SMARTT Toolpaths: Scalable Multi-head Assignment and Routing using Thermal- and Time-optimized Toolpaths

The SMARTT Toolpaths project aims to revolutionize multi-head Laser Wire Directed Energy Deposition (LW-DED) by enabling scalable, collision-free, and thermally optimized toolpath planning that dramatically increases throughput while improving part quality. Traditional single-head planning strategies or naive multi-head scheduling approaches fail to leverage the full potential of coordinated additive manufacturing due to high idle time, thermal distortion, and collision risks. SMARTT Toolpaths—short for Scalable Multi-head Assignment and Routing using Thermal- and Time-optimized Toolpaths—addresses these challenges through a novel, layered framework consisting of thermal optimization, productivity optimization, and machine learning.Layer 1, the Thermal Optimization Layer, models the part as a graph and sequences deposition paths to maximize thermal uniformity while preventing collision. This layer builds upon the SmartScan framework developed by the University of Michigan, which has demonstrated up to 92% improvements in thermal uniformity in LPBF systems. Layer 2, the Productivity Optimization Layer, assigns and schedules the thermally admissible toolpaths to each deposition head, minimizing non-productive times such as cooldowns, air travel, and workload imbalance. Layer 3, the Learning Layer, introduces a graph neural network (GNN) trained on simulation outputs to rapidly predict high-quality toolpaths for new geometries, enabling real-time and adaptive planning in Phase II.In Phase I, the team will demonstrate SMARTT Toolpaths on a complex honeycomb-shaped benchmark part—previously fabricated by ADDiTEC using single-head LW-DED—which experienced significant distortion. The honeycomb geometry is highly relevant to defense applications such as missile components, aircraft skins, and naval bulkheads due to its high stiffness-to-weight ratio, geometric complexity, and susceptibility to thermal buildup. Using two coordinated deposition heads, SMARTT Toolpaths will reduce build time by over 2× while eliminating thermal distortion, all without design modifications. The project will include virtual validation using ADDiTEC’s simulation platforms and, as a stretch goal, physical printing on ADDiTEC’s dual-head robotic system with real-time thermal monitoring.The SMARTT Toolpaths framework directly addresses DARPA’s CHAMP topic HR0011SB20254-06 by delivering cooperative, heuristic-driven scheduling with real-time simulation feedback, targeting >2× productivity, zero collisions, and thermal control. Phase II will scale the system to N-head configurations, generalize to arbitrary geometries, and integrate real-time learning, paving the way for deployment across defense and industrial sectors where rapid, large-format metal AM is needed.