What if we could develop an automatic rating system to assess the safety of RAS?

This project will design an automatic testing mechanism for the autonomous software that controls a robotic and autonomous system (RAS), including artificial intelligence and machine learning based controllers of RAS.

The testing mechanism will generate scenarios that are likely to cause dangerous or undesirable behavior in the RAS software being tested, by applying recent advances in decision making under uncertainty.

In the longer term, this project will form the core component of an automatic rating system to assess the safety of RAS – akin to the NCAP (New Car Assessment Programme) rating for cars.

Project team