Charlie C. L. Wang

This paper presents a field-based optimization method for process planning in hybrid manufacturing, where the goal is to determine a feasible and efficient sequence of additive and subtractive operations for fabricating a target shape. Existing approaches rely on deterministic or discretized formulations, which lead to unoptimized fabrication time and make planning sensitive to voxel resolution. They also do not explicitly account for distortion in intermediate structures formed during manufacturing. To address these limitations, we represent manufacturing states using continuous fields, which scale naturally to models with large dimensions and fine geometric features while enabling smoother fabrication sequences. Based on this representation, we formulate process planning as a numerical optimization problem under multiple objectives, including final shape conformance, intermediate structural strength, manufacturing time, self-supporting, and collision avoidance. Experimental results show that our method can generate distortion-aware process plans on a variety of models while substantially reducing fabrication time through up to a 72.6% reduction in the volume of extra support.

Functionally graded metamaterials offer a powerful route to spatially tailored properties and multifunctional performance. Their inverse design, however, remains challenging because spatially varying cellular topologies must remain geometrically compatible across neighboring regions, while multiphysics objectives often require time-consuming numerical evaluation. In this work, we introduce a compatibility-aware morphology learning framework for the inverse design of functionally graded metamaterials under expensive multiphysics evaluation. A low-dimensional morphology manifold is learned from cellular structure priors with explicit encoding of boundary compatibility, thereby embedding compatibility directly into the learned representation. This enables continuous geometric transitions among spatially varying unit-cell topologies. Coupled with latent-space optimization and prior-based dataset filtering, the proposed framework substantially reduces design complexity and makes inverse design tractable when target multiphysics are computationally expensive. Mechanical-acoustic inverse design is used as a representative example, in which acoustic response serves as the expensive physics to be simulated while mechanical performance is optimized concurrently. The resultant graded metamaterials are fabricated and experimentally validated through mechanical and acoustic absorption tests. Compared with specimens constructed from a single optimized cellular structure, the inversely designed graded metamaterials exhibit improvements of 58.1% in stiffness and 78.5% in failure loads under bending while maintaining high acoustic absorption.

To enable the design and manufacturing of optimized composite structures using fabric plies, we propose a field-driven optimization framework that jointly optimizes structural topology and manufacturable layers. A central challenge in this setting is the modeling and optimization of partial fabric layers with near-uniform thickness, which we formulate as a semi-continuous periodic scalar field parameterized by a continuous implicit neural vector field. Within this concurrent structure-layer optimization framework, we further derive a formulation of inter-layer anisotropic mechanical behavior that enables effective modeling of mechanical property transitions induced by partial-layer boundaries, together with additional objectives for manufacturability and field regularization. We validate the effectiveness of our approach through both numerical simulations and physical experiments, demonstrating that the optimized fabric-reinforced laminated composites achieve up to 43.8% higher stiffness compared to counterparts fabricated using planar fabric plies.

Infrared-assisted 3D compaction printing of continuous carbon fiber-reinforced thermoplastics on curved surfaces poses significant challenges for temperature regulation, because the distributed and transient thermal footprint of the infrared heat source interacts strongly with local surface geometry and feedrate. This work achieves consistent temperature control at the contact region through feedrate optimization enabled by a differentiable transient thermal-field modeling framework formulated for curved surfaces. A simulation-driven approach is used to identify path-dependent coefficient functions in the thermal model, and an efficient backtracking algorithm is developed to determine optimized printing feedrates. Together, these contributions enable quality-controlled desktop-scale continuous-carbon-fiber 3D compaction printing using low-power infrared heating. The proposed method is validated through physical experiments on a robotic system equipped with prototype desktop hardware.

Continuous carbon fiber placement on curved surfaces using low-power infrared heating poses significant challenges in temperature regulation, as the inherently distributed and transient thermal footprint of the heat source interacts strongly with local surface geometry and feedrate. In this work, we achieve consistent temperature control at the contact point during point-focused infrared heating through feedrate optimization enabled by a transient thermal field modeling framework formulated for curved surfaces. A simulation-driven approach is employed to identify path-dependent coefficient functions in the thermal model, and an efficient backtracking algorithm is developed to provide robust initial guesses for the feedrate optimization. Together, these contributions enable quality-controlled desktop continuous carbon fiber placement using low-power infrared heating. The effectiveness of the proposed method is validated through physical experiments conducted on a robotic system equipped with a prototype desktop carbon fiber placement hardware.

Flexible sensors are increasingly employed in soft robotics and wearable devices to provide proprioception of freeform deformations. Although supervised learning can train shape predictors from sensor signals, prediction accuracy strongly depends on sensor layout, which is typically determined heuristically or through trial-and-error. This work introduces a model-free, data-driven computational pipeline that jointly optimizes the number, length, and placement of flexible length-measurement sensors together with the parameters of a shape prediction network for large freeform deformations. Unlike model-based approaches, the proposed method relies solely on datasets of deformed shapes, without requiring physical simulation models, and is therefore broadly applicable to diverse robotic sensing tasks. The pipeline incorporates differentiable loss functions that account for both prediction accuracy and manufacturability constraints. By co-optimizing sensor layouts and network parameters, the method significantly improves deformation prediction accuracy over unoptimized layouts while ensuring practical feasibility. The effectiveness and generality of the approach are validated through numerical and physical experiments on multiple soft robotic and wearable systems.