Motion planning and control integrated with LLM for unstructured terrain with slope restrictions
To address the core challenges of autonomous driving motion planning in complex unstructured environments, this paper proposes an integrated solution that combines slope constraints, large language model parameter optimization, and large language model obstacle avoidance decision-making. First, a slope-aware 3D Hybrid A* path planning algorithm with a multi-objective cost function and an optimized heuristic function is proposed. Secondly, an adaptive parameter optimization framework integrating Hybrid A* with Large Language Model (LLM) is developed, dynamically adjusting key parameters through a closed-loop mechanism. Finally, an LLM-assisted model predictive control (MPC) scheme for trajectory tracking and obstacle avoidance is developed, which achieves a balanced trade-off between tracking accuracy and collision safety. The proposed obstacle avoidance method is compared with alternative approaches.
Authors
- Lei Cai (ORCID: https://orcid.org/0009-0007-9238-3799)
- Hsin Guan (ORCID: https://orcid.org/0000-0001-5027-3515)
- Guang Li
- Wei Sun
- Ming Yue
- Xing Meng
Institutions
- Jilin University (CN)
- Dalian University of Technology (CN)
- State Key Laboratory of Automotive Simulation and Control (CN)
Publication Details
- Journal
- Proceedings of the Institution of Mechanical Engineers Part D Journal of Automobile Engineering
- Published
- 2026-11-20
- DOI
- https://doi.org/10.1177/09544070261501325
- Primary Topic
- Robotic Path Planning Algorithms
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- National Key Research and Development Program of China