Edge Case Research: Self Driving Car Safety Series Edge Case Research

Phil Koopman, Co-Founder and CTO, Edge Case Research, addresses safety performance indicators covered in the recently published, UL 4600, the Standard for Safety for the Evaluation of Autonomous Products. A section of the 4600 standard is devoted to safety performance metrics.  Phil will take this opportunity to review a few of the 4600 SPIs as well as addressing the important differences between SPI's and KPI or Key Performance Indicators when it comes to developing a self driving car 

In this episode, Phil Koopman, Co-Founder & CTO, Edge Case Research, addresses conformance-based metrics.  Phil provides questions that can be asked to get more insight into whether a system partially or fully conforms to a standard.

Prof. Philip Koopman, Co-Founder, and CTO, Edge Case Research explains how using metrics to track the unknowns, or surprises can help determine if the self driving car is good at recognizing something weird has happened and if it will respond safely.  

Prof. Phil Koopman discusses approaches to ODD (Operational Design Domain) Metrics.  Phil addresses the importance of taking into account all the dimensions of the ODD to ensure the self driving car is prepared for all the interesting scenarios that it will encounter in the real world.  

Professor Phil Koopman, Co-Founder & CTO, Edge Case Research explains why prediction metrics based on object movement, speed, direction, and acceleration are critical for safe path planning for self driving cars. 

Prof. Phil Koopman, Co-Founder & CTO, Edge Case Research addresses approaches to perception metrics for self driving cars. Hear Phil discuss sensor performance, sensor fusion, classification accuracy, and the ability to track how often the system encounters novel things it isn't trained to see.