<p>To address the issues of low path planning efficiency and weak obstacle avoidance capabilities in open-pit mine blast hole filling manipulators, this paper proposes an improved algorithm to enhance the manipulator's performance. First, for inverse kinematics solving, the Dual learning strategy Particle Swarm Optimization (DlsPSO) algorithm is designed. It incorporates methods such as optimizing particle update mechanisms, applying perturbation functions for particle superposition perturbations, and adopting adaptive control parameter update schemes, thereby improving solution accuracy and stability. MATLAB simulations show that DlsPSO exhibits smaller errors and better stability in both single and random end-effector pose solutions. Second, in path planning, the improved RRT* (Rapidly-exploring Random Trees Star) algorithm introduces an elliptical sampling function to restrict the sampling area and reduce invalid nodes. It combines dynamic step size and obstacle-based variable step size mechanisms to lower collision probability, along with a path pruning strategy to eliminate redundant nodes. Simulation results indicate that compared with traditional RRT and RRT* algorithms, the improved RRT* achieves fewer average nodes, fewer collision counts, and shorter path lengths in both simple and complex obstacle environments. Finally, simulation verification using the blast hole filling manipulator demonstrates that the improved RRT* significantly reduces path length and node count, markedly lowers collision frequency, effectively avoids obstacles, and meets the high-precision motion requirements of the manipulator's end effector.</p>