CAREER: Integrated Digital Thread for Self-Evolving Cooperative Robotics Remanufacturing

Remanufacturing restores worn or damaged products to like-new performance, extending the life of high-value assets while reducing dependency on costly replacements and lowering supply chain vulnerability. Many repairs still depend on technician judgment that is difficult to document and is increasingly at risk as experienced workers retire faster than replacements can be trained. Although robots offer the potential to alleviate workforce shortages, today’s programmed automation is largely limited to repetitive operations and cannot replicate human-level reasoning and adaptability required to manage the unique geometries, uncertain damage states, and evolving conditions inherent to repair workflows. This Faculty Early Career Development (CAREER) project aims to create scientific and educational foundations for an integrated digital thread framework that enables autonomous, self-evolving cooperative robotic systems capable of additive repair. This project advances remanufacturing by moving from programmed automation toward cognitive automation, creating intelligent systems that leverage expert knowledge and continuously adapt to perform unique, customized operations across all remanufacturing steps. Further, this project will broaden participation through curriculum modules at the University of Connecticut, hands-on research and mentoring, summer programs with local schools and community colleges, and workforce development activities for manufacturers and small businesses. The overall research goal is to establish a mind-body-environment loop that integrates knowledge-based reasoning, physics-informed embodied interaction, and continuous environment-loop adaptation, to support adaptive repair actions and scalable deployment across emerging remanufacturing applications. Specific objectives include: (1) Develop a self-evolving, memory-augmented planning module to sense, diagnose, identify, and learn what processes are needed for the repair task, enabling generalizable context-aware reasoning. (2) Develop an embodied engine to decompose tasks, allocate subtasks to individual arms, optimize high degree of freedom motion plans, and execute non-planar slicing, ensure morphology-driven reconfiguration, and (3) Develop an adaptive digital twin for decision making based on multi-fidelity process data and physics-based simulation within a continuous environment loop, completing the mind-body-environment framework. Driven by the neuro-vector-symbolic architecture, this research integrates distributed sensory embeddings with structured symbolic knowledge, embodiment constraints and physics-based dynamics, and multi-fidelity simulation with experience-driven refinement. The resulting unified representation enables knowledge-driven reasoning, morphology-configured planning, and simulation-augmented adaptation. The system will be validated on multi-arm laboratory experiments and industrial case studies that include both reconstruction of damaged parts and modification to new specifications. This research advances foundational knowledge at the convergence of cognitive intelligence, embodied robotics, and advanced remanufacturing. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.