Join us online!

https://engineering.purdue.edu/crisp/seminar

Abstract
According to the 2015 vehicular accidents report [www.nhtsa.gov], there were more than 35000 fatal crashes and more than 6 million non-fatal crashes. Translating the 35K fatal crashes, over 3 trillion driven miles, to a FIT (failures in time, time = one billion hours) rate we get a fatality FIT rate in the range 250-500. A drive system, in a fully autonomous car, that replaces the human driver must at least be an order of magnitude more resilient. In fact the ISO 26262 Auto Safety Standard, stipulates a probabilistic metric for hardware failures (PMHF) to be at most 10 FITs. This target is almost two orders of magnitude improvement over driver related fatal accident FIT rate; notwithstanding, the fact that not all hardware failures result in fatalities.  Autonomous or self-driving car initiative is creating a new center stage for resilient computing and design for testability. This is very apparent from reading the ISO26262 specification, which is about the functional safety for automotive equipment applicable throughout the lifecycle of all automotive electronic and electrical safety-related systems. By way of a detailed review of the ISO26262 standard, this tutorial pays tribute to all of the important and significant ideas that have come through the past 40+ years of research in fault-tolerant computing and design-for-testability. One of the key use cases for self-driving car is neural-network based deep learning. Reliability models that explore the dynamic trade-offs between resiliency and performance requirements for deep learning are examined. Deep neural networks use the computational power of massively parallel processors in applications such as autonomous driving. Autonomous driving demands resiliency (as in safety and reliability) and trillions of operations per second of computing performance to process sensor data with extreme accuracy. This talk examines various approaches to achieve resiliency in autonomous cars and makes the case for design diversity based redundancy as a viable solution.