This new printing capability enables complex hydraulically actuated robots and robotic components to be automatically built, with no assembly required.
The multi-material 3D printing process, called Printable Hydraulics is capable of fabricating complete, functional, hydraulically actuated robotic structures in a single step, and can simultaneously print solid and liquid materials.
Robots produced in this way employ hydraulic channels to transmit force throughout the structure and incorporate large numbers of interconnected parts yet require no manual assembly which dramatically simplifies fabrication.
• No additional assembly is required because the forcetransmitting fluid is deposited at the same time as the robot’s solid body. This feature allows complex actuated structures that would be inconvenient or impossible to assemble manually.
• Printed hydraulics enables complex, intricate geometries that are infeasible with other 3D printing methods. For example, removing support material from tortuous capillary-like structures is often impossible. This is the case even with wax support when the aspect ratio of the channels is high, when it is not possible to include purging ports in the design, or when sealing these purging ports would impose onerous labor or design constraints.
• The use of an incompressible working fluid simplifies the control of complex fluid-actuated assemblies, relative to systems based on pneumatics.
• There is no need to purge air bubbles because the solid and fluid regions are fabricated together.
• Non-curing liquids are useful as an easily-removed support structure for subsequent layers; this approach is widely used in the examples we show.
• Compared to previous work employing kinematic linkages or gears in active 3D printed assemblies, printed hydraulics offers low-friction, low-backlash, high forcetransmission elements.
Designing solid and liquid printed geometries follows many of the same steps as a conventional CAD/3D-printing work-flow. The liquid parts, like the solid parts, must be specified via an interchange file (STL is a commonly used format) and a model material in the printer is assigned to that file. In the case of the printed liquid, the spoofed material type (TangoBlack+) should be assigned to the file specifying the liquid geometry.
Note that all references to direction are with respect to the printer’s coordinate system, rather than the coordinate system of the part. The Objet260 datasheet specifies an X/Y accuracy in the range of 20-85µm, and a Z accuracy of 30µm when printing with multiple materials. However, we observed that the resolution at liquid-solid interfaces when printing liquids is substantially coarser. We characterized the achievable print resolution when printing with liquids by creating various test geometries and printing many iterations of these geometries with different orientations on the build tray.
These tests revealed the primary challenge when printing with liquids: non-curing materials are moved by the roller and swept onto adjacent curing regions. The presence of the non-curing material inhibits the bonding between droplets of solidifying material within the current layer, and between subsequent layers. This effect is most pronounced at solid/liquid boundaries perpendicular to the print-head’s direction of travel (interfaces parallel to the Y axis), and is exacerbated by long unbroken segments of liquid.
The printer used in this work, a Stratasys Objet260 Connex, uses an inkjet head to deposit three different photopolymers simultaneously and achieves finished-part resolutions better than 100µm. The Objet260 uses eight print-heads with linear arrays of nozzles to deposit resins onto the build surface. These resins rapidly cure when exposed to the high-intensity UV light source mounted on the print head.
Three-dimensional models are broken up into thin slices, and printed from the bottom-up, layer by layer. The printer uses four heads for removable support (Objet printers use a soft, UV-cured solid for support) and allocates the remaining four heads to one or two model materials. Resins for the printer are supplied in plastic cartridges and these cartridges are labeled with an RFID chip, used by the printer to identify the material.
We define printed hydraulic parts as functional, fluidicallyactuated assemblies that employ non-solidifying liquids as an active, permanent, force-transmitting component. These parts, including the liquids, are printed in a single step, requiring no assembly. Such a printer can simultaneously fabricate solid and liquid regions within a structure. A print-head deposits individual droplets of material in a layer-by-layer build process. Each successive layer is deposited on the previous, and supports subsequent layers. Individual layers contain one or more material types, depending on the part geometry. Small droplet sizes (a 20-30 µm diameter is typical), enable finely spaced patterns of the constituent materials to be deposited. The use of solids with varying stiffness allows certain portions to be more flexible, enabling prescribed strain in response to applied fluid pressure. Supporting layers are provided either via removable curing support material or by non-curing liquid.
While we focus primarily on inkjet deposition in this paper, the printable hydraulics approach can also be applied to other 3D printing methods. For example, stereolithography uses focused light to selectively solidify photopolymers in a layer-by-layer process. Rather than allowing the uncured material to drain out of the model, certain regions of liquid could be permanently enclosed. Similarly, 3D printers based on fused deposition modeling (FDM) are now capable of depositing a variety of materials, including liquids, through interchangeable tool-heads. A dedicated nozzle with liquid could allow these multi-material FDM printers to create and then fill enclosed volumes with working fluid.
Creating robots has historically been a time consuming process. Constrained by available fabrication techniques, conventional robotic design practice dictates that engineers sequentially assemble robots from many discrete parts, with long concomitant assembly times. Massproduction achieves efficiency gains through optimizing each assembly step, but optimization requires that the design be fixed; even small changes become difficult and costly. Additionally, because many robots are unique or applicationspecific, relatively few opportunities to automate their assembly exist.
This situation is worsened by inevitable designfabricate-test-redesign iterations. Multi-material additive manufacturing techniques offer a compelling alternative fabrication approach, allowing materials with diverse mechanical properties to be placed at arbitrary locations within a structure, and enabling complex multi-part design iterations to be rapidly fabricated with trivial effort.
This work presents a new multi-material 3D printing process, Printable Hydraulics, capable of fabricating complete, functional, hydraulically actuated robotic structures in a single step. The key contribution of this work is a process that can simultaneously print solid and liquid materials. Robots produced in this manner employ hydraulic channels to transmit force throughout the structure and incorporate large numbers of interconnected parts yet require no manual assembly, dramatically simplifying fabrication.