Spatio-Temporal Joint Trajectory Planning for Autonomous Vehicles Based on Improved Constrained Iterative LQR

Area of Science:

• Robotics
• Control Systems
• Artificial Intelligence

Background:

• Autonomous driving requires sophisticated spatio-temporal joint trajectory planning.
• Traditional methods struggle with complex scenarios, highlighting limitations of sequential decoupling.
• Constrained Iterative Linear Quadratic Regulator (CILQR) shows promise but needs efficiency and adaptability improvements.

Purpose of the Study:

• To enhance the Constrained Iterative Linear Quadratic Regulator (CILQR) for autonomous driving trajectory planning.
• To improve computational efficiency, scenario adaptability, and driving comfort.
• To achieve more human-like driving and increased traffic efficiency.

Main Methods:

• Implemented a segmented barrier function truncation strategy with dynamic relaxation factors for stability.
• Introduced an adaptive weight parameter adjustment for acceleration and curvature planning.
• Integrated the hybrid A* algorithm to optimize initial trajectories and enhance iterative efficiency.

Main Results:

• Demonstrated substantial improvements in human-like driving performance (16.35% increase).
• Achieved a 12.65% average increase in traffic efficiency.
• Reduced computation time by 39.29% while maintaining driving comfort.

Conclusions:

• The improved CILQR method offers significant advancements in autonomous driving trajectory planning.
• The enhancements lead to more efficient, adaptable, and human-like autonomous navigation.
• Validated through simulations and real-vehicle tests, proving practical applicability.