Developed Risk Awareness Estimated System for Autonomous Vehicles

An important step in autonomous vehicle technology. The new system developed by the research team offers a risk-conscious approach to understand the environment of vehicles and make safe decisions. With multimodal orbit prediction method, the system’s behavior coverage is expanded but the limitations of existing solutions have been revealed.

• Part 19,74 percent of candidates in existing systems are in poor or weak category
• Even the best candidate is ranked in the average 33,9., which shows the candidate selection quality is problematic
• Register-based basic system fails to make reliable ranking between geometrically similar orbitals
• Research aims to solve deficiencies in heavy scene views
• In addition to risk awareness, the system is aimed to make more robust decisions against environmental ambiguities

Problems caused by missing or insufficient scene data are severe obstacles in real world applications of autonomous driving systems. Research makes risk assessment a central element to address these problems. Thus, the vehicle can take into account not only the possible options, but also those who are safe.

This development will allow autonomous vehicles to decide safer and more precisely. The results of the research have the potential to bring significant improvements in the industry.