Session Sheet - 2026 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2026): Multi-Axis and Robotic Enabled Additive Manufacturing

2026 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2026): Multi-Axis and Robotic Enabled Additive Manufacturing
Program Organizers: David Leigh, University of Texas at Austin

Monday 1:30 PM
August 3, 2026
Room: Brazos
Location: AT&T Center

1:30 PM
Physical AI++: A Multi-Agent Spatiotemporal Control Framework for Adaptive Solid Freeform Fabrication: Bharat Dwivedi¹; Arun Rebbapragada²; Indira Dwivedi³; Arka Rebbapragada²; Jeyakesavan Veerasamy⁴; ¹Eastlake High School; ²Carrollton Farmers Branch School District; ³Georgia Tech; ⁴University of Texas, Dallas
    Physical AI enabled closed-loop manufacturing systems integrate sensor fusion and process-planning. However this approach is largely geometric-centric, single agent in nature limiting the ability to control material evolution spatially and transiently. We present a novel concept Physical AI++, that uses mulit-agent frameform for Solid Freeform Fabrication to enable spatiotemporal control of material state, including temperature, structural continuity, and process dynamics as first class variables. The framework will coordinate multiple robotic agents to perform complementary roles including additive, subtractive processes, energy modulation, structural constraints and microstructural evolution. This cooperative material engagement will use multiple agents to manipulate shared material system enabling fabrication beyond layer-based paradigms. While we describe entire framework, we will demonstrate it in limited capacity through a case study. We will demonstrate material weaving through constrained conduits, analogous to cable routing in infrastructure development. We will show how continuity is sustained across complex regions.

1:50 PM
Robotic Reactive Extrusion AM for Large Scale Complex Geometry: Corrie Van Sice¹; Devin Young¹; Yuliang Feng¹; Yunlan Zhang¹; ¹The University of Texas at Austin
    Coastal infrastructure can benefit from large-scale architected geometries—such as triply periodic minimal surface (TPMS) lattices—which reduce wave forces and may enhance erosion resistance. Additive manufacturing (AM) is the de facto method for producing TPMS structures due to their geometric complexity, and robotic platforms enable significant kinematic and workspace flexibility for fabrication at architectural scale. Reactive extrusion AM (REAM) offers high deposition rates but has been limited to material characterization with minimal application to large or complex geometries. This work presents a ROS2-based robotic REAM platform for fabricating decimeter-scale (50 cm3) TPMS structures using a Yaskawa industrial manipulator. Closed-chain affordance planning is applied for the first time to AM, enabling real-time trajectory generation with minimal joint motion for future feedback-enabled planning and control. A software scheduler coordinates extrusion and robot motion execution. Gyroid unit cells are successfully fabricated, demonstrating the first application of robotic REAM to functionally-driven infrastructure.

2:10 PM
Volumetric Additive Repair: Near-Real Time Selective Multi-Axis Toolpathing: Tadeusz Kosmal¹; Christopher Williams¹; ¹Virginia Tech DREAMS Lab
    Additive manufacturing (AM) enables spatially selective material deposition, making it well suited for repairing damaged components. However, existing AM repair workflows remain heavily manual, often requiring human-guided segmentation and localization of damage from scanned geometry, as well as expert oversight to generate component-specific repair toolpaths. This lack of automation limits responsiveness to new geometries and constrains the scalability of AM-based repair. This work introduces a novel methodology that leverages in-situ geometric measurements to automatically generate conformal multi-axis repair toolpaths. Given the target final geometry and localized 3D scan data collected by a robot, the method can identify the compromised regions of a damaged component. Then, with the kinematic flexibility of multi-axis robotic AM deposition, material is selectively targeted to fix the damaged component. The robotic repair is validated by its use in scanning, identifying damage, generating and executing toolpath for a robotic Material Extrusion (MEX) AM processing system.

2:30 PM
A Framework for Energy-Efficient Repurposing of Retired Components Using Hybrid Additive–Subtractive Manufacturing: George Duke¹; Niechen Chen¹; ¹Northern Illinois University
    Scrapping end-of-life components leads to material loss, while complete material recovery through reprocessing requires substantial energy. This paper proposes a framework for repurposing retired components into new functional geometries using hybrid additive and subtractive manufacturing, with the objective of minimizing material wastage and energy usage during transformation. Rigid-body alignment between the source component and target geometry is optimized using Covariance Matrix Adaptation Evolution Strategy over the full three-dimensional rotation space, while translation is determined analytically for each rotation via fast Fourier transform cross-correlation. The aligned geometries are then compared to identify shared material to be retained, missing volume to be deposited, and excess volume to be removed, which are evaluated through a weighted cost function. Feasible manufacturing operations are sequentially planned using Monte Carlo Tree Search under evolving tool-accessibility constraints with a five-term cost model for processing, support, accessibility, reconfiguration, and tool change.

2:50 PM
A Unified Framework for Adaptive Path Planning and In-Process Control in Multi-Axis Robotic Additive Manufacturing: Walter Glockner¹; Jakob Hamilton¹; ¹Iowa State University
    Multi-axis additive manufacturing using localized deposition methods such as FFF, DED, and WAAM has long relied on pre-processed build plans. The stochastic nature of these processes poses challenges for motion planning, particularly given thermal gradients, part-level distortion, and anomaly formation that invalidate the geometric assumptions underlying offline toolpaths. Closed-loop control of deposition parameters has reduced process variability, but the union of online robotic path planning with in-process parameter adjustment remains absent, and current workflows are operator-dependent and decoupled from motion planning. This work unifies path planning, collision-aware trajectory generation, data acquisition, and processing within a single application for multi-axis deposition cells. The framework is robot- and sensor-agnostic while enforcing a standard for sensor-actuator communication, providing a foundation for online motion control in which toolpaths re-plan mid-build in response to build-state changes. We present the software architecture, data fusion strategy, and communication modes enabling adaptive in situ remediation.

3:10 PM Break

3:30 PM
Measuring Natural Frequencies in a 6 DOF Industrial Robot Performing Additive Manufacturing Tasks Using Experimental Modal Analysis: Mahati Kuram¹; Erik Komendera¹; ¹Virginia Tech
    The natural frequencies of a 6 DOF industrial robot are critical in predicting its dynamic behavior during additive manufacturing tasks such as 3D printing. Understanding these dynamic properties is necessary to improve precision, accuracy, and overall print quality. This paper aims to measure the natural frequencies of a 6 DOF robot using Experimental Modal Analysis. The robot system will first be modeled and simulated in the MuJoCo physics engine using Python. An accelerometer mounted on the end effector will be used to measure the dynamic response of the physical robot, and the experimental data will be compared with simulation results to evaluate model accuracy. Previous research has primarily applied Experimental Modal Analysis to robotic milling, where high-impact forces contribute significantly to deformation and positioning errors. This work extends those methods to additive manufacturing applications to assess their effectiveness and support future improvements in vibration-aware motion planning and structural accuracy.

3:50 PM
Fast Fragility-Aware Robotic Grasping of Metal Binder Jet 3D Printed Parts: Alexander Martinez-Marchese¹; Artyom Boyarov²; Bhumi Patel¹; Deepshikha Deepshikha¹; Chen Qian¹; Chinedum Okwudire¹; ¹University of Michigan; ²Stanford University
     The de-powdering stage in binder jetting additive manufacturing remains a manual and expensive post-processing step, creating a significant bottleneck that limits broader industrial adoption. This work presents an initial approach to automating the selection of both grasp location and applied gripping force for robotic handling, using material-specific surface friction data and the material’s maximum principal stress.We model the statistical distributions of these properties and scale them using two reference points: the minimum normal force required to prevent slipping and the maximum principal stress observed when applying a 1 N load at candidate grasp points. With this normalization, we evaluate and rank potential grasps by minimizing the probability of either slippage or structural failure, enabling the selection of an optimal grip strategy. The method successfully predicts which part/grasp combinations lead to part failure during testing.