Establishment of Safety Assurance WG

[mitsudome-r]

The new working group was approved by the TSC and we will be starting the working group calls from this week. If you are interested around safety assurance of autonomous driving technologies, feel free to join the call!

Working Group Goals

• Lead the development of continual safety assurance for autonomous driving systems based on Autoware.
• Design process to integrate safety assurance into the architecture, development, evaluation, and operation of the Autoware System.
• Research and development of trustworthy AI applied to mid integrity (SIL2) system within Autoware.

Working Group Scope
Some specific areas of interest of this WG include, but not limited to, the following:

• Understanding current state of Autoware related to safety and testing approaches.
• Investigate Runtime Safety Assurance and define reference architecture for enhanced safety.
• Identify ML based systems and their inherent safety issues (long-tail, open world classes, out-of-distribution, out-of-model scope, supervisor model, herds of dumb models (HDM), etc.)
• Definition of safety metrics and benchmarks to evaluate safety performance and track improvement.
• Integration/conformity with safety standards (IEC 61508 Ed3) and ISO PAS 8800 Road Vehicles – Safety and Artificial Intelligence

How to join the working group
The first meeting is scheduled from 2pm UTC on April 2nd. Like other working group, you can join to the call from our Autoware Events Calendar