Optimal Falsifications for Cyber-Physical Systems using Trajectory Splicing

In this talk, we address the problem of falsifying properties of complex cyber-physical sytems that model the interactions between a discrete controller and a continuous physical environment. Often, such systems that are so complex that a symbolic approach to falsification is beyond the ability of existing solvers. We present two recent promising approaches to falsification based on the ideas of trajectory robustness and trajectory splicing.

We show that the use of trajectory robustness can replace the standard Boolean interpretation of temporal logic properties with a real-valued interpretation that can guide the search for falsifying inputs. Furthermore, we present the idea of trajectory splicing that searches over multiple, possibly disconnected trajectory segments rather than a single trajectory. We present some promising results using our approach and discuss open challenges that will drive future research.

Note: All relevant concepts will be explained during the talk. Joint work with Aditya Zutshi (U. Colorado Boulder), Georgios Fainekos (Arizona State University), Jyotirmoy Deshmukh and James Kapinski (Toyota Technical Center).

Speaker Bios

Sriram Sankaranarayanan is an assistant professor of Computer Science at the University of Colorado, Boulder. His research interests include automatic techniques for reasoning about the behavior of computer and cyber-physical systems. Sriram obtained a PhD in 2005 from Stanford University where he was advised by Zohar Manna and Henny Sipma. Subsequently he worked as a research staff member at NEC research labs in Princeton, NJ. He has been on the faculty at CU Boulder since 2009. Sriram has been the recipient of awards including the President’s Gold Medal from IIT Kharagpur (2000), Siebel Scholarship (2005), the CAREER award from NSF (2009) and the Dean’s award for outstanding junior faculty for the College of Engineering at CU Boulder (2012).

Date:
Haut-parleurs:
Sriram Sankaranarayanan
Affiliation:
University of Colorado, Boulder

Taille: Microsoft Research Talks