Improving Trustworthiness of Self-driving Systems

Abstract

Self-Driving Vehicles (SDVs) are considered to be safety-critical system. They may jeopardize the lives of passengers in the vehicle and people in the street, or damaging public property such as the transportation infrastructure. According to the National Transportation Safety Board report [1] of an Uber self-driving crash, the accident was caused by the internal components of SDVs when the AI module failed to detect a victim. The autonomous system was implemented to give a human driver control of a vehicle on the unmanaged areas; however, the driver was distracted and did not react within the appropriate time.