Model-Based Synthesis and Safety Assurance of Intelligent Controllers for Autonomous Vehicles

点击翻译按钮获取中文摘要

Vehicles with limited self-driving capabilities are already on the market and some car makers have promised products capable of autonomous driving in an urban setting in 2020. Self-driving cars will eventually completely transform the automotive industry, replacing private car ownership by service-based products such as robotic cabs. The deployment of large-scale self-driving vehicle fleets will reduce the number of crashes and crash severity, reduce emissions, allow commuters to use their time more effectively, and free up spaces occupied by parked cars. While current self-driving technologies have improved immensely in recent years, a major challenge is assuring safe and appropriate operation of an autonomous vehicle in all traffic situations and all road conditions. New design methods and tools are needed to address this challenge. This Discovery grant will fund a cutting-edge research program to create new methods and tools for the development of safe and intelligent computer-based controls for self-driving cars. The expected results include (1) a controller design and associated machine-learning methods to enable controllers capable of intelligent decision making and human-like driving skills and (2) methods for safety assurance of systems that rely on machine learning. The program will use the Waterloo Autonomous Driving Testbed, a drive-by-wire Lincoln MKZ with an integrated sensor suite, and take advantage of Ontario's new Automated Vehicle Pilot Program to demonstrate and evaluate the new technology on public roads. Self-driving cars will have a major impact on our society and economy over the coming decades. It is expected that self-driving cars can reduce accident rates by as much as 80%. In Canada, this would save 1,600 lives a year and over $55 billion in health costs and lost productivity. Goldman Sachs predicts a driverless technology market of $96 billion as early as 2025. The proposed research program targets current major roadblocks to a wider adoption of self-driving cars: (1) intelligent decision making in traffic that mixes driverless and human-driven cars with other participants such as cyclists and pedestrians and (2) safety assurance of such systems. The program will generate cutting-edge research results in driverless technology, establish an open test and simulation benchmark to stimulate scientific and technological progress, create commercializable technology, and train thirteen HQP in systems and software engineering for autonomous driving over the next five years. Hands-on experience on a self-driving car will give these HQP a significant competitive advantage on the job market. The automotive sector is the single biggest contributor to Canada's manufacturing GDP. Canadian innovation and HQP training in driverless technology will contribute to maintaining Canada's role as a world leader in the automotive sector.