Resource Archive

The demand for compact beam-steering solutions has driven interest in gradient-index (GRIN) lenses, which shape wavefronts via spatially varying dielectric properties. Additive manufacturing enables their fabrication, but a central challenge remains: solving the inverse problem of determining the volumetric permittivity distribution for a desired field transformation. Unlike voxel-by-voxel topology optimization requiring thousands of full-wave simulations, our approach targets smooth GRIN profiles for propagation-based devices, reducing optimization to minutes on a standard computer. Here, we present an inverse-design framework that optimizes the geometrical optics propagation equations directly. We experimentally demonstrate the method by realizing a multi-feed GRIN lens based on a three-dimensional gyroid architecture fabricated via additive manufacturing. The device enables continuous beam scanning over a ±35∘ range with 33% fractional bandwidth, low scan loss, and excellent beam fidelity. Our results establish a direct pathway from electromagnetic functionality to manufacturable metamaterial devices, enabling scalable wavefront engineering.

We present the design, fabrication, and characterization of electrically large Luneburg lens (LL)-based antennas realized through vision-controlled inkjet-based additive manufacturing. Utilizing a low-loss cyclic olefin thermoset resin compatible with high-resolution material jetting, we demonstrate the fabrication of wideband gradient-index (GRIN) lenses operating at frequencies up to 100 GHz with apertures exceeding 30 wavelengths and realized gains exceeding 34 dBi.

Introduces Vision-Controlled Jetting (VCJ), an inkjet deposition process for creating complex composite systems and robots. A scanning system captures the 3D print geometry and enables a digital feedback loop that eliminates the need for mechanical planarizers. This contactless process allows the use of continuously curing chemistries and a broader range of materials, including tough epoxy and soft thiol-ene, leading to multi-material parts with high resolution and superior properties like increased solvent and UV resistance.

Introduces the Scalable Spectral Packing (SSP) algorithm to solve the NP-hard problem of densely packing 3D objects into a container for manufacturing (like 3D printing). The method uses a discrete voxel representation and formulates collisions as correlations of functions computed efficiently using the Fast Fourier Transform (FFT). Crucially, the method ensures the packing is "interlocking-free," meaning all objects can be removed without deformation or breakage. It achieves state-of-the-art performance in both density and speed.

Introduces Neural Jacobian Fields, an architecture using deep neural networks to learn to model and control diverse robots from vision alone. The method maps a video stream to a visuomotor Jacobian Field the sensitivity of all 3D points to the robot's actuators. It makes no assumptions about the robot's materials, actuation, or sensing, requiring only a single camera and training by observing random commands. This enables accurate closed-loop control for systems that are traditionally hard to model, such as multi-material and soft robots, potentially lowering the barrier to robotic automation.

Reveals a universal scaling law that links strand mechanics and connectivity to predict the intrinsic fracture energy of diverse stretchable networks. This law establishes a physical basis for fracture resistance, showing that Gamma is independent of the energy to rupture a strand but depends on the strand rupture force, breaking length, and connectivity. The law provides a framework for fabricating and designing tougher materials across various topologies, dimensionalities, and length scales, from nanoscale polymers to macroscale architected materials.

Presents a fully 3D-printed, multi-material robotic hand that is pneumatically actuated (controlled by air pressure). This demonstrates the capability of advanced fabrication techniques to create integrated, soft robotic manipulators with many degrees of freedom. The authors have worked on bio-inspired hands that combine soft and rigid elements to achieve high dexterity and sensitivity.