Multi-task Integrated Driving Decision and Control Method Oriented to Driving Demands

The integration of various driving tasks and demands has become a trend in the decision-making and control technology of autonomous vehicles. However, expanding driving conditions generate significant conflicts between demand indicators, posing greater challenges to algorithmic efficiency. To address this problem, this study proposes a real-time method based on nonlinear model predictive control theory. The proposed method features multiple driving functions and satisfies multiple requirements. An indicator scheduling strategy based on driving demand priority was first designed to handle the conflict between demand indicators and dynamically adjust the integration method of the indicators. Two indicator functions were next established based on frequency response analysis to balance calculation accuracy and complexity. An adaptive Lagrange discretization method was then designed to ensure control accuracy with fewer discrete points. Simulations and experiments validate the ability of the proposed method to achieve multiple tasks under normal and emergency driving conditions, satisfy performance requirements, and solve algorithms in less than 50 ms. The results present a novel perspective for improving real time efficiency of planning algorithm. © 2024 Chang'an University. All rights reserved.