A Kamm's Circle-Based Potential Risk Estimation Scheme in the Local Dynamic Map Computation Enhanced by Binary...

Area of Science:

• Robotics and Intelligent Systems
• Transportation Engineering
• Computer Science

Background:

• Autonomous vehicles (AVs) rely heavily on sensors, posing risks. High-definition maps (HD maps) can mitigate this by providing geographical context.
• Cooperative intelligent transport systems (C-ITS) and local dynamic maps (LDMs) are crucial for advanced AV safety and real-time risk assessment.
• Effective LDM implementation necessitates a robust database design capable of updating information on potential future vehicle interactions.

Purpose of the Study:

• To propose an advanced method for embedding future vehicle occupancy data into an LDM database.
• To enhance the risk assessment capabilities of LDMs for autonomous driving safety.
• To demonstrate a novel approach for managing dynamic information in C-ITS.

Main Methods:

• Formulating geographical future vehicle occupancy using Kamm's circle model.
• Developing a database embedding method utilizing Binary Decision Diagrams (BDDs).
• Implementing and testing the BDD-based occupancy data sharing in a ROS-based simulator.

Main Results:

• Successfully demonstrated the sharing of BDD-based occupancy data in a simulated environment.
• Utilized linked list-based BDDs for efficient data representation and manipulation.
• Showcased the effectiveness of algebraic operations on exchanged BDDs for managing future interactions and collision avoidance.

Conclusions:

• The proposed BDD-based method enables efficient management of future vehicle interactions within LDMs.
• This approach significantly contributes to the realization of ideal LDMs for enhanced autonomous vehicle safety.
• The study opens new possibilities for advanced C-ITS and safer autonomous mobility.