A Eulerian Circuit-Based Reinforcement Learning Path Optimization Method for 3D Concrete Printing (2025-10)¶

The printing path is a critical factor influencing both the forming quality and printing efficiency of 3D concrete printing. However, existing path planning methods are mainly based on empirical approaches, which perform poorly in terms of dynamic adaptability. This paper introduces a novel path planning method for 3D concrete printing that integrates Eulerian circuits with reinforcement learning. The printing task is formulated as a traversal problem, where a Eulerian circuit is utilized to achieve a “one-stroke” path, eliminating the non-print ratio and improving efficiency. Building on this, the Q-learning algorithm is employed to further optimize the path by reducing the turning times and lifting times of the printhead through a carefully designed reward function. Experimental results demonstrate that the proposed method not only minimizes redundant travel but also achieves significant improvements in path optimization. Notably, its advantages become increasingly evident as the complexity of the printed geometries grows, outperforming traditional approaches in intricate scenarios. The method also generates highly regular paths, paving the way for improved digitization and standardization in 3D concrete printing.