Follow Us On

Panel Discussion 1

State of AI and ML-January 2020

Collaboration for Safety of Autonomous Vehicles

Panel Chair/Moderator

John Currie

John Currie

Business Development Director at UL

Panelists

Miguel Acosta

Miguel Acosta

California Department of Motor Vehicles, CEA
Sagar Behere

Sagar Behere

Director Systems and Safety Engineering, Aurora
Mike Wagner

Mike Wagner

CEO, Edge Case Research
Benjamin Lewis

Benjamin Lewis

Director, Strategic Partnerships Automotive & Mobility Liberty Mutual
Liam Pedersen

Liam Pedersen

Deputy General Manager, Alliance Innovation Lab - Silicon Valley

Abstract

Advances in artificial intelligence and machine learning, along with compute and sensor technologies, are converging in autonomous vehicle (AV) systems, and are expected to transform the future of global mobility. The safe operation of autonomous vehicles concerns all stakeholders, including vehicle OEMs, insurers, governments, and of course the general public.   This panel discussion engages in a conversation about the need to  understand the risks and opportunities of autonomous, AI-enabled systems, and most importantly how to help ensure their safe deployment and operation in public spaces.  We bring together industry thought leaders to consider topics in AV safety and implications for society.

This will be a high level discussion, and not delve into specific legal or ethical concerns.  Rather, we seek to discuss important aspects for consideration, including which safety practices from the existing mobility or other industries may be applicable, or what may be considered best practices for enhancing safety. Some of the questions that are addressed are

Who owns safety (i.e. OEM, vehicle owner/operator, insurers)?  How is safety measured?  What is safe enough?  Is 3rd-party verification needed?  What standards exist/what standards are needed?  How will vehicle insurers quantify risk and set rates?  What is the role of municipal, state, and federal governments?  Role of simulation, test track, and public road testing?  If AV algorithms can learn and self-modify, how will safety compliance be ensured over the life of the vehicle?