A Study on Functional Safety, SOTIF and RSS from the Perspective of Human-Automation Interaction 2021-01-0858

As perhaps the primary conduit of the physical expression of human freedom of movement, transportation in its various forms plays a critical role in human society. The availability of ready transport has shaped the fabric, infrastructure, and, to some degree, even the culture of whole human society. Now, automated driving, in various forms of Advanced Driver Assistance Systems (ADAS) and self-driving systems, is promising a safe, efficient, and productive life to humans by relieving driver from active driving tasks in the context of road transportation. However, recent high-profile crashes, e.g., fatalities involving Uber test autonomous vehicle or Tesla car, have undermined the automated-driving promise of enhanced safety. On the other hand, in automotive industry, there are safety approaches like ISO 26262 and others in-development which all are expected to mitigate the safety risk resulting from automated driving. This contribution firstly reviewed the emerging human-automation interaction issues relating to automated-driving safety. Then, with the respect to these issues, the relevant automotive safety approaches - ISO 26262, ISO 21448 (Safety of the Intended Functionality, SOTIF) and RSS (Responsibility-Sensitive Safety model), were discussed to highlight the blind spots in them which may impair the safety promise of automated driving where now machine learning mechanisms are prevalent. Next, we briefly introduced the system-thinking tools in human factors discipline which are expected to enhance automated-driving technology to address human-automation interaction issues. Finally, we concluded the findings and argued that automated-driving safety risks should be managed not only from technological perspective but also from human-factors perspective, and the philosophy of human use of driving automation can benefit the development of a safe, reliable, and trustworthy Automated Driving System (ADS).