Spatial-temporal risk field-based coupled dynamic-static driving risk assessment and trajectory planning in weaving...

Area of Science:

• Transportation Engineering
• Robotics
• Artificial Intelligence

Background:

• Traditional connected and automated vehicle (CAV) technologies struggle with dynamic risk prediction in complex scenarios like weaving segments.
• Existing risk assessment methods for CAVs often lack the foresight needed for optimal trajectory planning, limiting potential benefits.

Purpose of the Study:

• To develop an advanced risk assessment paradigm for CAVs, specifically addressing dynamic risks in weaving segments.
• To enhance trajectory planning for CAVs by improving risk foreseeability and optimizing for safety, comfort, and efficiency.

Main Methods:

• Proposed a three-dimensional spatial-temporal risk field (STRF) incorporating spatial-temporal distances and a specialized field for weaving segments.
• Developed a parameter calibration method using YOLO-based machine vision and aerial video data to improve STRF accuracy.
• Designed a STRF-based trajectory planning method integrating risk occupancy maps, dynamic iterative sampling, and quadratic programming.

Main Results:

• The proposed STRF demonstrated superior risk foreseeability compared to traditional risk fields.
• The STRF-based trajectory planning method significantly improved safety, reduced lane-change completion time, and minimized speed fluctuations.
• The method achieved robust safety performance by considering both dynamic and static risk factors during planning.

Conclusions:

• The spatial-temporal coupled risk assessment paradigm and STRF offer a significant advancement in predicting and mitigating risks for CAVs.
• The developed trajectory planning method enhances CAV performance in weaving segments, outperforming baseline schemes and human driving.
• This research provides a foundation for safer and more efficient operation of CAVs in complex traffic environments.