The proposed research seeks to merge data-driven methods, symbolic knowledge, and domain expertise to enhance the planning, optimization, and control of industrial robots and processes in robot-based manufacturing. By incorporating sensor data, we can leverage additional information to monitor and control industrial processes, leading to improved processing time and quality. Data-driven methods offer the ability to model and control complex systems without a complete understanding of their mathematical representation. However, they may lack interpretability and domain-specific knowledge.In contrast, symbolic representations, such as logic relations, first principles models, and physics-based knowledge, provide a structured and interpretable framework for capturing the underlying processes. By combining data-driven models with symbolic representations, we can harness the strengths of both approaches. Data-driven methods can uncover paIerns and relationships from various process scenarios, augmenting the symbolic knowledge and providing insights that may not be explicitly captured by symbolic representations alone. Symbolic approaches, on the other hand, provide a structured framework to incorporate domain expertise and interpretability into the optimization and control routines.