi-startup: launching a Kickstarter to gauge interest and show traction. We even thought, "Hey, we could build a real stadium-sized 3D printer for our promotional video!" But in the end, while it would've been a whole lot of fun, neither Erik nor I were keen on the idea's long-term economics. Nor were we super passionate about it.
Naturally, the stadium-sized version would probably be the most fun aspect of the whole project. We'd still be game for building a stadium-sized 3D printer just for the fun of it -- oh, the bragging rights! If you think you can help get us access to one...
Only after flushing out our own thinking, we ran some internet searches and discovered a number of slightly related old projects.
Fraunhofer started exploring winch robots back in 2007. In 2011 they issued a press release about the IPAnema robot, whose goal was to cheaply (yet accurately) install solar concentrator systems for large thermal solar plants:
The idea of combining the power of a crane with the speed and accuracy of a robot is something other people have tried to do, but computer processing power was simply not advanced enough for cable robots to succeed until now. Our IPAnema robot actually consists almost entirely of cables and winches.
Unlike cranes that, owing to their swinging loads, must move slowly, IPAnema can accelerate quickly in full control of its load thanks to its automatically controlled winches.
Here are some photos, concept drawings, and videos of their early prototypes:
Like Fraunhofer, I think this "coordinated crane" approach could be quite powerful for robot construction too.
We also came across DIY winch robot by Ben Peters from MIT. He describes his Spiderbot project thusly:
The SpiderBot is a cable-suspended robotic gantry system that provides an easily deployable platform from which to print large structures. The body composed of a deposition nozzle, a reservoir of material, and parallel winching electric motors. Cables from the robot are connected to stable points high in the environment, such as large trees or buildings. This actuation arrangement is capable of moving large distances without the need for more conventional linear guides, much like a spider does. The system is easy to set up for mobile projects, and will afford sufficient printing resolution and build volume. Expanding foam can be deposited to create a building-scale printed object rapidly. Another material type of interest is the extrusion or spinning of tension elements, like rope or cable. With tension elements, unique structures such as bridges or webs can be wrapped, woven, or strung around environmental features or previously printed materials.
In fact, Ben even conceived of the stadium-scale robot. I guess that just goes to show: Good ideas are almost never new. A few things to note: (1) It looks like Ben never actually finished his robot. Right now, he only has some basic positioning capabilities, but no printing. (2) It looks like the robot currently relies entirely on winch odometry and system geometry for positioning accuracy. This seems less than ideal. Our suggestion of using depth sensing could dramatically improve this aspect. (3) Ben's decision to put the winches on the stage is (in my opinion) probably backwards. Putting them on the stage has the benefit of local motor control (versus wireless). But it also drastically increases the weight of the stage, and thereby increases the size, weight, and capacity of the winches. That's a brutal positive feedback loop in the design, though I could imagine situations (eg. very light FDM components) where having the winches locally is better for simplicity and cost.
In summary, I think winch robots could be a compelling alternative to the other giant 3D printing and rapid prototyping robots that currently exist. At the very least, it's probably worth further exploration. I just wish I had more time.