Optimizing Robotic Task Sequencing and Trajectory Planning (TSTP): A Multi-Objective Hybrid Optimization Approach Coupled with Reinforcement Learning (DRL)

The integrated task scheduling and path planning issue in remote laser welding (RLW) is examined in this research. This work addresses the task of integrated task scheduling and path planning in RLW, proposing a novel Multi-Objective Hybrid Optimization Approach coupled with Deep Reinforcement Learning (DRL). The proposed study introduces a hybrid framework for task sequencing based on Giant Armadillo Optimization (GAO) and Zebra Optimization Algorithm (ZOA), which is used to schedule robotic operations dynamically and efficiently. Unlike the conventional static methods, the proposed method achieves the goal of balancing conflict objectives of task completion time and energy consumption with collision-free operation. A major contribution is the adoption of Proximal Policy Optimization (PPO) for trajectory planning, whereby robots are driven autonomously to navigate through unknown dynamic environments that may contain unobserved obstacles. The findings show a very high improvement in the efficiency of task execution and result in 0.028% FNR and 0.024% FPR, respectively. This shows the capability of the system to be adaptive and robust under real-world scenarios for robotic applications. These results highlight the efficiency and robustness of our method for real-world applications.