Nonlinear Path Optimization Algorithm for Mining Trucks Based on Two-Layer Trust Region Strategy

This paper proposes a nonlinear path optimization algorithm for mining trucks based on double-layer trust region strategy, aiming at addressing path planning in the complex environment of mining areas. Initially, to ensure kinematic constraints and obstacle avoidance requirements in the outer loop of the optimization process, an "eight-neighborhood spiral" drivable obstacle avoidance tunnel method was introduced on the basis of the classic drivable tunnel technology. The method expands the search to eight neighborhood directions through the dynamic adjustments of search modes, to facilitate adjustments to collision path points for obstacle avoidance in the tunnel. Then, a state-space discrete model was constructed for path optimization. The adoption of a backtracking linear search method in the inner loop allows for fine-tuning the inner-loop search direction and step size. Dynamic adjustments to the trust region radius ensure effective response to nonlinear characteristics in the path optimization model. Finally, comparative experiments were conducted using real trucks, yielding the following results. Compared with major algorithms popularized in academic studies, such as LIOM, DL-IAPS, and OBCA, the average solution time of this algorithm was at least 2.75% shorter in the shovel loading operation scenario. Compared with the discrete point smoothing and multiple optimization algorithms widely applied in the industry, the algorithm exhibited an improvement in the limiting effect of curvatures and curvature change rates by about 16.36% and 28.07%, and 8.46% and 19.61%, respectively. In obstacle avoiding experiments, the application of the proposed algorithm increased the distance from obstacles to a certain extent, while maintaining a low curvature. These optimization results not only demonstrate an improvement in the planning efficiency of driverless mining trucks, but also a significant enhancement in their operational safety in complex environments.