Verifying safety for systems with neural-network-based controllers such as the autonomous F1/10 racing car.
About this Event
Autonomous systems interacting with the physical world, collecting data, processing it using machine learning algorithms, and making decisions, have the potential to transform a wide range of applications including medicine and transportation.
Realizing this potential requires that the system designers can provide high assurance regarding safe and predictable behavior. This motivates research on formally verifying safety (such as avoidance of collisions) of closed-loop systems with controllers based on learning algorithms.
In this talk, I will use the experimental platform of the autonomous F1/10 racing car to highlight research challenges for verifying safety for systems with neural-network-based controllers.
Our solution to safety verification, incorporated in the tool Verisig at Penn, builds upon techniques for symbolic computation of the set of reachable states of hybrid (mixed discrete-continuous) systems. The case study consists of training the controller using reinforcement learning in a simulation environment, verifying the trained controller using Verisig, and validating the controller by deploying it on the F1/10 racing car.
Rajeev Alur is Zisman Family Professor of Computer and Information Science at University of Pennsylvania. He obtained his bachelor’s degree in computer science from IIT Kanpur in 1987 and PhD in computer science from Stanford University in 1991. Before joining Penn in 1997, he was with Computing Science Research Center at Bell Labs.
His research is focused on formal methods for system design, and spans cyber-physical systems, programming languages, and theoretical computer science.
He is a Fellow of the AAAS, a Fellow of the ACM, a Fellow of the IEEE, an Alfred P. Sloan Faculty Fellow, and a Simons Investigator. He was awarded the inaugural Computer-Aided Verification award, the inaugural Alonzo Church award by ACM SIGLOG / EATCS / EACSL / Kurt Goedel Society, Distinguished Alumnus Award by IIT Kanpur for his work on timed automata. Prof. Alur has served as the chair of ACM SIGBED (Special Interest Group on Embedded Systems), the general chair of LICS, and the lead PI of the NSF Expeditions in Computing center ExCAPE (Expeditions in Computer Augmented Program Engineering). He is the author of the textbook Principles of Cyber-Physical Systems (MIT Press).