Making Autonomous Systems Safe - One in a Billion Failure Rate

Watch a comprehensive lecture where Richard Murray, the Thomas E. and Doris Everhart Professor at Caltech, explores the critical challenges of ensuring safety in autonomous systems. Delve into the complex world of self-driving cars and safety-critical decision-making protocols, examining how engineers can validate system safety to meet society's expectation of less than one critical failure per billion operations. Learn about the integration of model-based simulations, log-based regression, real-world operations, and structured testing methodologies. Explore the crucial role of formal specifications, assume/guarantee contracts, and test synthesis in developing reliable autonomous systems. Gain insights into the sophisticated combinations of model-based and data-driven algorithms that power real-time decision-making in autonomous vehicles, drawing from Murray's extensive research in feedback control, networked systems, synthetic biology, and autonomy.