Key points are not available for this paper at this time.
The welding process has significant importance across several sectors, as it guarantees integrated components' structural soundness and dependability. Robotic welding, distinguished by the use of automated procedures facilitated by robotic arms, has garnered considerable recognition due to its exceptional accuracy and productivity. The conventional techniques need help attaining the most favourable welding results due to workpiece location and trajectory accuracy. The need arises for the advancement of sophisticated optimization methodologies. This paper presents a smart manufacturing paradigm to tackle the problem using Reinforcement Learning for Robotic Welding Process Optimization (RL-RWPO). The RL-RWPO framework combines Vision-Based and CAD-Based methodologies, facilitating a holistic comprehension of the workpiece and its intended arrangement. Using 3-D calibration and workpiece placement techniques significantly improves the precision of welding processes. Reinforcement Learning is used in trajectory optimization to ensure accurate and efficient welding operations. The results of the simulation indicate significant enhancements, as seen by the average Weld Seam Quality of 93.27%, Welding Speed of 50.51 mm/s, Weld Penetration Depth of 5.25 mm, Trajectory Accuracy of 3.04 mm, and Welding Efficiency of 91.52%. The RL-RWPO method can improve the effectiveness and excellence of robotic welding practices in several industrial sectors.
Sathick et al. (Sat,) studied this question.