6 November 2023 in IEEE Transactions on Vehicular Technology
Trajectory planning for autonomous driving is challenging in unstructured scenarios such as mining sites. Existing studies mainly resort to a heuristic search-based planner to find a feasible trajectory. In the case of the narrow area, the heuristic function of the planner suffers from over-expansion problems, which may result in heavy computation burden and memory usage as well as potential failure. Therefore, a novel trajectory planning approach, namely learning and optimization-based trajectory planning (LOTP), is proposed, which is featured by a hierarchical structure consisting of two modules: (1) path searching, (2) speed profile generation. Firstly, a path searching method based on deep learning and Monte-Carlo tree search is proposed to generate a coarse path connecting starting and terminal points. Then, the path is smoothed using path optimization and provided to the speed-planning module as the reference. Next, a speed planning method based on quadratic optimization is developed, which allows to seek maximum driving comfort and energy saving. Last, extensive simulation experiments were conducted in the real environment of mining sites. The results verified that LOTP enhances the computational efficiency and success rate of path planning and helps generate an optimal speed profile. Furthermore, LOTP exhibits desirable potential for the practical application of autonomous driving at mining sites. Source implementation will be released as an open-source code.
1 September 2025 in IEEE Transactions on Vehicular Technology
Parking trajectory planning in loading/dumping areas of open-pit mines presents unique challenges compared to standard parking scenarios for passenger vehicles. These include irregular workspaces, kinematic constraints of heavy trucks, and customized trajectory requirements, which collectively formulate a mixed-integer nonlinear programming (MINLP) problem with non-differentiable logical condition constraints. Mainstream MINLP solvers struggle to directly handle logical conditional constraints involving integer variables and usually fail to meet the real-time requirements of practical applications. This study presents a nearly global optimal trajectory planning method that developed a search-then-optimize framework based on the divide-and-conquer strategy, separating the difficulties into distinct stages. A multiple-priority-queue search-based method is introduced in the search stage, leveraging its gradient-free nature to tackle non-differentiable variables and generate an initial solution near the global optimal solution. In the numerical optimization stage, the problem is reformulated as a fully differentiable NLP, refining the initial trajectory to enhance quality. Extensive simulations and real-vehicle tests confirm that this method effectively addresses the above difficulties, producing efficient and trackable trajectories.