Why do self-driving cars crash? King’s College London researchers think they have the answer

Researchers at King’s College London have developed a new method to analyze autonomous vehicle crashes by applying 'actual causality' to AI-driven cyber-physical systems. Traditional safety methods rely on broad failure statistics, which predict future risk but fail to explain why specific accidents occur. By utilizing a 'responsibility-guided search algorithm,' the team can retrospectively trace chains of complex observations and decisions to pinpoint the exact cause of a crash with high computational efficiency. This breakthrough is intended to improve accountability, facilitate system repairs, and foster public trust as autonomous technology integrates further into real-world environments.

(Source：The Brighter Side of News)