复杂构件激光切割轨迹规划算法 (特邀)

Objective In the laser cutting of complex components, achieving synchronous laser head pose control and avoiding collisions remain significant challenges, primarily due to frequent and abrupt attitude adjustments during the process. To address these issues, this paper proposes a segmented trajectory planning algorithm tailored for laser cutting of complex curved surfaces. By integrating 3D point cloud processing and an improved multi-chromosome genetic algorithm, the method aims to realize coordinated optimization of motion path and laser head orientation, thereby enhancing machining efficiency, process reliability, and operational safety. Methods Taking the laser cutting of large and complex automotive door inner panels as a case study, the workpiece surface data is first represented as a 3D point cloud model. After downsampling to reduce data density, the point cloud is segmented into distinct regions based on surface normal vectors using Principal Component Analysis (PCA). A multi-threshold boundary point extraction method combined with clustering algorithms is employed to accurately identify and extract the boundary points of the door inner panel based on their geometric features. Subsequently, a contour connection algorithm is used to cluster these boundary points into closed cutting contours, which are further classified according to the regions their points belong to —contours in the same category share similar geometric characteristics, enabling concentrated machining that minimizes drastic laser head attitude changes and collision risks caused by frequent cross-region switching. An enhanced multi chromosome genetic algorithm is designed and improved, where three chromosomes correspond to three types of decision variables respectively, realizing classified and segmented cutting of the laser head and autonomous selection of entry and exit points. Additionally, a laser head return-to-origin mechanism is introduced to ensure safe switching and smooth transition of cutting paths between different machining regions. The proposed algorithm formulates the trajectory planning task as a hierarchically constrained Generalized Traveling Salesman Problem (GTSP) for efficient optimization. Results and Discussions The proposed algorithm is compared with four conventional optimization methods —standard Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), and Greedy Algorithm —through comprehensive simulations (Fig.8). Comparative evaluation results demonstrate that the proposed method achieves superior performance in key process indicators: it improves laser head pose smoothness by 18.2% and obstacle avoidance safety by 19.1% compared to the next-best performing algorithms. This effectively suppresses drastic changes in the normal vector of the laser head attitude between adjacent machining loops, providing sufficient buffer space for attitude transitions and reducing overcutting, collision risks, and equipment wear caused by abrupt acceleration/deceleration of machine axes. Although slight trade-offs exist in non-productive travel distance and convergence speed due to the need for refined processing of attitude and safety constraints, the algorithm maintains competitive computational efficiency and global optimization capability, avoiding the local optimality limitation of the Greedy Algorithm. It achieves an effective balance between machining stability, safety, and optimization efficiency, with overall performance more aligned with the core requirements of practical production for stability and safety, thus offering a more engineering practical solution for laser cutting trajectory planning. Conclusions This study presents an effective solution for trajectory planning in complex surface laser cutting by integrating point cloud-based segmentation with multi-objective evolutionary optimization. The proposed segmented trajectory planning scheme successfully realizes the collaborative optimization of path length, pose continuity, and operational safety, providing a robust and engineering-practical approach for high-precision laser manufacturing of complex components. Future research will focus on integrating trajectory smoothing techniques such as B-spline and Bezier curve fitting for post-processing of planned trajectories, optimizing transition paths between adjacent contours to reduce micro-vibrations during laser head movement and further improve machining stability and surface quality. Additionally, the algorithm will be closely integrated with laser cutting processes, and laser processing parameters will be dynamically adjusted in real-time based on trajectory characteristics and changes on the basis of "segmented planning," truly realizing closed-loop optimization from "trajectory generation" to "process adaptation" and promoting the upgrading of laser cutting towards intelligence and adaptability.