A year ago, I recognized that Intel’s end-to-end platform for autonomous driving wasn’t complete.
Large gaps existed. We couldn’t create the amazing experiences we wanted to, and, more important, we weren’t able to complete the vision for saving lives. To this end, Intel acquired Mobileye – a company with best-in-class technology and visionary leadership. Its proven track record and the millions of production vehicles already equipped with Mobileye technology are laying the foundation for tomorrow’s autonomous vehicles.
Today I can confidently say that both Intel and Mobileye are better together. Our combined portfolio spans in-vehicle computing with Mobileye EyeQ SoCs (systems-on-chip), all the way to the cloud with Intel® Xeon® processors. Nesting in Mobileye puts Intel at the forefront of the autonomous revolution.
Data is the world’s most valuable resource. I said it in 2016, and it remains true today. Even more exciting is the ability for computing to unlock data in powerful ways that we have only imagined – society-affecting ways like autonomous driving.
The road to autonomy will require important groundbreaking collaborations in many areas – the most important being safety. Safety is not new to Mobileye: Its products, such as its EyeQ SoC-powered collision avoidance solutions, are already installed in some 27 million vehicles globally.
However, artificial Intelligence (AI) systems used in autonomous vehicles today for decision-making are probabilistic in nature. These techniques are excellent for comfort and assertiveness, but require an additional software layer for safety, as even a very small chance of a safety risk is not acceptable. Instead what we need is a formal deterministic system that sits on top of this AI and provides a safety envelope. That’s why our acquisition of Mobileye has helped us push the boundaries of safety. Last year, we developed and published a safety framework called Responsibility-Sensitive Safety (RSS) for the industry to consider as a starting point for this important work.
RSS formally defines what it means to drive safely and the boundaries of where human-like assertive driving becomes unsafe driving. It adds a non-proprietary and definitive safety and transparency layer to verify decisions proposed by any developer’s driving policy (planning) software, which are typically probabilistic and opaque.
Put simply, planning or policy is how you get from point A to point B; RSS is what will keep you safe along the way.
We have called for industry, government regulators, academia and consumer groups to work together to align on an open, transparent and verifiable method to assure the safety of autonomous vehicle decision-making.
We believe that society will only embrace automated vehicles when they demonstrate the ability to improve safety and save lives, not by modest amounts, but by multiple orders of magnitude.
Today at Automobil Elektronik, I spoke to a group of corporate executives from the automotive industry and urged them to band together in this safety challenge. The automotive industry can’t do it alone. We need and welcome feedback and collaboration from governments, regulators, standards bodies and academia to align on an open, transparent and verifiable solution for the safety of autonomous vehicles.
At Intel and Mobileye, we’re stronger together as we make a safe driverless future a reality.
Brian Krzanich is the chief executive officer of Intel Corporation.