Grant: Alliance Grants

The proposed project aims to enhance the safety and assurance of driving automation systems (DAS), such as, such as advanced driver-assistance systems (ADAS) and autonomous vehicles (AVs). These systems must make decisions in an open traffic environment, which they can neither control nor fully predict. This makes the safety validation of such systems extremely challenging. One class of scenarios that these systems must handle are those involving dynamic occlusions, i.e., occlusions caused by other vehicles in traffic. Dynamic occlusions create a safety risk, where a vehicle may collide with other vehicles or pedestrians who suddenly emerge from such occlusions. The objective of this research is to develop methods to systematically generate test scenarios to assure safe handling of dynamic occlusions by DAS. The main challenge is that collisions are rare, and occlusion-caused collision scenarios are even rarer. The proposed methods will use real traffic data to generate realistic scenarios where cars with driving automation may face these hidden dangers. These scenarios can then be used to test DAS in simulation to assure their safe handling. The propose collaboration between Ford and University of Waterloo researchers will address key technical challenges of critical scenario generation and strengthen Ford's market leadership in safe and highly capable DAS. The project will benefit society by informing industry standards, improving safety assurance of DAS and potentially other autonomous robots, and training HQP in DAS technology. With the automotive sector being Canada's biggest contributor to manufacturing GDP, the program's innovation and highly qualified personnel (HQP) training will help maintain Canada's role as a world leader in the automotive sector and robotics.